diff --git a/CMakeLists.txt b/CMakeLists.txt
index 0e535f5d..32da575c 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -50,6 +50,39 @@ set(GLM_ROOT_DIR "${CMAKE_SOURCE_DIR}/external")
find_package(GLM REQUIRED)
include_directories(${GLM_INCLUDE_DIRS})
+# Include tinygltf header
+include_directories("${CMAKE_SOURCE_DIR}/external/include")
+
+# Include OIDN headers
+include_directories("${CMAKE_SOURCE_DIR}/external/include/OpenImageDenoise")
+
+# Define OIDN libraries
+set(OIDN_LIBRARIES
+ "${CMAKE_SOURCE_DIR}/external/lib/OpenImageDenoise.lib"
+ "${CMAKE_SOURCE_DIR}/external/lib/OpenImageDenoise_core.lib"
+ "${CMAKE_SOURCE_DIR}/external/lib/OpenImageDenoise_device_cuda.lib"
+ "${CMAKE_SOURCE_DIR}/external/lib/OpenImageDenoise_device_cpu.lib"
+ "${CMAKE_SOURCE_DIR}/external/lib/OpenImageDenoise_utils.lib"
+ "${CMAKE_SOURCE_DIR}/external/lib/tbb12.lib"
+)
+
+# Define the destination directory
+set(OIDN_DEST_DIR "${CMAKE_RUNTIME_OUTPUT_DIRECTORY}")
+
+# Add a custom command to copy the OIDN libraries
+foreach(OIDN_LIB ${OIDN_LIBRARIES})
+ get_filename_component(OIDN_LIB_NAME ${OIDN_LIB} NAME)
+ add_custom_command(
+ OUTPUT "${OIDN_DEST_DIR}/${OIDN_LIB_NAME}"
+ COMMAND ${CMAKE_COMMAND} -E copy_if_different ${OIDN_LIB} ${OIDN_DEST_DIR}
+ DEPENDS ${OIDN_LIB}
+ )
+ list(APPEND OIDN_COPY_TARGETS "${OIDN_DEST_DIR}/${OIDN_LIB_NAME}")
+endforeach()
+
+# Add a custom target to execute the copy command
+add_custom_target(copy_oidn_libs ALL DEPENDS ${OIDN_COPY_TARGETS})
+
set(headers
src/main.h
src/image.h
@@ -66,6 +99,7 @@ set(headers
set(sources
src/main.cpp
src/stb.cpp
+ src/tinygltf.cpp
src/image.cpp
src/glslUtility.cpp
src/pathtrace.cu
@@ -120,11 +154,22 @@ elseif(CMAKE_VERSION VERSION_LESS "3.24.0")
else()
set_target_properties(${CMAKE_PROJECT_NAME} PROPERTIES CUDA_ARCHITECTURES native)
endif()
+
+# Link OIDN libraries only in Release mode
target_link_libraries(${CMAKE_PROJECT_NAME}
${LIBRARIES}
cudadevrt
+ $<$:${OIDN_LIBRARIES}>
#stream_compaction # TODO: uncomment if using your stream compaction
- )
+)
+
+# Add dependencies for copying OIDN libraries only in Release mode
+add_custom_command(TARGET ${CMAKE_PROJECT_NAME} POST_BUILD
+ COMMAND ${CMAKE_COMMAND} -E echo "Copying OIDN libraries..."
+ COMMAND ${CMAKE_COMMAND} -E copy_if_different $<$:${OIDN_LIBRARIES}> ${OIDN_DEST_DIR}
+ COMMENT "Copying OIDN libraries in Release mode"
+)
+
target_compile_options(${CMAKE_PROJECT_NAME} PRIVATE "$<$,$>:-G;-src-in-ptx>")
-target_compile_options(${CMAKE_PROJECT_NAME} PRIVATE "$<$,$>:-lineinfo;-src-in-ptx>")
-set_property(DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR} PROPERTY VS_STARTUP_PROJECT ${CMAKE_PROJECT_NAME})
+target_compile_options(${CMAKE_PROJECT_NAME} PRIVATE "$<$,$>:-lineinfo;-src-in-ptx>;--extended-lambda")
+set_property(DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR} PROPERTY VS_STARTUP_PROJECT ${CMAKE_PROJECT_NAME})
\ No newline at end of file
diff --git a/README.md b/README.md
index 110697ce..88576bda 100644
--- a/README.md
+++ b/README.md
@@ -3,11 +3,205 @@ CUDA Path Tracer
**University of Pennsylvania, CIS 565: GPU Programming and Architecture, Project 3**
-* (TODO) YOUR NAME HERE
-* Tested on: (TODO) Windows 22, i7-2222 @ 2.22GHz 22GB, GTX 222 222MB (Moore 2222 Lab)
+* Matt Schwartz
+* Tested on: Windows 10 22H2, Intel(R) Core(TM) i7-10750H CPU @ 2.60GHz, NVIDIA GeForce RTX 2060
-### (TODO: Your README)
+
+
+
-*DO NOT* leave the README to the last minute! It is a crucial part of the
-project, and we will not be able to grade you without a good README.
+
+
+

+
+
+# Background
+
+Welcome to my CUDA-based path tracer! What is a path tracer, I hear you asking? In its simplest form, it's a program that follows the paths of light rays as they bounce around a scene - absorbing into materials, scattering off others, refracting in yet others, etc. - and colors the pixels on your screen according to what those rays intersect and how they interact with those materials. If you follow physically-based rules for how those light rays interact with objects, you can get very realistic looking scenes!
+
+One important note about path tracers - it's impractical to follow every light ray in existence! We really only care about the ones that make it to our "eye" / "camera"; those are the ones we see. To find those rays, the ones we see, we actually operate in reverse! We "shoot" rays out in all directions from the camera, and follow their paths until they terminate at a light source (or until we say "enough is enough!"). I like to think of it like echolocation, but with light.
+
+That's enough context, for now - let's take a tour through some path-traced images and the features implemented in this project.
+
+# Features
+## Visual Features
+
+### Diffuse, reflective, refractive, and emissive surfaces!
+
+
+One of the marvelous abilities of path tracers is how easy it is to model a wide variety of different materials. Simply changing the distribution of paths that a light ray scattering off a material can take will change a material from metallic, to plastic, to glassy, and more.
+
+The simplest type of material is a "diffuse" one. In terms of light scattering, a diffuse materially is equally likely to scatter incoming light in any outgoing direction (*simplification). In more common terms, we might call such a material "matte". Here's what it looks like:
+
+
+

+
+
+
+
+
+On the other hand, if incoming light scatters off a surface at the same angle it came in at, we get a reflective surface:
+
+
+
+
(The black in the center is reflecting the void that our "camera" is sitting in
+
+
+
+
+
+Light isn't limited to absorbing and scattering, however. It can also *refract* - or bend through materials that *transmit* light through their volumes. For example - glass, plastic, crystals, etc. The physical quantity that determines how much light bends in these media is called the index of refraction; by changing this value, we can easily render a variety of transmissive media!
+
+
+

+
+
+
+
+
+### Depth of field
+
+In the real world, whether viewed through a camera or the human eye, there is a certain range of distance in which objects are sharply in focus. This range is called depth of field. Objects outside of this range - either too close or too far away from the viewer - are blurry. You can try this by holding a finger in front of your own eye and looking past it; it should appear blurry. The source of the blur comes from the fact that light rays must focus through a lens (of your eye or a camera). Physics tells us that a lens of a given aperture size and radius of curvature has a given distance of focus. Light rays outside that distance do not come to a neat focus point after traveling through the lens.
+
+In the path tracer, this effect can be physically simulated. Remember that we trace rays in reverse - starting from the view point. For depth of field, we imagine a lens of a given aperture size, and jitter our ray origins around in that lens, adjusting their directions such that they point towards the plane of focus.
+
+
+

+
+
+### Stochastic antialising
+
+One important point about path tracing that I haven't touched on yet, is that to get a nice result, we actually have to trace multiple light rays *per* pixel. Why? Well, your screen only has a finite number of pixels, and each one may span an area of a material or multiple materials. If we only trace one path in the direction of that pixel, we're essentially saying "take the color at the middle of the pixel and use it to color the *whole* pixel". It's not a bad first approach, but it can create artifacts in the render known as aliasing. To resolve these artifacts, we can take multiple samples in each pixel and average them. Not unexpectedly, that's called *anti*aliasing (or, in this case, "stochastic sampled antialiasing," since there's more than one way to achieve this).
+
+### Denoising
+
+One of the cons of antialiasing in this way, is that it makes the image "noisier" - or more static-filled - and requires us to take more samples to converge on the true colors of each pixel. To help mitigate this, I integrated Intel's Open Image Denoiser (OIDN) to help clean up and speed up the render results. OIDN includes a pretrained ML model that is specially adapted to clean up noise in stochastic path tracers such as this one. You give it some key information about the scene: for each pixel, the path-traced pixel color, the albedo (material base color), and surface normal (direction). It gives you back some very slick denoised results (which I've secretly been showing this whole time in each render).
+
+### Arbitrary mesh loading and intersection
+
+Now that we can represent materials, visual effects, and get clean, noise-free renders, the real fun begins. At this stage, the path tracer can only render spheres and cubes. That's because, for each different shape, I have to write code that mathematically tests if a light ray intersects the shape. For spheres and cubes, that's easy. But what about for a spaceship? For more complex shapes, we need to be able to break down a mesh into a triangles and test each triangle individually for intersections. Not so difficult in theory, but if we want to render an arbitrary mesh, we'll need a way to represent it and load it into the program. There are many file types commonly used to represent 3D models - `.obj`, `.fbx`, my choice for this project, `.gltf` (Graphics Language Transmission Format), and others. `glTF` is a rich file type that includes information about the scene's 3D models, cameras, lights, textures, and more. For that richness, you must pay the price - it's a little more complex to load and parse than, say, a `.obj` file. But, if you follow this nifty [Khronos group glTF reference cheat sheet](https://www.khronos.org/files/gltf20-reference-guide.pdf), you get arbitrary meshes:
+
+
+

+
+
+### Texture mapping and normal mapping
+
+As I'm sure you'll agree, that space ship looks a little bare. Sure, we could use our previous work to make it metallic, glassy, etc., but few real objects are so uniform. What we need, is texture! If we can map points on our 3D spaceship model to the right points on a colorful 2D image, we can "paint" the spaceship with that image. Simple in concept, very difficult in implementation. Without going into too much detail, I'll enumerate some of the challenges that I encountered:
+1. Loading the texture from glTF (turns out, if your 3D model has a certain number of triangles, glTF changes the data type it uses to represent them!)
+1. Getting the texture from CPU to GPU memory.
+1. Getting the right 2D UV coordinates to sample the texture.
+1. Debugging! (See appendix for bloopers here...) Boy is it difficult to diagnose the issue when the problem is on the GPU-side!
+
+But after many painful hours... voila!
+
+
+

+
+
+Similar to texture-mapping, we can also enhance our models by performing a technique known as "bump" or "normal" mapping. All of our 3D models have normals at each vertex that dictate the direction of the surface at that point. These normals are used by the path tracer to calculate the direction of scattered light when it hits a given point on the model. However, with only so many vertices on a model, it can be difficult to get smooth and realistic results. We could increase the vertex count of our model, but that means more intersections to compute and other costly calculations. Instead, we can once again use 2D textures to enhance our model. We encode the direction of normals (x,y,z) as the red-green-blue components of the image, and again "paint" them onto to each point on the model's surface. The model's geometry stays as coarse as before, but its response to light becomes much more granular.
+
+### Environment mapping
+
+Lastly, one final visual cherry-on-top. All of our scenes so far have been indoors - that's because, until now, outdoors has been a big black void. When our rays don't intersect anything, we simply color those pixels black. What if, instead, we use a high definition image - one that is specifically captured to map onto a sphere - and we color those empty pixels according to the image. This is environment mapping. The key is, that any lights that bounce off our objects in our scene and into the environment map void will also reflect the color of that environment map. This helps breathe some depth and life into our scene.
+
+Putting it all together:
+
+
+
+
+
+
+## Performance features
+
+With all these visual features, the frame rate on this path tracer sure has become abysmal. I believe at my low point, I was taking ~3 samples per pixel, per second, for a relatively low-poly scene. For the above spaceship scene, a reasonable image would take a couple minutes to render, and I had already done a few optimizations by then.
+
+Let's discuss some of the ways I improved the performance of the path tracer to be capable of rendering more complex scenes in reasonable times. Along the way, I'll provide some analysis of the performance improvements; unless otherwise specified, tests were done in a simple cornell box scene with the Ironman model featured at the top of the README - about 30K triangles.
+
+### Stream compaction for early ray termination
+
+When a ray bounces and doesn't intersect anything, we stop following it - we set it's `remainingBounces` to 0. Doing so means that any GPU thread following a ray with 0 bounces remaining can retire early. However, we can do better than this; rather than returning from the thread early, what if we simply never launch the thread at all? That would leave fewer threads idle and more GPU resources available to calculate actively bouncing light rays.
+
+To achieve this, after each bounce of all of our rays, we perform *stream compaction*. That is, we (in parallel) take our array of rays, and sort them so that all rays that are still bouncing are at the front. Then, for the next bounce, we only launch threads for those rays, and give the GPU only the data pertaining to those rays. Less data transfer, fewer calculations to do = better performance!
+
+
+
+
+
+### Material sorting
+
+Another, similar but slightly more subtle optimization we can do is to sort rays by the type of material they intersected with, *before* shading them. Why would this help? Wouldn't it just make things slower because now we have to sort all these rays? Yes, and no.
+
+On the GPU, threads are partitioned into groups called warps. These warps operate in lock-step; each thread in a warp performs the same instruction on different data. This is the nature of the GPU; it's what allows it to operate so quickly, and also what can cause it to operate sluggishly, without the proper care. If the GPU instruction contains a branch, an if-statement for example, and different threads in a warp evaluate that condition differently, they *diverge*. Essentially, group A carries out instructions from branch A, while group B waits, then vice-versa. Our beautiful parallelism has been serialized! In the context of our path tracer, our shading code will certainly contain different behavior for different material interactions. Thus, to reduce divergence, we sort our rays by material intersection type *before* shading. The hope is that the increase in speed due to GPU coherence outweighs the cost of the sorting!
+
+
+
+
+
+
+Well, that *was* the hope - it seems the path tracer performs better without the sort! Granted, the data above comes from a scene with few materials, so the benefit of a sorting pass isn't too great. I tried some more complex scenes, with more materials as well, but the performance improvements were not statistically significant. If there was an improvement, it was shadowed by other, more significant factors.
+
+### Russian roulette termination
+
+Not all rays are equal - darker rays (those that have interacted with dark materials / haven't hit a light) don't contribute as much to the final render result as lighter rays. We can save ourselves some compute time, with little affect on the final result, by terminating these rays early. The trick is determining which rays are worth terminating, and doing so in a way that doesn't bias the image.
+
+An easy heuristic we can use is to evaluate the *luminance* of a ray. The luminance is a particular combination of the red, green, and blue color channels of the ray; a combination that mimics the sensitivity of human eyes (which are more receptive to green light). Higher luminance effectively means brighter rays. We normalize that luminance to the range [0, 1], and choose a random number in that same range for each ray we shoot. Comparing the two, we decide whether or not to keep the ray.
+
+A couple caveats:
+- In order to not "lose" energy from the scene and bias the image, russian roulette also involves "redistributing" that energy to unterminated rays by bumping up their colors by a factor proportional to the luminance.
+- In my experimentation, I found this effect to be a little too strong. It seemed to wash out shadows as well. I decided to limit its effect by only applying it to rays after a given number of bounces (say, depth > 2).
+
+
+
+
+
+Not bad for such an easy change to implement! Obviously, results will vary depending on how well lit a scene is, how many bounces you want each ray to perform, and more.
+
+### Bounding Volume Hierarchy (BVH)
+
+Last, but definitely not least, of the performance optimizations is Bounding Volume Hierarchies, or BVH. As discussed in the arbitrary mesh loading section, each object is composed of triangles, and every ray needs to test for intersection against every one of those triangles, every sample. You can see how this gets very expensive, very quickly. The low-poly model of the spaceship shown earlier has around ~4k triangles. In an 800x800 image, that's around 2.5 *billion* intersection tests for that object alone, per sample. The GPU is fast, but not that fast. And imagine if we want to render higher-poly models or higher resolution images!
+
+In comes BVH to the rescue; a spatial-partitioning scheme. That sounds intimidating, but the concept is actually quite straightforward. We take our model and divide it into two boxes, or bounding volumes. Then, we doing intersection testing, we test these boxes. Let's say our ray misses one of those boxes - we get to eliminate *half* of the triangles in the model as possible intersectors. We can take this approach further and divide each of the two boxes into two more, and so on. Each time we test a box, we can potentially get rid of half our remaining triangles. The time complexity of this approach is thus `O(log(n))`, where n is the number of triangles in the model. Compare this to our original `O(n)` and it's clear how major this improvement is. Here's a cool visualazation of BVH sourced from a [fantastic video](https://www.youtube.com/watch?v=C1H4zIiCOaI) on ray tracing by Sebastian Lague:
+
+
+

+
+
+And for the FPS comparison:
+
+
+
+
+Now, this is for our 30K triangle Ironman scene. The impact of BVH can vary drastically depending on the triangle count of the scene, the partitioning scheme used, and even the viewing angle (imagine viewing a train head on, versus from the side. Many more intersection tests to do head on). In a more complex scene (the interior home scene shown earlier), which had ~250K triangles among 60 meshes, BVH is the only reason I could get a render at all! A measly 0.8 FPS, but better than nothing!
+
+Also worth noting, my BVH partitions meshes by longest axis (so on each iteration, whichever cardinal axis is longest is the one chosen to split the mesh up further). This is an OK heuristic, but some meshes don't play nicely. For example - imagine a picture frame. On the first iteration, we split it in half; no problems yet. On the second iteration, however, when we split each half in half again, it's likely that 2 of the new bounding boxes contain 0 triangles. For this shape, BVH converges very slowly, and we actually create *more* work for ourselves. To improve upon this basic partitioning scheme, I could implement a [surface area heuristic](https://pbr-book.org/3ed-2018/Primitives_and_Intersection_Acceleration/Bounding_Volume_Hierarchies).
+
+# Bloopers
+No project is complete without a few bloopers, so please enjoy these grainy, terrible renders at my expense!
+
+
+

+
Texture mapping was, perhaps surprisingly, the most difficult part of this project for me. Mismatched euler angle orders between GLM and our utility functions, neglecting to transform normals appropriately, and so much more. Here's my sad, mis-textured spaceship:
+
+
+
+
+
+
+
+

+
The sphere was actually behind the Ironman model in 3D space! When doing intersection tests, I had neglected to transform the intersection distance back into world space, and so objects with non-uniform scale were behaving quite oddly!
+
+
+# References / Credits
+
+"Bedroom Interior Design" (https://skfb.ly/oGPYN) by gozdemrl is licensed under Creative Commons Attribution (http://creativecommons.org/licenses/by/4.0/).
+
+"IRONMAN - TONY STARK MARVEL" (https://skfb.ly/o9Pxq) by Felipe.Sanches.Alves is licensed under Creative Commons Attribution (http://creativecommons.org/licenses/by/4.0/).
+
+"Low-poly Spaceship" (https://skfb.ly/YU9x) by Shophorn is licensed under Creative Commons Attribution (http://creativecommons.org/licenses/by/4.0/).
+
+[Meadow HDRI from Polyhaven](https://polyhaven.com/a/meadow_2)
+
+[Workshop HDRI from Polyhaven](https://polyhaven.com/a/industrial_pipe_and_valve_01)
\ No newline at end of file
diff --git a/external/include/OpenImageDenoise/config.h b/external/include/OpenImageDenoise/config.h
new file mode 100644
index 00000000..b78a324d
--- /dev/null
+++ b/external/include/OpenImageDenoise/config.h
@@ -0,0 +1,81 @@
+// Copyright 2018 Intel Corporation
+// SPDX-License-Identifier: Apache-2.0
+
+#pragma once
+
+#define OIDN_VERSION_MAJOR 2
+#define OIDN_VERSION_MINOR 3
+#define OIDN_VERSION_PATCH 0
+#define OIDN_VERSION 20300
+#define OIDN_VERSION_STRING "2.3.0"
+
+/* #undef OIDN_API_NAMESPACE */
+#define OIDN_STATIC_LIB
+
+#if defined(OIDN_API_NAMESPACE)
+ #define OIDN_API_NAMESPACE_BEGIN namespace {
+ #define OIDN_API_NAMESPACE_END }
+ #define OIDN_API_NAMESPACE_USING using namespace ;
+ #define OIDN_API_EXTERN_C
+ #define OIDN_NAMESPACE ::oidn
+ #define OIDN_NAMESPACE_C _oidn
+ #define OIDN_NAMESPACE_BEGIN namespace { namespace oidn {
+ #define OIDN_NAMESPACE_END }}
+#else
+ #define OIDN_API_NAMESPACE_BEGIN
+ #define OIDN_API_NAMESPACE_END
+ #define OIDN_API_NAMESPACE_USING
+ #if defined(__cplusplus)
+ #define OIDN_API_EXTERN_C extern "C"
+ #else
+ #define OIDN_API_EXTERN_C
+ #endif
+ #define OIDN_NAMESPACE oidn
+ #define OIDN_NAMESPACE_C oidn
+ #define OIDN_NAMESPACE_BEGIN namespace oidn {
+ #define OIDN_NAMESPACE_END }
+#endif
+
+#define OIDN_NAMESPACE_USING using namespace OIDN_NAMESPACE;
+
+#if defined(OIDN_STATIC_LIB)
+ #define OIDN_API_IMPORT OIDN_API_EXTERN_C
+ #define OIDN_API_EXPORT OIDN_API_EXTERN_C
+#elif defined(_WIN32)
+ #define OIDN_API_IMPORT OIDN_API_EXTERN_C __declspec(dllimport)
+ #define OIDN_API_EXPORT OIDN_API_EXTERN_C __declspec(dllexport)
+#else
+ #define OIDN_API_IMPORT OIDN_API_EXTERN_C
+ #define OIDN_API_EXPORT OIDN_API_EXTERN_C __attribute__((visibility ("default")))
+#endif
+
+#if defined(OpenImageDenoise_EXPORTS)
+ #define OIDN_API OIDN_API_EXPORT
+#else
+ #define OIDN_API OIDN_API_IMPORT
+#endif
+
+#if defined(_WIN32)
+ #define OIDN_DEPRECATED(msg) __declspec(deprecated(msg))
+#else
+ #define OIDN_DEPRECATED(msg) __attribute__((deprecated(msg)))
+#endif
+
+#if !defined(OIDN_DEVICE_CPU)
+ #define OIDN_DEVICE_CPU
+#endif
+#if !defined(OIDN_DEVICE_SYCL)
+/* #undef OIDN_DEVICE_SYCL */
+#endif
+#if !defined(OIDN_DEVICE_CUDA)
+ #define OIDN_DEVICE_CUDA
+#endif
+#if !defined(OIDN_DEVICE_HIP)
+/* #undef OIDN_DEVICE_HIP */
+#endif
+#if !defined(OIDN_DEVICE_METAL)
+/* #undef OIDN_DEVICE_METAL */
+#endif
+
+#define OIDN_FILTER_RT
+#define OIDN_FILTER_RTLIGHTMAP
diff --git a/external/include/OpenImageDenoise/config.h.in b/external/include/OpenImageDenoise/config.h.in
new file mode 100644
index 00000000..0d79a643
--- /dev/null
+++ b/external/include/OpenImageDenoise/config.h.in
@@ -0,0 +1,81 @@
+// Copyright 2018 Intel Corporation
+// SPDX-License-Identifier: Apache-2.0
+
+#pragma once
+
+#define OIDN_VERSION_MAJOR @OIDN_VERSION_MAJOR@
+#define OIDN_VERSION_MINOR @OIDN_VERSION_MINOR@
+#define OIDN_VERSION_PATCH @OIDN_VERSION_PATCH@
+#define OIDN_VERSION @OIDN_VERSION_NUMBER@
+#define OIDN_VERSION_STRING "@OIDN_VERSION_MAJOR@.@OIDN_VERSION_MINOR@.@OIDN_VERSION_PATCH@@OIDN_VERSION_NOTE@"
+
+#cmakedefine OIDN_API_NAMESPACE @OIDN_API_NAMESPACE@
+#cmakedefine OIDN_STATIC_LIB
+
+#if defined(OIDN_API_NAMESPACE)
+ #define OIDN_API_NAMESPACE_BEGIN namespace @OIDN_API_NAMESPACE@ {
+ #define OIDN_API_NAMESPACE_END }
+ #define OIDN_API_NAMESPACE_USING using namespace @OIDN_API_NAMESPACE@;
+ #define OIDN_API_EXTERN_C
+ #define OIDN_NAMESPACE @OIDN_API_NAMESPACE@::oidn
+ #define OIDN_NAMESPACE_C @OIDN_API_NAMESPACE@_oidn
+ #define OIDN_NAMESPACE_BEGIN namespace @OIDN_API_NAMESPACE@ { namespace oidn {
+ #define OIDN_NAMESPACE_END }}
+#else
+ #define OIDN_API_NAMESPACE_BEGIN
+ #define OIDN_API_NAMESPACE_END
+ #define OIDN_API_NAMESPACE_USING
+ #if defined(__cplusplus)
+ #define OIDN_API_EXTERN_C extern "C"
+ #else
+ #define OIDN_API_EXTERN_C
+ #endif
+ #define OIDN_NAMESPACE oidn
+ #define OIDN_NAMESPACE_C oidn
+ #define OIDN_NAMESPACE_BEGIN namespace oidn {
+ #define OIDN_NAMESPACE_END }
+#endif
+
+#define OIDN_NAMESPACE_USING using namespace OIDN_NAMESPACE;
+
+#if defined(OIDN_STATIC_LIB)
+ #define OIDN_API_IMPORT OIDN_API_EXTERN_C
+ #define OIDN_API_EXPORT OIDN_API_EXTERN_C
+#elif defined(_WIN32)
+ #define OIDN_API_IMPORT OIDN_API_EXTERN_C __declspec(dllimport)
+ #define OIDN_API_EXPORT OIDN_API_EXTERN_C __declspec(dllexport)
+#else
+ #define OIDN_API_IMPORT OIDN_API_EXTERN_C
+ #define OIDN_API_EXPORT OIDN_API_EXTERN_C __attribute__((visibility ("default")))
+#endif
+
+#if defined(OpenImageDenoise_EXPORTS)
+ #define OIDN_API OIDN_API_EXPORT
+#else
+ #define OIDN_API OIDN_API_IMPORT
+#endif
+
+#if defined(_WIN32)
+ #define OIDN_DEPRECATED(msg) __declspec(deprecated(msg))
+#else
+ #define OIDN_DEPRECATED(msg) __attribute__((deprecated(msg)))
+#endif
+
+#if !defined(OIDN_DEVICE_CPU)
+ #cmakedefine OIDN_DEVICE_CPU
+#endif
+#if !defined(OIDN_DEVICE_SYCL)
+ #cmakedefine OIDN_DEVICE_SYCL
+#endif
+#if !defined(OIDN_DEVICE_CUDA)
+ #cmakedefine OIDN_DEVICE_CUDA
+#endif
+#if !defined(OIDN_DEVICE_HIP)
+ #cmakedefine OIDN_DEVICE_HIP
+#endif
+#if !defined(OIDN_DEVICE_METAL)
+ #cmakedefine OIDN_DEVICE_METAL
+#endif
+
+#cmakedefine OIDN_FILTER_RT
+#cmakedefine OIDN_FILTER_RTLIGHTMAP
\ No newline at end of file
diff --git a/external/include/OpenImageDenoise/oidn.h b/external/include/OpenImageDenoise/oidn.h
new file mode 100644
index 00000000..1d184667
--- /dev/null
+++ b/external/include/OpenImageDenoise/oidn.h
@@ -0,0 +1,517 @@
+// Copyright 2018 Intel Corporation
+// SPDX-License-Identifier: Apache-2.0
+
+#pragma once
+
+#include
+#include
+#include
+
+#include "config.h"
+
+#if defined(__cplusplus)
+ #if defined(SYCL_LANGUAGE_VERSION)
+ #include
+ #define OIDN_SYCL_HPP
+ #elif defined(SYCL_FEATURE_SET_FULL) || defined(SYCL_FEATURE_SET_REDUCED)
+ #define OIDN_SYCL_HPP // not using a SYCL compiler but SYCL headers are included
+ #else
+ namespace sycl
+ {
+ class device;
+ class queue;
+ class event;
+ }
+ #endif
+#endif
+
+typedef struct CUstream_st* cudaStream_t;
+typedef struct ihipStream_t* hipStream_t;
+
+#if defined(__OBJC__)
+ @protocol MTLDevice;
+ @protocol MTLCommandQueue;
+ @protocol MTLBuffer;
+
+ typedef id MTLDevice_id;
+ typedef id MTLCommandQueue_id;
+ typedef id MTLBuffer_id;
+#else
+ typedef void* MTLDevice_id;
+ typedef void* MTLCommandQueue_id;
+ typedef void* MTLBuffer_id;
+#endif
+
+OIDN_API_NAMESPACE_BEGIN
+
+// -------------------------------------------------------------------------------------------------
+// Physical Device
+// -------------------------------------------------------------------------------------------------
+
+#define OIDN_UUID_SIZE 16u // size of a universally unique identifier (UUID) of a physical device
+#define OIDN_LUID_SIZE 8u // size of a locally unique identifier (LUID) of a physical device
+
+// Returns the number of supported physical devices.
+OIDN_API int oidnGetNumPhysicalDevices();
+
+// Gets a boolean parameter of the physical device.
+OIDN_API bool oidnGetPhysicalDeviceBool(int physicalDeviceID, const char* name);
+
+// Gets an integer parameter of the physical device.
+OIDN_API int oidnGetPhysicalDeviceInt(int physicalDeviceID, const char* name);
+
+// Gets an unsigned integer parameter of the physical device.
+inline unsigned int oidnGetPhysicalDeviceUInt(int physicalDeviceID, const char* name)
+{
+ return (unsigned int)oidnGetPhysicalDeviceInt(physicalDeviceID, name);
+}
+
+// Gets a string parameter of the physical device.
+OIDN_API const char* oidnGetPhysicalDeviceString(int physicalDeviceID, const char* name);
+
+// Gets an opaque data parameter of the physical device.
+OIDN_API const void* oidnGetPhysicalDeviceData(int physicalDeviceID, const char* name,
+ size_t* byteSize);
+
+// -------------------------------------------------------------------------------------------------
+// Device
+// -------------------------------------------------------------------------------------------------
+
+// Device types
+typedef enum
+{
+ OIDN_DEVICE_TYPE_DEFAULT = 0, // select device automatically
+
+ OIDN_DEVICE_TYPE_CPU = 1, // CPU device
+ OIDN_DEVICE_TYPE_SYCL = 2, // SYCL device
+ OIDN_DEVICE_TYPE_CUDA = 3, // CUDA device
+ OIDN_DEVICE_TYPE_HIP = 4, // HIP device
+ OIDN_DEVICE_TYPE_METAL = 5, // Metal device
+} OIDNDeviceType;
+
+// Error codes
+typedef enum
+{
+ OIDN_ERROR_NONE = 0, // no error occurred
+ OIDN_ERROR_UNKNOWN = 1, // an unknown error occurred
+ OIDN_ERROR_INVALID_ARGUMENT = 2, // an invalid argument was specified
+ OIDN_ERROR_INVALID_OPERATION = 3, // the operation is not allowed
+ OIDN_ERROR_OUT_OF_MEMORY = 4, // not enough memory to execute the operation
+ OIDN_ERROR_UNSUPPORTED_HARDWARE = 5, // the hardware (e.g. CPU) is not supported
+ OIDN_ERROR_CANCELLED = 6, // the operation was cancelled by the user
+} OIDNError;
+
+// Error callback function
+typedef void (*OIDNErrorFunction)(void* userPtr, OIDNError code, const char* message);
+
+// Device handle
+typedef struct OIDNDeviceImpl* OIDNDevice;
+
+// Returns whether the CPU device is supported.
+OIDN_API bool oidnIsCPUDeviceSupported();
+
+#if defined(__cplusplus)
+// Returns whether the specified SYCL device is supported.
+OIDN_API bool oidnIsSYCLDeviceSupported(const sycl::device* device);
+#endif
+
+// Returns whether the specified CUDA device is supported.
+OIDN_API bool oidnIsCUDADeviceSupported(int deviceID);
+
+// Returns whether the specified HIP device is supported.
+OIDN_API bool oidnIsHIPDeviceSupported(int deviceID);
+
+// Returns whether the specified Metal device is supported.
+OIDN_API bool oidnIsMetalDeviceSupported(MTLDevice_id device);
+
+// Creates a device of the specified type.
+OIDN_API OIDNDevice oidnNewDevice(OIDNDeviceType type);
+
+// Creates a device from a physical device specified by its ID (0 to oidnGetNumPhysicalDevices()-1).
+OIDN_API OIDNDevice oidnNewDeviceByID(int physicalDeviceID);
+
+// Creates a device from a physical device specified by its UUID.
+OIDN_API OIDNDevice oidnNewDeviceByUUID(const void* uuid);
+
+// Creates a device from a physical device specified by its LUID.
+OIDN_API OIDNDevice oidnNewDeviceByLUID(const void* luid);
+
+// Creates a device from a physical device specified by its PCI address.
+OIDN_API OIDNDevice oidnNewDeviceByPCIAddress(int pciDomain, int pciBus, int pciDevice,
+ int pciFunction);
+
+#if defined(__cplusplus)
+// Creates a device from the specified list of SYCL queues.
+// The queues should belong to different SYCL sub-devices (Xe Stack/Tile) of the same SYCL
+// root-device (GPU).
+OIDN_API OIDNDevice oidnNewSYCLDevice(const sycl::queue* queues, int numQueues);
+#endif
+
+// Creates a device from the specified pairs of CUDA device IDs and streams (null stream
+// corresponds to the default stream). Currently only one device ID/stream is supported.
+OIDN_API OIDNDevice oidnNewCUDADevice(const int* deviceIDs, const cudaStream_t* streams,
+ int numPairs);
+
+// Creates a device from the specified pairs of HIP device IDs and streams (null stream
+// corresponds to the default stream). Currently only one device ID/stream is supported.
+OIDN_API OIDNDevice oidnNewHIPDevice(const int* deviceIDs, const hipStream_t* streams,
+ int numPairs);
+
+// Creates a device from the specified list of Metal command queues.
+// Currently only one queue is supported.
+OIDN_API OIDNDevice oidnNewMetalDevice(const MTLCommandQueue_id* commandQueues, int numQueues);
+
+// Retains the device (increments the reference count).
+OIDN_API void oidnRetainDevice(OIDNDevice device);
+
+// Releases the device (decrements the reference count).
+OIDN_API void oidnReleaseDevice(OIDNDevice device);
+
+// Sets a boolean parameter of the device.
+OIDN_API void oidnSetDeviceBool(OIDNDevice device, const char* name, bool value);
+
+OIDN_DEPRECATED("oidnSetDevice1b is deprecated. Use oidnSetDeviceBool instead.")
+inline void oidnSetDevice1b(OIDNDevice device, const char* name, bool value)
+{
+ oidnSetDeviceBool(device, name, value);
+}
+
+// Sets an integer parameter of the device.
+OIDN_API void oidnSetDeviceInt(OIDNDevice device, const char* name, int value);
+
+OIDN_DEPRECATED("oidnSetDevice1i is deprecated. Use oidnSetDeviceInt instead.")
+inline void oidnSetDevice1i(OIDNDevice device, const char* name, int value)
+{
+ oidnSetDeviceInt(device, name, value);
+}
+
+// Sets an unsigned integer parameter of the device.
+inline void oidnSetDeviceUInt(OIDNDevice device, const char* name, unsigned int value)
+{
+ oidnSetDeviceInt(device, name, (int)value);
+}
+
+// Gets a boolean parameter of the device.
+OIDN_API bool oidnGetDeviceBool(OIDNDevice device, const char* name);
+
+OIDN_DEPRECATED("oidnGetDevice1b is deprecated. Use oidnGetDeviceBool instead.")
+inline bool oidnGetDevice1b(OIDNDevice device, const char* name)
+{
+ return oidnGetDeviceBool(device, name);
+}
+
+// Gets an integer parameter of the device.
+OIDN_API int oidnGetDeviceInt(OIDNDevice device, const char* name);
+
+// Gets an unsigned integer parameter of the device.
+inline unsigned int oidnGetDeviceUInt(OIDNDevice device, const char* name)
+{
+ return (unsigned int)oidnGetDeviceInt(device, name);
+}
+
+OIDN_DEPRECATED("oidnGetDevice1i is deprecated. Use oidnGetDeviceInt instead.")
+inline int oidnGetDevice1i(OIDNDevice device, const char* name)
+{
+ return oidnGetDeviceInt(device, name);
+}
+
+// Sets the error callback function of the device.
+OIDN_API void oidnSetDeviceErrorFunction(OIDNDevice device, OIDNErrorFunction func, void* userPtr);
+
+// Returns the first unqueried error code stored in the device for the current thread, optionally
+// also returning a string message (if not NULL), and clears the stored error. Can be called with
+// a NULL device as well to check for per-thread global errors (e.g. why a device creation or
+// physical device query has failed).
+OIDN_API OIDNError oidnGetDeviceError(OIDNDevice device, const char** outMessage);
+
+// Commits all previous changes to the device.
+// Must be called before first using the device (e.g. creating filters).
+OIDN_API void oidnCommitDevice(OIDNDevice device);
+
+// Waits for all asynchronous operations running on the device to complete.
+OIDN_API void oidnSyncDevice(OIDNDevice device);
+
+// -------------------------------------------------------------------------------------------------
+// Buffer
+// -------------------------------------------------------------------------------------------------
+
+// Formats for images and other data stored in buffers
+typedef enum
+{
+ OIDN_FORMAT_UNDEFINED = 0,
+
+ // 32-bit single-precision floating-point scalar and vector formats
+ OIDN_FORMAT_FLOAT = 1,
+ OIDN_FORMAT_FLOAT2,
+ OIDN_FORMAT_FLOAT3,
+ OIDN_FORMAT_FLOAT4,
+
+ // 16-bit half-precision floating-point scalar and vector formats
+ OIDN_FORMAT_HALF = 257,
+ OIDN_FORMAT_HALF2,
+ OIDN_FORMAT_HALF3,
+ OIDN_FORMAT_HALF4,
+} OIDNFormat;
+
+// Storage modes for buffers
+typedef enum
+{
+ OIDN_STORAGE_UNDEFINED = 0,
+
+ // stored on the host, accessible by both host and device
+ OIDN_STORAGE_HOST = 1,
+
+ // stored on the device, *not* accessible by the host
+ OIDN_STORAGE_DEVICE = 2,
+
+ // automatically migrated between host and device, accessible by both
+ // *not* supported by all devices, "managedMemorySupported" device parameter should be checked
+ OIDN_STORAGE_MANAGED = 3,
+} OIDNStorage;
+
+// External memory type flags
+typedef enum
+{
+ OIDN_EXTERNAL_MEMORY_TYPE_FLAG_NONE = 0,
+
+ // opaque POSIX file descriptor handle
+ OIDN_EXTERNAL_MEMORY_TYPE_FLAG_OPAQUE_FD = 1 << 0,
+
+ // file descriptor handle for a Linux dma_buf
+ OIDN_EXTERNAL_MEMORY_TYPE_FLAG_DMA_BUF = 1 << 1,
+
+ // NT handle
+ OIDN_EXTERNAL_MEMORY_TYPE_FLAG_OPAQUE_WIN32 = 1 << 2,
+
+ // global share (KMT) handle
+ OIDN_EXTERNAL_MEMORY_TYPE_FLAG_OPAQUE_WIN32_KMT = 1 << 3,
+
+ // NT handle returned by IDXGIResource1::CreateSharedHandle referring to a Direct3D 11 texture
+ // resource
+ OIDN_EXTERNAL_MEMORY_TYPE_FLAG_D3D11_TEXTURE = 1 << 4,
+
+ // global share (KMT) handle returned by IDXGIResource::GetSharedHandle referring to a Direct3D 11
+ // texture resource
+ OIDN_EXTERNAL_MEMORY_TYPE_FLAG_D3D11_TEXTURE_KMT = 1 << 5,
+
+ // NT handle returned by IDXGIResource1::CreateSharedHandle referring to a Direct3D 11 resource
+ OIDN_EXTERNAL_MEMORY_TYPE_FLAG_D3D11_RESOURCE = 1 << 6,
+
+ // global share (KMT) handle returned by IDXGIResource::GetSharedHandle referring to a Direct3D 11
+ // resource
+ OIDN_EXTERNAL_MEMORY_TYPE_FLAG_D3D11_RESOURCE_KMT = 1 << 7,
+
+ // NT handle returned by ID3D12Device::CreateSharedHandle referring to a Direct3D 12 heap
+ // resource
+ OIDN_EXTERNAL_MEMORY_TYPE_FLAG_D3D12_HEAP = 1 << 8,
+
+ // NT handle returned by ID3D12Device::CreateSharedHandle referring to a Direct3D 12 committed
+ // resource
+ OIDN_EXTERNAL_MEMORY_TYPE_FLAG_D3D12_RESOURCE = 1 << 9,
+} OIDNExternalMemoryTypeFlag;
+
+// Buffer handle
+typedef struct OIDNBufferImpl* OIDNBuffer;
+
+// Creates a buffer accessible to both the host and device.
+OIDN_API OIDNBuffer oidnNewBuffer(OIDNDevice device, size_t byteSize);
+
+// Creates a buffer with the specified storage mode.
+OIDN_API OIDNBuffer oidnNewBufferWithStorage(OIDNDevice device, size_t byteSize, OIDNStorage storage);
+
+// Creates a shared buffer from memory allocated and owned by the user and accessible to the device.
+OIDN_API OIDNBuffer oidnNewSharedBuffer(OIDNDevice device, void* devPtr, size_t byteSize);
+
+// Creates a shared buffer by importing external memory from a POSIX file descriptor.
+OIDN_API OIDNBuffer oidnNewSharedBufferFromFD(OIDNDevice device,
+ OIDNExternalMemoryTypeFlag fdType,
+ int fd, size_t byteSize);
+
+// Creates a shared buffer by importing external memory from a Win32 handle.
+OIDN_API OIDNBuffer oidnNewSharedBufferFromWin32Handle(OIDNDevice device,
+ OIDNExternalMemoryTypeFlag handleType,
+ void* handle, const void* name, size_t byteSize);
+
+// Creates a shared buffer from a Metal buffer.
+// Only buffers with shared or private storage and hazard tracking are supported.
+OIDN_API OIDNBuffer oidnNewSharedBufferFromMetal(OIDNDevice device, MTLBuffer_id buffer);
+
+// Gets the size of the buffer in bytes.
+OIDN_API size_t oidnGetBufferSize(OIDNBuffer buffer);
+
+// Gets the storage mode of the buffer.
+OIDN_API OIDNStorage oidnGetBufferStorage(OIDNBuffer buffer);
+
+// Gets a pointer to the buffer data, which is accessible to the device but not necessarily to
+// the host as well, depending on the storage mode. Null pointer may be returned if the buffer
+// is empty or getting a pointer to data with device storage is not supported by the device.
+OIDN_API void* oidnGetBufferData(OIDNBuffer buffer);
+
+// Copies data from a region of the buffer to host memory.
+OIDN_API void oidnReadBuffer(OIDNBuffer buffer, size_t byteOffset, size_t byteSize, void* dstHostPtr);
+
+// Copies data from a region of the buffer to host memory asynchronously.
+OIDN_API void oidnReadBufferAsync(OIDNBuffer buffer,
+ size_t byteOffset, size_t byteSize, void* dstHostPtr);
+
+// Copies data to a region of the buffer from host memory.
+OIDN_API void oidnWriteBuffer(OIDNBuffer buffer,
+ size_t byteOffset, size_t byteSize, const void* srcHostPtr);
+
+// Copies data to a region of the buffer from host memory asynchronously.
+OIDN_API void oidnWriteBufferAsync(OIDNBuffer buffer,
+ size_t byteOffset, size_t byteSize, const void* srcHostPtr);
+
+// Retains the buffer (increments the reference count).
+OIDN_API void oidnRetainBuffer(OIDNBuffer buffer);
+
+// Releases the buffer (decrements the reference count).
+OIDN_API void oidnReleaseBuffer(OIDNBuffer buffer);
+
+// -------------------------------------------------------------------------------------------------
+// Filter
+// -------------------------------------------------------------------------------------------------
+
+// Filter quality/performance modes
+typedef enum
+{
+ OIDN_QUALITY_DEFAULT = 0, // default quality
+
+ OIDN_QUALITY_FAST = 4, // high performance (for interactive/real-time preview rendering)
+ OIDN_QUALITY_BALANCED = 5, // balanced quality/performance (for interactive/real-time rendering)
+ OIDN_QUALITY_HIGH = 6, // high quality (for final-frame rendering)
+} OIDNQuality;
+
+// Progress monitor callback function
+typedef bool (*OIDNProgressMonitorFunction)(void* userPtr, double n);
+
+// Filter handle
+typedef struct OIDNFilterImpl* OIDNFilter;
+
+// Creates a filter of the specified type (e.g. "RT").
+OIDN_API OIDNFilter oidnNewFilter(OIDNDevice device, const char* type);
+
+// Retains the filter (increments the reference count).
+OIDN_API void oidnRetainFilter(OIDNFilter filter);
+
+// Releases the filter (decrements the reference count).
+OIDN_API void oidnReleaseFilter(OIDNFilter filter);
+
+// Sets an image parameter of the filter with data stored in a buffer.
+// If pixelByteStride and/or rowByteStride are zero, these will be computed automatically.
+OIDN_API void oidnSetFilterImage(OIDNFilter filter, const char* name,
+ OIDNBuffer buffer, OIDNFormat format,
+ size_t width, size_t height,
+ size_t byteOffset,
+ size_t pixelByteStride, size_t rowByteStride);
+
+// Sets an image parameter of the filter with data owned by the user and accessible to the device.
+// If pixelByteStride and/or rowByteStride are zero, these will be computed automatically.
+OIDN_API void oidnSetSharedFilterImage(OIDNFilter filter, const char* name,
+ void* devPtr, OIDNFormat format,
+ size_t width, size_t height,
+ size_t byteOffset,
+ size_t pixelByteStride, size_t rowByteStride);
+
+// Unsets an image parameter of the filter that was previously set.
+OIDN_API void oidnUnsetFilterImage(OIDNFilter filter, const char* name);
+
+OIDN_DEPRECATED("oidnRemoveFilterImage is deprecated. Use oidnUnsetFilterImage instead.")
+inline void oidnRemoveFilterImage(OIDNFilter filter, const char* name)
+{
+ oidnUnsetFilterImage(filter, name);
+}
+
+// Sets an opaque data parameter of the filter owned by the user and accessible to the host.
+OIDN_API void oidnSetSharedFilterData(OIDNFilter filter, const char* name,
+ void* hostPtr, size_t byteSize);
+
+// Notifies the filter that the contents of an opaque data parameter has been changed.
+OIDN_API void oidnUpdateFilterData(OIDNFilter filter, const char* name);
+
+// Unsets an opaque data parameter of the filter that was previously set.
+OIDN_API void oidnUnsetFilterData(OIDNFilter filter, const char* name);
+
+OIDN_DEPRECATED("oidnRemoveFilterData is deprecated. Use oidnUnsetFilterData instead.")
+inline void oidnRemoveFilterData(OIDNFilter filter, const char* name)
+{
+ oidnUnsetFilterData(filter, name);
+}
+
+// Sets a boolean parameter of the filter.
+OIDN_API void oidnSetFilterBool(OIDNFilter filter, const char* name, bool value);
+
+OIDN_DEPRECATED("oidnSetFilter1b is deprecated. Use oidnSetFilterBool instead.")
+inline void oidnSetFilter1b(OIDNFilter filter, const char* name, bool value)
+{
+ oidnSetFilterBool(filter, name, value);
+}
+
+// Gets a boolean parameter of the filter.
+OIDN_API bool oidnGetFilterBool(OIDNFilter filter, const char* name);
+
+OIDN_DEPRECATED("oidnGetFilter1b is deprecated. Use oidnGetFilterBool instead.")
+inline bool oidnGetFilter1b(OIDNFilter filter, const char* name)
+{
+ return oidnGetFilterBool(filter, name);
+}
+
+// Sets an integer parameter of the filter.
+OIDN_API void oidnSetFilterInt(OIDNFilter filter, const char* name, int value);
+
+OIDN_DEPRECATED("oidnSetFilter1i is deprecated. Use oidnSetFilterInt instead.")
+inline void oidnSetFilter1i(OIDNFilter filter, const char* name, int value)
+{
+ oidnSetFilterInt(filter, name, value);
+}
+
+// Gets an integer parameter of the filter.
+OIDN_API int oidnGetFilterInt(OIDNFilter filter, const char* name);
+
+OIDN_DEPRECATED("oidnGetFilter1i is deprecated. Use oidnGetFilterInt instead.")
+inline int oidnGetFilter1i(OIDNFilter filter, const char* name)
+{
+ return oidnGetFilterInt(filter, name);
+}
+
+// Sets a float parameter of the filter.
+OIDN_API void oidnSetFilterFloat(OIDNFilter filter, const char* name, float value);
+
+OIDN_DEPRECATED("oidnSetFilter1f is deprecated. Use oidnSetFilterFloat instead.")
+inline void oidnSetFilter1f(OIDNFilter filter, const char* name, float value)
+{
+ oidnSetFilterFloat(filter, name, value);
+}
+
+// Gets a float parameter of the filter.
+OIDN_API float oidnGetFilterFloat(OIDNFilter filter, const char* name);
+
+OIDN_DEPRECATED("oidnGetFilter1f is deprecated. Use oidnGetFilterFloat instead.")
+inline float oidnGetFilter1f(OIDNFilter filter, const char* name)
+{
+ return oidnGetFilterFloat(filter, name);
+}
+
+// Sets the progress monitor callback function of the filter.
+OIDN_API void oidnSetFilterProgressMonitorFunction(OIDNFilter filter,
+ OIDNProgressMonitorFunction func, void* userPtr);
+
+// Commits all previous changes to the filter.
+// Must be called before first executing the filter.
+OIDN_API void oidnCommitFilter(OIDNFilter filter);
+
+// Executes the filter.
+OIDN_API void oidnExecuteFilter(OIDNFilter filter);
+
+// Executes the filter asynchronously.
+OIDN_API void oidnExecuteFilterAsync(OIDNFilter filter);
+
+#if defined(__cplusplus)
+// Executes the filter of a SYCL device using the specified dependent events asynchronously, and
+// optionally returns an event for completion.
+OIDN_API void oidnExecuteSYCLFilterAsync(OIDNFilter filter,
+ const sycl::event* depEvents, int numDepEvents,
+ sycl::event* doneEvent);
+#endif
+
+OIDN_API_NAMESPACE_END
diff --git a/external/include/OpenImageDenoise/oidn.hpp b/external/include/OpenImageDenoise/oidn.hpp
new file mode 100644
index 00000000..4588d421
--- /dev/null
+++ b/external/include/OpenImageDenoise/oidn.hpp
@@ -0,0 +1,1082 @@
+// Copyright 2018 Intel Corporation
+// SPDX-License-Identifier: Apache-2.0
+
+#pragma once
+
+#include "oidn.h"
+#include
+#include
+#include
+#include
+#include
+#include
+#include
+#include
+
+OIDN_NAMESPACE_BEGIN
+
+ // -----------------------------------------------------------------------------------------------
+ // Flags helper type
+ // -----------------------------------------------------------------------------------------------
+
+ template
+ struct IsFlag
+ {
+ static constexpr bool value = false;
+ };
+
+ template
+ class Flags
+ {
+ public:
+ static_assert(IsFlag::value, "not a flag type");
+
+ using MaskType = typename std::underlying_type::type;
+
+ constexpr Flags() noexcept : mask(0) {}
+ constexpr Flags(FlagT flag) noexcept : mask(static_cast(flag)) {}
+ constexpr Flags(const Flags& b) noexcept = default;
+ constexpr explicit Flags(MaskType mask) noexcept : mask(mask) {}
+
+ constexpr bool operator !() const noexcept { return !mask; }
+
+ constexpr Flags operator &(const Flags& b) const noexcept { return Flags(mask & b.mask); }
+ constexpr Flags operator |(const Flags& b) const noexcept { return Flags(mask | b.mask); }
+ constexpr Flags operator ^(const Flags& b) const noexcept { return Flags(mask ^ b.mask); }
+
+ Flags& operator =(const Flags& b) noexcept = default;
+
+ Flags& operator &=(const Flags& b) noexcept
+ {
+ mask &= b.mask;
+ return *this;
+ }
+
+ Flags& operator |=(const Flags& b) noexcept
+ {
+ mask |= b.mask;
+ return *this;
+ }
+
+ Flags& operator ^=(const Flags& b) noexcept
+ {
+ mask ^= b.mask;
+ return *this;
+ }
+
+ constexpr bool operator ==(const Flags& b) const noexcept { return mask == b.mask; }
+ constexpr bool operator !=(const Flags& b) const noexcept { return mask != b.mask; }
+
+ constexpr explicit operator bool() const noexcept { return mask; }
+ constexpr explicit operator MaskType() const noexcept { return mask; }
+
+ private:
+ MaskType mask;
+ };
+
+ template
+ inline constexpr Flags operator &(FlagT a, const Flags& b) noexcept
+ {
+ return Flags(a) & b;
+ }
+
+ template
+ inline constexpr Flags operator |(FlagT a, const Flags& b) noexcept
+ {
+ return Flags(a) | b;
+ }
+
+ template
+ inline constexpr Flags operator ^(FlagT a, const Flags& b) noexcept
+ {
+ return Flags(a) ^ b;
+ }
+
+ template::value, bool>::type = true>
+ inline constexpr Flags operator &(FlagT a, FlagT b) noexcept
+ {
+ return Flags(a) & b;
+ }
+
+ template::value, bool>::type = true>
+ inline constexpr Flags operator |(FlagT a, FlagT b) noexcept
+ {
+ return Flags(a) | b;
+ }
+
+ template::value, bool>::type = true>
+ inline constexpr Flags operator ^(FlagT a, FlagT b) noexcept
+ {
+ return Flags(a) ^ b;
+ }
+
+ // -----------------------------------------------------------------------------------------------
+ // Buffer
+ // -----------------------------------------------------------------------------------------------
+
+ // Formats for images and other data stored in buffers
+ enum class Format
+ {
+ Undefined = OIDN_FORMAT_UNDEFINED,
+
+ // 32-bit single-precision floating-point scalar and vector formats
+ Float = OIDN_FORMAT_FLOAT,
+ Float2 = OIDN_FORMAT_FLOAT2,
+ Float3 = OIDN_FORMAT_FLOAT3,
+ Float4 = OIDN_FORMAT_FLOAT4,
+
+ // 16-bit half-precision floating-point scalar and vector formats
+ Half = OIDN_FORMAT_HALF,
+ Half2 = OIDN_FORMAT_HALF2,
+ Half3 = OIDN_FORMAT_HALF3,
+ Half4 = OIDN_FORMAT_HALF4,
+ };
+
+ // Storage modes for buffers
+ enum class Storage
+ {
+ Undefined = OIDN_STORAGE_UNDEFINED,
+
+ // stored on the host, accessible by both host and device
+ Host = OIDN_STORAGE_HOST,
+
+ // stored on the device, *not* accessible by the host
+ Device = OIDN_STORAGE_DEVICE,
+
+ // automatically migrated between host and device, accessible by both
+ // *not* supported by all devices, "managedMemorySupported" device parameter should be checked
+ Managed = OIDN_STORAGE_MANAGED,
+ };
+
+ // External memory type flags
+ enum class ExternalMemoryTypeFlag
+ {
+ None = OIDN_EXTERNAL_MEMORY_TYPE_FLAG_NONE,
+
+ // opaque POSIX file descriptor handle
+ OpaqueFD = OIDN_EXTERNAL_MEMORY_TYPE_FLAG_OPAQUE_FD,
+
+ // file descriptor handle for a Linux dma_buf
+ DMABuf = OIDN_EXTERNAL_MEMORY_TYPE_FLAG_DMA_BUF,
+
+ // NT handle
+ OpaqueWin32 = OIDN_EXTERNAL_MEMORY_TYPE_FLAG_OPAQUE_WIN32,
+
+ // global share (KMT) handle
+ OpaqueWin32KMT = OIDN_EXTERNAL_MEMORY_TYPE_FLAG_OPAQUE_WIN32_KMT,
+
+ // NT handle returned by IDXGIResource1::CreateSharedHandle referring to a Direct3D 11
+ // texture resource
+ D3D11Texture = OIDN_EXTERNAL_MEMORY_TYPE_FLAG_D3D11_TEXTURE,
+
+ // global share (KMT) handle returned by IDXGIResource::GetSharedHandle referring to a
+ // Direct3D 11 texture resource
+ D3D11TextureKMT = OIDN_EXTERNAL_MEMORY_TYPE_FLAG_D3D11_TEXTURE_KMT,
+
+ // NT handle returned by IDXGIResource1::CreateSharedHandle referring to a Direct3D 11
+ // resource
+ D3D11Resource = OIDN_EXTERNAL_MEMORY_TYPE_FLAG_D3D11_RESOURCE,
+
+ // global share (KMT) handle returned by IDXGIResource::GetSharedHandle referring to a
+ // Direct3D 11 resource
+ D3D11ResourceKMT = OIDN_EXTERNAL_MEMORY_TYPE_FLAG_D3D11_RESOURCE_KMT,
+
+ // NT handle returned by ID3D12Device::CreateSharedHandle referring to a Direct3D 12
+ // heap resource
+ D3D12Heap = OIDN_EXTERNAL_MEMORY_TYPE_FLAG_D3D12_HEAP,
+
+ // NT handle returned by ID3D12Device::CreateSharedHandle referring to a Direct3D 12
+ // committed resource
+ D3D12Resource = OIDN_EXTERNAL_MEMORY_TYPE_FLAG_D3D12_RESOURCE,
+ };
+
+ template<> struct IsFlag { static constexpr bool value = true; };
+ using ExternalMemoryTypeFlags = Flags;
+
+ // Buffer object with automatic reference counting
+ class BufferRef
+ {
+ public:
+ BufferRef() : handle(nullptr) {}
+ BufferRef(OIDNBuffer handle) : handle(handle) {}
+
+ BufferRef(const BufferRef& other) : handle(other.handle)
+ {
+ if (handle)
+ oidnRetainBuffer(handle);
+ }
+
+ BufferRef(BufferRef&& other) noexcept : handle(other.handle)
+ {
+ other.handle = nullptr;
+ }
+
+ BufferRef& operator =(const BufferRef& other)
+ {
+ if (&other != this)
+ {
+ if (other.handle)
+ oidnRetainBuffer(other.handle);
+ if (handle)
+ oidnReleaseBuffer(handle);
+ handle = other.handle;
+ }
+ return *this;
+ }
+
+ BufferRef& operator =(BufferRef&& other) noexcept
+ {
+ std::swap(handle, other.handle);
+ return *this;
+ }
+
+ BufferRef& operator =(OIDNBuffer other)
+ {
+ if (other)
+ oidnRetainBuffer(other);
+ if (handle)
+ oidnReleaseBuffer(handle);
+ handle = other;
+ return *this;
+ }
+
+ ~BufferRef()
+ {
+ if (handle)
+ oidnReleaseBuffer(handle);
+ }
+
+ OIDNBuffer getHandle() const
+ {
+ return handle;
+ }
+
+ operator bool() const
+ {
+ return handle != nullptr;
+ }
+
+ // Releases the buffer (decrements the reference count).
+ void release()
+ {
+ if (handle)
+ {
+ oidnReleaseBuffer(handle);
+ handle = nullptr;
+ }
+ }
+
+ // Gets the size of the buffer in bytes.
+ size_t getSize() const
+ {
+ return oidnGetBufferSize(handle);
+ }
+
+ // Gets the storage mode of the buffer.
+ Storage getStorage() const
+ {
+ return static_cast(oidnGetBufferStorage(handle));
+ }
+
+ // Gets a pointer to the buffer data, which is accessible to the device but not necessarily to
+ // the host as well, depending on the storage mode. Null pointer may be returned if the buffer
+ // is empty or getting a pointer to data with device storage is not supported by the device.
+ void* getData() const
+ {
+ return oidnGetBufferData(handle);
+ }
+
+ // Copies data from a region of the buffer to host memory.
+ void read(size_t byteOffset, size_t byteSize, void* dstHostPtr) const
+ {
+ oidnReadBuffer(handle, byteOffset, byteSize, dstHostPtr);
+ }
+
+ // Copies data from a region of the buffer to host memory asynchronously.
+ void readAsync(size_t byteOffset, size_t byteSize, void* dstHostPtr) const
+ {
+ oidnReadBufferAsync(handle, byteOffset, byteSize, dstHostPtr);
+ }
+
+ // Copies data to a region of the buffer from host memory.
+ void write(size_t byteOffset, size_t byteSize, const void* srcHostPtr)
+ {
+ oidnWriteBuffer(handle, byteOffset, byteSize, srcHostPtr);
+ }
+
+ // Copies data to a region of the buffer from host memory asynchronously.
+ void writeAsync(size_t byteOffset, size_t byteSize, const void* srcHostPtr)
+ {
+ oidnWriteBufferAsync(handle, byteOffset, byteSize, srcHostPtr);
+ }
+
+ private:
+ OIDNBuffer handle;
+ };
+
+ // -----------------------------------------------------------------------------------------------
+ // Filter
+ // -----------------------------------------------------------------------------------------------
+
+ // Filter quality/performance modes
+ enum class Quality
+ {
+ Default = OIDN_QUALITY_DEFAULT, // default quality
+
+ Fast = OIDN_QUALITY_FAST, // high performance (for interactive/real-time preview rendering)
+ Balanced = OIDN_QUALITY_BALANCED, // balanced quality/performance (for interactive/real-time rendering)
+ High = OIDN_QUALITY_HIGH, // high quality (for final-frame rendering)
+ };
+
+ // Progress monitor callback function
+ using ProgressMonitorFunction = OIDNProgressMonitorFunction;
+
+ // Filter object with automatic reference counting
+ class FilterRef
+ {
+ public:
+ FilterRef() : handle(nullptr) {}
+ FilterRef(OIDNFilter handle) : handle(handle) {}
+
+ FilterRef(const FilterRef& other) : handle(other.handle)
+ {
+ if (handle)
+ oidnRetainFilter(handle);
+ }
+
+ FilterRef(FilterRef&& other) noexcept : handle(other.handle)
+ {
+ other.handle = nullptr;
+ }
+
+ FilterRef& operator =(const FilterRef& other)
+ {
+ if (&other != this)
+ {
+ if (other.handle)
+ oidnRetainFilter(other.handle);
+ if (handle)
+ oidnReleaseFilter(handle);
+ handle = other.handle;
+ }
+ return *this;
+ }
+
+ FilterRef& operator =(FilterRef&& other) noexcept
+ {
+ std::swap(handle, other.handle);
+ return *this;
+ }
+
+ FilterRef& operator =(OIDNFilter other)
+ {
+ if (other)
+ oidnRetainFilter(other);
+ if (handle)
+ oidnReleaseFilter(handle);
+ handle = other;
+ return *this;
+ }
+
+ ~FilterRef()
+ {
+ if (handle)
+ oidnReleaseFilter(handle);
+ }
+
+ OIDNFilter getHandle() const
+ {
+ return handle;
+ }
+
+ operator bool() const
+ {
+ return handle != nullptr;
+ }
+
+ // Releases the filter (decrements the reference count).
+ void release()
+ {
+ if (handle)
+ {
+ oidnReleaseFilter(handle);
+ handle = nullptr;
+ }
+ }
+
+ // Sets an image parameter of the filter with data stored in a buffer.
+ void setImage(const char* name,
+ const BufferRef& buffer, Format format,
+ size_t width, size_t height,
+ size_t byteOffset = 0,
+ size_t pixelByteStride = 0, size_t rowByteStride = 0)
+ {
+ oidnSetFilterImage(handle, name,
+ buffer.getHandle(), static_cast(format),
+ width, height,
+ byteOffset,
+ pixelByteStride, rowByteStride);
+ }
+
+ // Sets an image parameter of the filter with data owned by the user and accessible to the device.
+ void setImage(const char* name,
+ void* devPtr, Format format,
+ size_t width, size_t height,
+ size_t byteOffset = 0,
+ size_t pixelByteStride = 0, size_t rowByteStride = 0)
+ {
+ oidnSetSharedFilterImage(handle, name,
+ devPtr, static_cast(format),
+ width, height,
+ byteOffset,
+ pixelByteStride, rowByteStride);
+ }
+
+ // Unsets an image parameter of the filter that was previously set.
+ void unsetImage(const char* name)
+ {
+ oidnUnsetFilterImage(handle, name);
+ }
+
+ OIDN_DEPRECATED("removeImage is deprecated. Use unsetImage instead.")
+ void removeImage(const char* name)
+ {
+ oidnUnsetFilterImage(handle, name);
+ }
+
+ // Sets an opaque data parameter of the filter owned by the user and accessible to the host.
+ void setData(const char* name, void* hostPtr, size_t byteSize)
+ {
+ oidnSetSharedFilterData(handle, name, hostPtr, byteSize);
+ }
+
+ // Notifies the filter that the contents of an opaque data parameter has been changed.
+ void updateData(const char* name)
+ {
+ oidnUpdateFilterData(handle, name);
+ }
+
+ // Unsets an opaque data parameter of the filter that was previously set.
+ void unsetData(const char* name)
+ {
+ oidnUnsetFilterData(handle, name);
+ }
+
+ OIDN_DEPRECATED("removeData is deprecated. Use unsetData instead.")
+ void removeData(const char* name)
+ {
+ oidnUnsetFilterData(handle, name);
+ }
+
+ // Sets a boolean parameter of the filter.
+ void set(const char* name, bool value)
+ {
+ oidnSetFilterBool(handle, name, value);
+ }
+
+ // Sets an integer parameter of the filter.
+ void set(const char* name, int value)
+ {
+ oidnSetFilterInt(handle, name, value);
+ }
+
+ void set(const char* name, Quality value)
+ {
+ oidnSetFilterInt(handle, name, static_cast(value));
+ }
+
+ // Sets a float parameter of the filter.
+ void set(const char* name, float value)
+ {
+ oidnSetFilterFloat(handle, name, value);
+ }
+
+ // Gets a parameter of the filter.
+ template
+ T get(const char* name) const;
+
+ // Sets the progress monitor callback function of the filter.
+ void setProgressMonitorFunction(ProgressMonitorFunction func, void* userPtr = nullptr)
+ {
+ oidnSetFilterProgressMonitorFunction(handle, func, userPtr);
+ }
+
+ // Commits all previous changes to the filter.
+ void commit()
+ {
+ oidnCommitFilter(handle);
+ }
+
+ // Executes the filter.
+ void execute()
+ {
+ oidnExecuteFilter(handle);
+ }
+
+ // Executes the filter asynchronously.
+ void executeAsync()
+ {
+ oidnExecuteFilterAsync(handle);
+ }
+
+ #if defined(OIDN_SYCL_HPP)
+ // Executes the filter of a SYCL device using the specified dependent events asynchronously, and
+ // optionally returns an event for completion.
+ sycl::event executeAsync(const std::vector& depEvents)
+ {
+ sycl::event doneEvent;
+ oidnExecuteSYCLFilterAsync(handle, depEvents.data(), static_cast(depEvents.size()), &doneEvent);
+ return doneEvent;
+ }
+ #endif
+
+ private:
+ OIDNFilter handle;
+ };
+
+ template<>
+ inline bool FilterRef::get(const char* name) const
+ {
+ return oidnGetFilterBool(handle, name);
+ }
+
+ template<>
+ inline int FilterRef::get(const char* name) const
+ {
+ return oidnGetFilterInt(handle, name);
+ }
+
+ template<>
+ inline Quality FilterRef::get(const char* name) const
+ {
+ return static_cast(oidnGetFilterInt(handle, name));
+ }
+
+ template<>
+ inline float FilterRef::get(const char* name) const
+ {
+ return oidnGetFilterFloat(handle, name);
+ }
+
+ // -----------------------------------------------------------------------------------------------
+ // Device
+ // -----------------------------------------------------------------------------------------------
+
+ // Device types
+ enum class DeviceType
+ {
+ Default = OIDN_DEVICE_TYPE_DEFAULT, // select device automatically
+
+ CPU = OIDN_DEVICE_TYPE_CPU, // CPU device
+ SYCL = OIDN_DEVICE_TYPE_SYCL, // SYCL device
+ CUDA = OIDN_DEVICE_TYPE_CUDA, // CUDA device
+ HIP = OIDN_DEVICE_TYPE_HIP, // HIP device
+ Metal = OIDN_DEVICE_TYPE_METAL, // Metal device
+ };
+
+ // Error codes
+ enum class Error
+ {
+ None = OIDN_ERROR_NONE, // no error occurred
+ Unknown = OIDN_ERROR_UNKNOWN, // an unknown error occurred
+ InvalidArgument = OIDN_ERROR_INVALID_ARGUMENT, // an invalid argument was specified
+ InvalidOperation = OIDN_ERROR_INVALID_OPERATION, // the operation is not allowed
+ OutOfMemory = OIDN_ERROR_OUT_OF_MEMORY, // not enough memory to execute the operation
+ UnsupportedHardware = OIDN_ERROR_UNSUPPORTED_HARDWARE, // the hardware (e.g. CPU) is not supported
+ Cancelled = OIDN_ERROR_CANCELLED, // the operation was cancelled by the user
+ };
+
+ // Error callback function
+ typedef void (*ErrorFunction)(void* userPtr, Error code, const char* message);
+
+ // Opaque universally unique identifier (UUID) of a physical device
+ struct UUID
+ {
+ uint8_t bytes[OIDN_UUID_SIZE];
+ };
+
+ // Opaque locally unique identifier (LUID) of a physical device
+ struct LUID
+ {
+ union
+ {
+ struct
+ {
+ uint32_t low;
+ int32_t high;
+ };
+ uint8_t bytes[OIDN_LUID_SIZE];
+ };
+ };
+
+ // Device object with automatic reference counting
+ class DeviceRef
+ {
+ public:
+ DeviceRef() : handle(nullptr) {}
+ DeviceRef(OIDNDevice handle) : handle(handle) {}
+
+ DeviceRef(const DeviceRef& other) : handle(other.handle)
+ {
+ if (handle)
+ oidnRetainDevice(handle);
+ }
+
+ DeviceRef(DeviceRef&& other) noexcept : handle(other.handle)
+ {
+ other.handle = nullptr;
+ }
+
+ DeviceRef& operator =(const DeviceRef& other)
+ {
+ if (&other != this)
+ {
+ if (other.handle)
+ oidnRetainDevice(other.handle);
+ if (handle)
+ oidnReleaseDevice(handle);
+ handle = other.handle;
+ }
+ return *this;
+ }
+
+ DeviceRef& operator =(DeviceRef&& other) noexcept
+ {
+ std::swap(handle, other.handle);
+ return *this;
+ }
+
+ DeviceRef& operator =(OIDNDevice other)
+ {
+ if (other)
+ oidnRetainDevice(other);
+ if (handle)
+ oidnReleaseDevice(handle);
+ handle = other;
+ return *this;
+ }
+
+ ~DeviceRef()
+ {
+ if (handle)
+ oidnReleaseDevice(handle);
+ }
+
+ OIDNDevice getHandle() const
+ {
+ return handle;
+ }
+
+ operator bool() const
+ {
+ return handle != nullptr;
+ }
+
+ // Releases the device (decrements the reference count).
+ void release()
+ {
+ if (handle)
+ {
+ oidnReleaseDevice(handle);
+ handle = nullptr;
+ }
+ }
+
+ // Sets a boolean parameter of the device.
+ void set(const char* name, bool value)
+ {
+ oidnSetDeviceBool(handle, name, value);
+ }
+
+ // Sets an integer parameter of the device.
+ void set(const char* name, int value)
+ {
+ oidnSetDeviceInt(handle, name, value);
+ }
+
+ // Sets an unsigned integer parameter of the device.
+ void set(const char* name, unsigned int value)
+ {
+ oidnSetDeviceUInt(handle, name, value);
+ }
+
+ // Gets a parameter of the device.
+ template
+ T get(const char* name) const;
+
+ // Sets the error callback function of the device.
+ void setErrorFunction(ErrorFunction func, void* userPtr = nullptr)
+ {
+ oidnSetDeviceErrorFunction(handle, reinterpret_cast(func), userPtr);
+ }
+
+ // Returns the first unqueried error code and clears the stored error.
+ // Can be called for a null device as well to check for global errors (e.g. why a device
+ // creation or physical device query has failed.
+ Error getError()
+ {
+ return static_cast(oidnGetDeviceError(handle, nullptr));
+ }
+
+ // Returns the first unqueried error code and string message, and clears the stored error.
+ // Can be called for a null device as well to check why a device creation failed.
+ Error getError(const char*& outMessage)
+ {
+ return static_cast(oidnGetDeviceError(handle, &outMessage));
+ }
+
+ // Commits all previous changes to the device.
+ // Must be called before first using the device (e.g. creating filters).
+ void commit()
+ {
+ oidnCommitDevice(handle);
+ }
+
+ // Waits for all asynchronous operations running on the device to complete.
+ void sync()
+ {
+ oidnSyncDevice(handle);
+ }
+
+ // Creates a buffer accessible to both the host and device.
+ BufferRef newBuffer(size_t byteSize) const
+ {
+ return oidnNewBuffer(handle, byteSize);
+ }
+
+ // Creates a buffer with the specified storage mode.
+ BufferRef newBuffer(size_t byteSize, Storage storage) const
+ {
+ return oidnNewBufferWithStorage(handle, byteSize, static_cast(storage));
+ }
+
+ // Creates a shared buffer from memory allocated and owned by the user and accessible to the
+ // device.
+ BufferRef newBuffer(void* ptr, size_t byteSize) const
+ {
+ return oidnNewSharedBuffer(handle, ptr, byteSize);
+ }
+
+ // Creates a shared buffer by importing external memory from a POSIX file descriptor.
+ BufferRef newBuffer(ExternalMemoryTypeFlag fdType, int fd, size_t byteSize) const
+ {
+ return oidnNewSharedBufferFromFD(
+ handle, static_cast(fdType), fd, byteSize);
+ }
+
+ // Creates a shared buffer by importing external memory from a Win32 handle.
+ BufferRef newBuffer(ExternalMemoryTypeFlag handleType, void* handle, const void* name, size_t byteSize) const
+ {
+ return oidnNewSharedBufferFromWin32Handle(
+ this->handle, static_cast(handleType), handle, name, byteSize);
+ }
+
+ // Creates a shared buffer from a Metal buffer.
+ // Only buffers with shared or private storage and hazard tracking are supported.
+ #if defined(__OBJC__)
+ BufferRef newBuffer(id buffer) const
+ {
+ return oidnNewSharedBufferFromMetal(handle, buffer);
+ }
+ #endif
+
+ // Creates a filter of the specified type (e.g. "RT").
+ FilterRef newFilter(const char* type) const
+ {
+ return oidnNewFilter(handle, type);
+ }
+
+ private:
+ OIDNDevice handle;
+ };
+
+ template<>
+ inline bool DeviceRef::get(const char* name) const
+ {
+ return oidnGetDeviceBool(handle, name);
+ }
+
+ template<>
+ inline int DeviceRef::get(const char* name) const
+ {
+ return oidnGetDeviceInt(handle, name);
+ }
+
+ template<>
+ inline unsigned int DeviceRef::get(const char* name) const
+ {
+ return oidnGetDeviceUInt(handle, name);
+ }
+
+ template<>
+ inline DeviceType DeviceRef::get(const char* name) const
+ {
+ return static_cast(oidnGetDeviceInt(handle, name));
+ }
+
+ template<>
+ inline ExternalMemoryTypeFlags DeviceRef::get(const char* name) const
+ {
+ return ExternalMemoryTypeFlags(oidnGetDeviceInt(handle, name));
+ }
+
+ // Returns the first unqueried per-thread global error code and clears the stored error.
+ inline Error getError()
+ {
+ return static_cast(oidnGetDeviceError(nullptr, nullptr));
+ }
+
+ // Returns the first unqueried per-thread global error code and string message, and clears the
+ // stored error.
+ inline Error getError(const char*& outMessage)
+ {
+ return static_cast(oidnGetDeviceError(nullptr, &outMessage));
+ }
+
+ // Returns whether the CPU device is supported.
+ inline bool isCPUDeviceSupported()
+ {
+ return oidnIsCPUDeviceSupported();
+ }
+
+ // Returns whether the specified SYCL device is supported.
+#if defined(OIDN_SYCL_HPP)
+ inline bool isSYCLDeviceSupported(const sycl::device& device)
+ {
+ return oidnIsSYCLDeviceSupported(&device);
+ }
+#endif
+
+ // Returns whether the specified CUDA device is supported.
+ inline bool isCUDADeviceSupported(int deviceID)
+ {
+ return oidnIsCUDADeviceSupported(deviceID);
+ }
+
+ // Returns whether the specified HIP device is supported.
+ inline bool isHIPDeviceSupported(int deviceID)
+ {
+ return oidnIsHIPDeviceSupported(deviceID);
+ }
+
+ // Returns whether the specified Metal device is supported.
+ inline bool isMetalDeviceSupported(MTLDevice_id device)
+ {
+ return oidnIsMetalDeviceSupported(device);
+ }
+
+ // Creates a device of the specified type.
+ inline DeviceRef newDevice(DeviceType type = DeviceType::Default)
+ {
+ return DeviceRef(oidnNewDevice(static_cast(type)));
+ }
+
+ // Creates a device from a physical device specified by its ID (0 to getNumPhysicalDevices()-1).
+ inline DeviceRef newDevice(int physicalDeviceID)
+ {
+ return DeviceRef(oidnNewDeviceByID(physicalDeviceID));
+ }
+
+ // Creates a device from a physical device specified by its UUID.
+ inline DeviceRef newDevice(const UUID& uuid)
+ {
+ return DeviceRef(oidnNewDeviceByUUID(uuid.bytes));
+ }
+
+ // Creates a device from a physical device specified by its LUID.
+ inline DeviceRef newDevice(const LUID& luid)
+ {
+ return DeviceRef(oidnNewDeviceByLUID(luid.bytes));
+ }
+
+ // Creates a device from a physical device specified by its PCI address.
+ inline DeviceRef newDevice(int pciDomain, int pciBus, int pciDevice, int pciFunction)
+ {
+ return DeviceRef(oidnNewDeviceByPCIAddress(pciDomain, pciBus, pciDevice, pciFunction));
+ }
+
+#if defined(OIDN_SYCL_HPP)
+ // Creates a device from the specified SYCL queue.
+ inline DeviceRef newSYCLDevice(const sycl::queue& queue)
+ {
+ return DeviceRef(oidnNewSYCLDevice(&queue, 1));
+ }
+
+ // Creates a device from the specified list of SYCL queues.
+ // The queues should belong to different SYCL sub-devices (Xe Stack/Tile) of the same SYCL
+ // root-device (GPU).
+ inline DeviceRef newSYCLDevice(const std::vector& queues)
+ {
+ return DeviceRef(oidnNewSYCLDevice(queues.data(), static_cast(queues.size())));
+ }
+#endif
+
+ // Creates a device from the specified CUDA device ID and stream (null stream corresponds to the
+ // default stream).
+ inline DeviceRef newCUDADevice(int deviceID, cudaStream_t stream)
+ {
+ return DeviceRef(oidnNewCUDADevice(&deviceID, &stream, 1));
+ }
+
+ // Creates a device from the specified pairs of CUDA device IDs and streams (null stream
+ // corresponds to the default stream). Currently only one device ID/stream is supported.
+ inline DeviceRef newCUDADevice(const std::vector& deviceIDs,
+ const std::vector& streams)
+ {
+ assert(deviceIDs.size() == streams.size());
+ return DeviceRef(oidnNewCUDADevice(deviceIDs.data(), streams.data(),
+ static_cast(streams.size())));
+ }
+
+ // Creates a device from the specified HIP device ID and stream (null stream corresponds to the
+ // default stream).
+ inline DeviceRef newHIPDevice(int deviceID, hipStream_t stream)
+ {
+ return DeviceRef(oidnNewHIPDevice(&deviceID, &stream, 1));
+ }
+
+ // Creates a device from the specified pairs of HIP device IDs and streams (null stream
+ // corresponds to the default stream). Currently only one device ID/stream is supported.
+ inline DeviceRef newHIPDevice(const std::vector& deviceIDs,
+ const std::vector& streams)
+ {
+ assert(deviceIDs.size() == streams.size());
+ return DeviceRef(oidnNewHIPDevice(deviceIDs.data(), streams.data(),
+ static_cast(streams.size())));
+ }
+
+ // Creates a device from the specified Metal command queue.
+ inline DeviceRef newMetalDevice(MTLCommandQueue_id commandQueue)
+ {
+ return DeviceRef(oidnNewMetalDevice(&commandQueue, 1));
+ }
+
+ // Creates a device from the specified list of Metal command queues.
+ // Currently only one queue is supported.
+ inline DeviceRef newMetalDevice(const std::vector& commandQueues)
+ {
+ return DeviceRef(oidnNewMetalDevice(commandQueues.data(), static_cast(commandQueues.size())));
+ }
+
+ // -----------------------------------------------------------------------------------------------
+ // Physical Device
+ // -----------------------------------------------------------------------------------------------
+
+ class PhysicalDeviceRef
+ {
+ public:
+ PhysicalDeviceRef() : id(-1) {}
+ PhysicalDeviceRef(int id) : id(id) {}
+
+ PhysicalDeviceRef& operator =(int other)
+ {
+ id = other;
+ return *this;
+ }
+
+ int getID() const
+ {
+ return id;
+ }
+
+ operator bool() const
+ {
+ return id >= 0;
+ }
+
+ // Gets a paramter of the physical device.
+ template
+ T get(const char* name) const;
+
+ // Gets an opaque data parameter of the physical device.
+ std::pair getData(const char* name) const
+ {
+ size_t byteSize = 0;
+ const void* ptr = oidnGetPhysicalDeviceData(id, name, &byteSize);
+ return {ptr, byteSize};
+ }
+
+ // Creates a device from the physical device.
+ DeviceRef newDevice()
+ {
+ return DeviceRef(oidnNewDeviceByID(id));
+ }
+
+ private:
+ int id;
+ };
+
+ // Returns the number of supported physical devices.
+ inline int getNumPhysicalDevices()
+ {
+ return oidnGetNumPhysicalDevices();
+ }
+
+ template<>
+ inline bool PhysicalDeviceRef::get(const char* name) const
+ {
+ return oidnGetPhysicalDeviceBool(id, name);
+ }
+
+ template<>
+ inline int PhysicalDeviceRef::get(const char* name) const
+ {
+ return oidnGetPhysicalDeviceInt(id, name);
+ }
+
+ template<>
+ inline unsigned int PhysicalDeviceRef::get(const char* name) const
+ {
+ return oidnGetPhysicalDeviceUInt(id, name);
+ }
+
+ template<>
+ inline DeviceType PhysicalDeviceRef::get(const char* name) const
+ {
+ return static_cast(oidnGetPhysicalDeviceInt(id, name));
+ }
+
+ template<>
+ inline const char* PhysicalDeviceRef::get(const char* name) const
+ {
+ return oidnGetPhysicalDeviceString(id, name);
+ }
+
+ template<>
+ inline std::string PhysicalDeviceRef::get(const char* name) const
+ {
+ const char* str = oidnGetPhysicalDeviceString(id, name);
+ return str ? str : "";
+ }
+
+ template<>
+ inline UUID PhysicalDeviceRef::get(const char* name) const
+ {
+ UUID uuid{};
+ auto data = getData(name);
+ if (data.first != nullptr)
+ {
+ if (data.second == sizeof(uuid.bytes))
+ std::memcpy(uuid.bytes, data.first, sizeof(uuid.bytes));
+ else
+ getData(""); // invoke an error
+ }
+ return uuid;
+ }
+
+ template<>
+ inline LUID PhysicalDeviceRef::get(const char* name) const
+ {
+ LUID luid{};
+ auto data = getData(name);
+ if (data.first != nullptr)
+ {
+ if (data.second == sizeof(luid.bytes))
+ std::memcpy(luid.bytes, data.first, sizeof(luid.bytes));
+ else
+ getData(""); // invoke an error
+ }
+ return luid;
+ }
+
+OIDN_NAMESPACE_END
diff --git a/external/include/stb_image.h b/external/include/stb_image.h
index b9b265fa..9eedabed 100644
--- a/external/include/stb_image.h
+++ b/external/include/stb_image.h
@@ -1,5 +1,5 @@
-/* stb_image - v2.06 - public domain image loader - http://nothings.org/stb_image.h
- no warranty implied; use at your own risk
+/* stb_image - v2.30 - public domain image loader - http://nothings.org/stb
+ no warranty implied; use at your own risk
Do this:
#define STB_IMAGE_IMPLEMENTATION
@@ -21,17 +21,20 @@
avoid problematic images and only need the trivial interface
JPEG baseline & progressive (12 bpc/arithmetic not supported, same as stock IJG lib)
- PNG 1/2/4/8-bit-per-channel (16 bpc not supported)
+ PNG 1/2/4/8/16-bit-per-channel
TGA (not sure what subset, if a subset)
BMP non-1bpp, non-RLE
- PSD (composited view only, no extra channels)
+ PSD (composited view only, no extra channels, 8/16 bit-per-channel)
GIF (*comp always reports as 4-channel)
HDR (radiance rgbE format)
PIC (Softimage PIC)
PNM (PPM and PGM binary only)
+ Animated GIF still needs a proper API, but here's one way to do it:
+ http://gist.github.com/urraka/685d9a6340b26b830d49
+
- decode from memory or through FILE (define STBI_NO_STDIO to remove code)
- decode from arbitrary I/O callbacks
- SIMD acceleration on x86/x64 (SSE2) and ARM (NEON)
@@ -39,177 +42,88 @@
Full documentation under "DOCUMENTATION" below.
- Revision 2.00 release notes:
-
- - Progressive JPEG is now supported.
-
- - PPM and PGM binary formats are now supported, thanks to Ken Miller.
-
- - x86 platforms now make use of SSE2 SIMD instructions for
- JPEG decoding, and ARM platforms can use NEON SIMD if requested.
- This work was done by Fabian "ryg" Giesen. SSE2 is used by
- default, but NEON must be enabled explicitly; see docs.
-
- With other JPEG optimizations included in this version, we see
- 2x speedup on a JPEG on an x86 machine, and a 1.5x speedup
- on a JPEG on an ARM machine, relative to previous versions of this
- library. The same results will not obtain for all JPGs and for all
- x86/ARM machines. (Note that progressive JPEGs are significantly
- slower to decode than regular JPEGs.) This doesn't mean that this
- is the fastest JPEG decoder in the land; rather, it brings it
- closer to parity with standard libraries. If you want the fastest
- decode, look elsewhere. (See "Philosophy" section of docs below.)
-
- See final bullet items below for more info on SIMD.
-
- - Added STBI_MALLOC, STBI_REALLOC, and STBI_FREE macros for replacing
- the memory allocator. Unlike other STBI libraries, these macros don't
- support a context parameter, so if you need to pass a context in to
- the allocator, you'll have to store it in a global or a thread-local
- variable.
-
- - Split existing STBI_NO_HDR flag into two flags, STBI_NO_HDR and
- STBI_NO_LINEAR.
- STBI_NO_HDR: suppress implementation of .hdr reader format
- STBI_NO_LINEAR: suppress high-dynamic-range light-linear float API
-
- - You can suppress implementation of any of the decoders to reduce
- your code footprint by #defining one or more of the following
- symbols before creating the implementation.
-
- STBI_NO_JPEG
- STBI_NO_PNG
- STBI_NO_BMP
- STBI_NO_PSD
- STBI_NO_TGA
- STBI_NO_GIF
- STBI_NO_HDR
- STBI_NO_PIC
- STBI_NO_PNM (.ppm and .pgm)
-
- - You can request *only* certain decoders and suppress all other ones
- (this will be more forward-compatible, as addition of new decoders
- doesn't require you to disable them explicitly):
-
- STBI_ONLY_JPEG
- STBI_ONLY_PNG
- STBI_ONLY_BMP
- STBI_ONLY_PSD
- STBI_ONLY_TGA
- STBI_ONLY_GIF
- STBI_ONLY_HDR
- STBI_ONLY_PIC
- STBI_ONLY_PNM (.ppm and .pgm)
-
- Note that you can define multiples of these, and you will get all
- of them ("only x" and "only y" is interpreted to mean "only x&y").
-
- - If you use STBI_NO_PNG (or _ONLY_ without PNG), and you still
- want the zlib decoder to be available, #define STBI_SUPPORT_ZLIB
-
- - Compilation of all SIMD code can be suppressed with
- #define STBI_NO_SIMD
- It should not be necessary to disable SIMD unless you have issues
- compiling (e.g. using an x86 compiler which doesn't support SSE
- intrinsics or that doesn't support the method used to detect
- SSE2 support at run-time), and even those can be reported as
- bugs so I can refine the built-in compile-time checking to be
- smarter.
-
- - The old STBI_SIMD system which allowed installing a user-defined
- IDCT etc. has been removed. If you need this, don't upgrade. My
- assumption is that almost nobody was doing this, and those who
- were will find the built-in SIMD more satisfactory anyway.
-
- - RGB values computed for JPEG images are slightly different from
- previous versions of stb_image. (This is due to using less
- integer precision in SIMD.) The C code has been adjusted so
- that the same RGB values will be computed regardless of whether
- SIMD support is available, so your app should always produce
- consistent results. But these results are slightly different from
- previous versions. (Specifically, about 3% of available YCbCr values
- will compute different RGB results from pre-1.49 versions by +-1;
- most of the deviating values are one smaller in the G channel.)
-
- - If you must produce consistent results with previous versions of
- stb_image, #define STBI_JPEG_OLD and you will get the same results
- you used to; however, you will not get the SIMD speedups for
- the YCbCr-to-RGB conversion step (although you should still see
- significant JPEG speedup from the other changes).
-
- Please note that STBI_JPEG_OLD is a temporary feature; it will be
- removed in future versions of the library. It is only intended for
- near-term back-compatibility use.
-
-
- Latest revision history:
- 2.06 (2015-04-19) fix bug where PSD returns wrong '*comp' value
- 2.05 (2015-04-19) fix bug in progressive JPEG handling, fix warning
- 2.04 (2015-04-15) try to re-enable SIMD on MinGW 64-bit
- 2.03 (2015-04-12) additional corruption checking
- stbi_set_flip_vertically_on_load
- fix NEON support; fix mingw support
- 2.02 (2015-01-19) fix incorrect assert, fix warning
- 2.01 (2015-01-17) fix various warnings
- 2.00b (2014-12-25) fix STBI_MALLOC in progressive JPEG
- 2.00 (2014-12-25) optimize JPEG, including x86 SSE2 & ARM NEON SIMD
- progressive JPEG
- PGM/PPM support
- STBI_MALLOC,STBI_REALLOC,STBI_FREE
- STBI_NO_*, STBI_ONLY_*
- GIF bugfix
- 1.48 (2014-12-14) fix incorrectly-named assert()
- 1.47 (2014-12-14) 1/2/4-bit PNG support (both grayscale and paletted)
- optimize PNG
- fix bug in interlaced PNG with user-specified channel count
+LICENSE
+
+ See end of file for license information.
+
+RECENT REVISION HISTORY:
+
+ 2.30 (2024-05-31) avoid erroneous gcc warning
+ 2.29 (2023-05-xx) optimizations
+ 2.28 (2023-01-29) many error fixes, security errors, just tons of stuff
+ 2.27 (2021-07-11) document stbi_info better, 16-bit PNM support, bug fixes
+ 2.26 (2020-07-13) many minor fixes
+ 2.25 (2020-02-02) fix warnings
+ 2.24 (2020-02-02) fix warnings; thread-local failure_reason and flip_vertically
+ 2.23 (2019-08-11) fix clang static analysis warning
+ 2.22 (2019-03-04) gif fixes, fix warnings
+ 2.21 (2019-02-25) fix typo in comment
+ 2.20 (2019-02-07) support utf8 filenames in Windows; fix warnings and platform ifdefs
+ 2.19 (2018-02-11) fix warning
+ 2.18 (2018-01-30) fix warnings
+ 2.17 (2018-01-29) bugfix, 1-bit BMP, 16-bitness query, fix warnings
+ 2.16 (2017-07-23) all functions have 16-bit variants; optimizations; bugfixes
+ 2.15 (2017-03-18) fix png-1,2,4; all Imagenet JPGs; no runtime SSE detection on GCC
+ 2.14 (2017-03-03) remove deprecated STBI_JPEG_OLD; fixes for Imagenet JPGs
+ 2.13 (2016-12-04) experimental 16-bit API, only for PNG so far; fixes
+ 2.12 (2016-04-02) fix typo in 2.11 PSD fix that caused crashes
+ 2.11 (2016-04-02) 16-bit PNGS; enable SSE2 in non-gcc x64
+ RGB-format JPEG; remove white matting in PSD;
+ allocate large structures on the stack;
+ correct channel count for PNG & BMP
+ 2.10 (2016-01-22) avoid warning introduced in 2.09
+ 2.09 (2016-01-16) 16-bit TGA; comments in PNM files; STBI_REALLOC_SIZED
See end of file for full revision history.
============================ Contributors =========================
- Image formats Bug fixes & warning fixes
- Sean Barrett (jpeg, png, bmp) Marc LeBlanc
- Nicolas Schulz (hdr, psd) Christpher Lloyd
- Jonathan Dummer (tga) Dave Moore
- Jean-Marc Lienher (gif) Won Chun
- Tom Seddon (pic) the Horde3D community
- Thatcher Ulrich (psd) Janez Zemva
- Ken Miller (pgm, ppm) Jonathan Blow
- Laurent Gomila
- Aruelien Pocheville
- Extensions, features Ryamond Barbiero
- Jetro Lauha (stbi_info) David Woo
- Martin "SpartanJ" Golini (stbi_info) Martin Golini
- James "moose2000" Brown (iPhone PNG) Roy Eltham
- Ben "Disch" Wenger (io callbacks) Luke Graham
- Omar Cornut (1/2/4-bit PNG) Thomas Ruf
- Nicolas Guillemot (vertical flip) John Bartholomew
- Ken Hamada
- Optimizations & bugfixes Cort Stratton
- Fabian "ryg" Giesen Blazej Dariusz Roszkowski
- Arseny Kapoulkine Thibault Reuille
- Paul Du Bois
- Guillaume George
- If your name should be here but Jerry Jansson
- isn't, let Sean know. Hayaki Saito
- Johan Duparc
- Ronny Chevalier
- Michal Cichon
- Tero Hanninen
- Sergio Gonzalez
- Cass Everitt
- Engin Manap
- Martins Mozeiko
- Joseph Thomson
- Phil Jordan
-
-LICENSE
-
-This software is in the public domain. Where that dedication is not
-recognized, you are granted a perpetual, irrevocable license to copy,
-distribute, and modify this file as you see fit.
-
+ Image formats Extensions, features
+ Sean Barrett (jpeg, png, bmp) Jetro Lauha (stbi_info)
+ Nicolas Schulz (hdr, psd) Martin "SpartanJ" Golini (stbi_info)
+ Jonathan Dummer (tga) James "moose2000" Brown (iPhone PNG)
+ Jean-Marc Lienher (gif) Ben "Disch" Wenger (io callbacks)
+ Tom Seddon (pic) Omar Cornut (1/2/4-bit PNG)
+ Thatcher Ulrich (psd) Nicolas Guillemot (vertical flip)
+ Ken Miller (pgm, ppm) Richard Mitton (16-bit PSD)
+ github:urraka (animated gif) Junggon Kim (PNM comments)
+ Christopher Forseth (animated gif) Daniel Gibson (16-bit TGA)
+ socks-the-fox (16-bit PNG)
+ Jeremy Sawicki (handle all ImageNet JPGs)
+ Optimizations & bugfixes Mikhail Morozov (1-bit BMP)
+ Fabian "ryg" Giesen Anael Seghezzi (is-16-bit query)
+ Arseny Kapoulkine Simon Breuss (16-bit PNM)
+ John-Mark Allen
+ Carmelo J Fdez-Aguera
+
+ Bug & warning fixes
+ Marc LeBlanc David Woo Guillaume George Martins Mozeiko
+ Christpher Lloyd Jerry Jansson Joseph Thomson Blazej Dariusz Roszkowski
+ Phil Jordan Dave Moore Roy Eltham
+ Hayaki Saito Nathan Reed Won Chun
+ Luke Graham Johan Duparc Nick Verigakis the Horde3D community
+ Thomas Ruf Ronny Chevalier github:rlyeh
+ Janez Zemva John Bartholomew Michal Cichon github:romigrou
+ Jonathan Blow Ken Hamada Tero Hanninen github:svdijk
+ Eugene Golushkov Laurent Gomila Cort Stratton github:snagar
+ Aruelien Pocheville Sergio Gonzalez Thibault Reuille github:Zelex
+ Cass Everitt Ryamond Barbiero github:grim210
+ Paul Du Bois Engin Manap Aldo Culquicondor github:sammyhw
+ Philipp Wiesemann Dale Weiler Oriol Ferrer Mesia github:phprus
+ Josh Tobin Neil Bickford Matthew Gregan github:poppolopoppo
+ Julian Raschke Gregory Mullen Christian Floisand github:darealshinji
+ Baldur Karlsson Kevin Schmidt JR Smith github:Michaelangel007
+ Brad Weinberger Matvey Cherevko github:mosra
+ Luca Sas Alexander Veselov Zack Middleton [reserved]
+ Ryan C. Gordon [reserved] [reserved]
+ DO NOT ADD YOUR NAME HERE
+
+ Jacko Dirks
+
+ To add your name to the credits, pick a random blank space in the middle and fill it.
+ 80% of merge conflicts on stb PRs are due to people adding their name at the end
+ of the credits.
*/
#ifndef STBI_INCLUDE_STB_IMAGE_H
@@ -218,10 +132,8 @@ distribute, and modify this file as you see fit.
// DOCUMENTATION
//
// Limitations:
-// - no 16-bit-per-channel PNG
// - no 12-bit-per-channel JPEG
// - no JPEGs with arithmetic coding
-// - no 1-bit BMP
// - GIF always returns *comp=4
//
// Basic usage (see HDR discussion below for HDR usage):
@@ -231,13 +143,13 @@ distribute, and modify this file as you see fit.
// // ... x = width, y = height, n = # 8-bit components per pixel ...
// // ... replace '0' with '1'..'4' to force that many components per pixel
// // ... but 'n' will always be the number that it would have been if you said 0
-// stbi_image_free(data)
+// stbi_image_free(data);
//
// Standard parameters:
-// int *x -- outputs image width in pixels
-// int *y -- outputs image height in pixels
-// int *comp -- outputs # of image components in image file
-// int req_comp -- if non-zero, # of image components requested in result
+// int *x -- outputs image width in pixels
+// int *y -- outputs image height in pixels
+// int *channels_in_file -- outputs # of image components in image file
+// int desired_channels -- if non-zero, # of image components requested in result
//
// The return value from an image loader is an 'unsigned char *' which points
// to the pixel data, or NULL on an allocation failure or if the image is
@@ -245,11 +157,12 @@ distribute, and modify this file as you see fit.
// with each pixel consisting of N interleaved 8-bit components; the first
// pixel pointed to is top-left-most in the image. There is no padding between
// image scanlines or between pixels, regardless of format. The number of
-// components N is 'req_comp' if req_comp is non-zero, or *comp otherwise.
-// If req_comp is non-zero, *comp has the number of components that _would_
-// have been output otherwise. E.g. if you set req_comp to 4, you will always
-// get RGBA output, but you can check *comp to see if it's trivially opaque
-// because e.g. there were only 3 channels in the source image.
+// components N is 'desired_channels' if desired_channels is non-zero, or
+// *channels_in_file otherwise. If desired_channels is non-zero,
+// *channels_in_file has the number of components that _would_ have been
+// output otherwise. E.g. if you set desired_channels to 4, you will always
+// get RGBA output, but you can check *channels_in_file to see if it's trivially
+// opaque because e.g. there were only 3 channels in the source image.
//
// An output image with N components has the following components interleaved
// in this order in each pixel:
@@ -261,14 +174,50 @@ distribute, and modify this file as you see fit.
// 4 red, green, blue, alpha
//
// If image loading fails for any reason, the return value will be NULL,
-// and *x, *y, *comp will be unchanged. The function stbi_failure_reason()
-// can be queried for an extremely brief, end-user unfriendly explanation
-// of why the load failed. Define STBI_NO_FAILURE_STRINGS to avoid
-// compiling these strings at all, and STBI_FAILURE_USERMSG to get slightly
+// and *x, *y, *channels_in_file will be unchanged. The function
+// stbi_failure_reason() can be queried for an extremely brief, end-user
+// unfriendly explanation of why the load failed. Define STBI_NO_FAILURE_STRINGS
+// to avoid compiling these strings at all, and STBI_FAILURE_USERMSG to get slightly
// more user-friendly ones.
//
// Paletted PNG, BMP, GIF, and PIC images are automatically depalettized.
//
+// To query the width, height and component count of an image without having to
+// decode the full file, you can use the stbi_info family of functions:
+//
+// int x,y,n,ok;
+// ok = stbi_info(filename, &x, &y, &n);
+// // returns ok=1 and sets x, y, n if image is a supported format,
+// // 0 otherwise.
+//
+// Note that stb_image pervasively uses ints in its public API for sizes,
+// including sizes of memory buffers. This is now part of the API and thus
+// hard to change without causing breakage. As a result, the various image
+// loaders all have certain limits on image size; these differ somewhat
+// by format but generally boil down to either just under 2GB or just under
+// 1GB. When the decoded image would be larger than this, stb_image decoding
+// will fail.
+//
+// Additionally, stb_image will reject image files that have any of their
+// dimensions set to a larger value than the configurable STBI_MAX_DIMENSIONS,
+// which defaults to 2**24 = 16777216 pixels. Due to the above memory limit,
+// the only way to have an image with such dimensions load correctly
+// is for it to have a rather extreme aspect ratio. Either way, the
+// assumption here is that such larger images are likely to be malformed
+// or malicious. If you do need to load an image with individual dimensions
+// larger than that, and it still fits in the overall size limit, you can
+// #define STBI_MAX_DIMENSIONS on your own to be something larger.
+//
+// ===========================================================================
+//
+// UNICODE:
+//
+// If compiling for Windows and you wish to use Unicode filenames, compile
+// with
+// #define STBI_WINDOWS_UTF8
+// and pass utf8-encoded filenames. Call stbi_convert_wchar_to_utf8 to convert
+// Windows wchar_t filenames to utf8.
+//
// ===========================================================================
//
// Philosophy
@@ -281,15 +230,15 @@ distribute, and modify this file as you see fit.
//
// Sometimes I let "good performance" creep up in priority over "easy to maintain",
// and for best performance I may provide less-easy-to-use APIs that give higher
-// performance, in addition to the easy to use ones. Nevertheless, it's important
+// performance, in addition to the easy-to-use ones. Nevertheless, it's important
// to keep in mind that from the standpoint of you, a client of this library,
-// all you care about is #1 and #3, and stb libraries do not emphasize #3 above all.
+// all you care about is #1 and #3, and stb libraries DO NOT emphasize #3 above all.
//
// Some secondary priorities arise directly from the first two, some of which
-// make more explicit reasons why performance can't be emphasized.
+// provide more explicit reasons why performance can't be emphasized.
//
// - Portable ("ease of use")
-// - Small footprint ("easy to maintain")
+// - Small source code footprint ("easy to maintain")
// - No dependencies ("ease of use")
//
// ===========================================================================
@@ -321,13 +270,6 @@ distribute, and modify this file as you see fit.
// (at least this is true for iOS and Android). Therefore, the NEON support is
// toggled by a build flag: define STBI_NEON to get NEON loops.
//
-// The output of the JPEG decoder is slightly different from versions where
-// SIMD support was introduced (that is, for versions before 1.49). The
-// difference is only +-1 in the 8-bit RGB channels, and only on a small
-// fraction of pixels. You can force the pre-1.49 behavior by defining
-// STBI_JPEG_OLD, but this will disable some of the SIMD decoding path
-// and hence cost some performance.
-//
// If for some reason you do not want to use any of SIMD code, or if
// you have issues compiling it, you can disable it entirely by
// defining STBI_NO_SIMD.
@@ -336,11 +278,10 @@ distribute, and modify this file as you see fit.
//
// HDR image support (disable by defining STBI_NO_HDR)
//
-// stb_image now supports loading HDR images in general, and currently
-// the Radiance .HDR file format, although the support is provided
-// generically. You can still load any file through the existing interface;
-// if you attempt to load an HDR file, it will be automatically remapped to
-// LDR, assuming gamma 2.2 and an arbitrary scale factor defaulting to 1;
+// stb_image supports loading HDR images in general, and currently the Radiance
+// .HDR file format specifically. You can still load any file through the existing
+// interface; if you attempt to load an HDR file, it will be automatically remapped
+// to LDR, assuming gamma 2.2 and an arbitrary scale factor defaulting to 1;
// both of these constants can be reconfigured through this interface:
//
// stbi_hdr_to_ldr_gamma(2.2f);
@@ -372,18 +313,59 @@ distribute, and modify this file as you see fit.
//
// iPhone PNG support:
//
-// By default we convert iphone-formatted PNGs back to RGB, even though
-// they are internally encoded differently. You can disable this conversion
-// by by calling stbi_convert_iphone_png_to_rgb(0), in which case
-// you will always just get the native iphone "format" through (which
-// is BGR stored in RGB).
+// We optionally support converting iPhone-formatted PNGs (which store
+// premultiplied BGRA) back to RGB, even though they're internally encoded
+// differently. To enable this conversion, call
+// stbi_convert_iphone_png_to_rgb(1).
//
// Call stbi_set_unpremultiply_on_load(1) as well to force a divide per
// pixel to remove any premultiplied alpha *only* if the image file explicitly
// says there's premultiplied data (currently only happens in iPhone images,
// and only if iPhone convert-to-rgb processing is on).
//
-
+// ===========================================================================
+//
+// ADDITIONAL CONFIGURATION
+//
+// - You can suppress implementation of any of the decoders to reduce
+// your code footprint by #defining one or more of the following
+// symbols before creating the implementation.
+//
+// STBI_NO_JPEG
+// STBI_NO_PNG
+// STBI_NO_BMP
+// STBI_NO_PSD
+// STBI_NO_TGA
+// STBI_NO_GIF
+// STBI_NO_HDR
+// STBI_NO_PIC
+// STBI_NO_PNM (.ppm and .pgm)
+//
+// - You can request *only* certain decoders and suppress all other ones
+// (this will be more forward-compatible, as addition of new decoders
+// doesn't require you to disable them explicitly):
+//
+// STBI_ONLY_JPEG
+// STBI_ONLY_PNG
+// STBI_ONLY_BMP
+// STBI_ONLY_PSD
+// STBI_ONLY_TGA
+// STBI_ONLY_GIF
+// STBI_ONLY_HDR
+// STBI_ONLY_PIC
+// STBI_ONLY_PNM (.ppm and .pgm)
+//
+// - If you use STBI_NO_PNG (or _ONLY_ without PNG), and you still
+// want the zlib decoder to be available, #define STBI_SUPPORT_ZLIB
+//
+// - If you define STBI_MAX_DIMENSIONS, stb_image will reject images greater
+// than that size (in either width or height) without further processing.
+// This is to let programs in the wild set an upper bound to prevent
+// denial-of-service attacks on untrusted data, as one could generate a
+// valid image of gigantic dimensions and force stb_image to allocate a
+// huge block of memory and spend disproportionate time decoding it. By
+// default this is set to (1 << 24), which is 16777216, but that's still
+// very big.
#ifndef STBI_NO_STDIO
#include
@@ -393,7 +375,7 @@ distribute, and modify this file as you see fit.
enum
{
- STBI_default = 0, // only used for req_comp
+ STBI_default = 0, // only used for desired_channels
STBI_grey = 1,
STBI_grey_alpha = 2,
@@ -401,17 +383,21 @@ enum
STBI_rgb_alpha = 4
};
+#include
typedef unsigned char stbi_uc;
+typedef unsigned short stbi_us;
#ifdef __cplusplus
extern "C" {
#endif
+#ifndef STBIDEF
#ifdef STB_IMAGE_STATIC
#define STBIDEF static
#else
#define STBIDEF extern
#endif
+#endif
//////////////////////////////////////////////////////////////////////////////
//
@@ -429,34 +415,64 @@ typedef struct
int (*eof) (void *user); // returns nonzero if we are at end of file/data
} stbi_io_callbacks;
-STBIDEF stbi_uc *stbi_load (char const *filename, int *x, int *y, int *comp, int req_comp);
-STBIDEF stbi_uc *stbi_load_from_memory (stbi_uc const *buffer, int len , int *x, int *y, int *comp, int req_comp);
-STBIDEF stbi_uc *stbi_load_from_callbacks(stbi_io_callbacks const *clbk , void *user, int *x, int *y, int *comp, int req_comp);
+////////////////////////////////////
+//
+// 8-bits-per-channel interface
+//
+
+STBIDEF stbi_uc *stbi_load_from_memory (stbi_uc const *buffer, int len , int *x, int *y, int *channels_in_file, int desired_channels);
+STBIDEF stbi_uc *stbi_load_from_callbacks(stbi_io_callbacks const *clbk , void *user, int *x, int *y, int *channels_in_file, int desired_channels);
#ifndef STBI_NO_STDIO
-STBIDEF stbi_uc *stbi_load_from_file (FILE *f, int *x, int *y, int *comp, int req_comp);
+STBIDEF stbi_uc *stbi_load (char const *filename, int *x, int *y, int *channels_in_file, int desired_channels);
+STBIDEF stbi_uc *stbi_load_from_file (FILE *f, int *x, int *y, int *channels_in_file, int desired_channels);
// for stbi_load_from_file, file pointer is left pointing immediately after image
#endif
+#ifndef STBI_NO_GIF
+STBIDEF stbi_uc *stbi_load_gif_from_memory(stbi_uc const *buffer, int len, int **delays, int *x, int *y, int *z, int *comp, int req_comp);
+#endif
+
+#ifdef STBI_WINDOWS_UTF8
+STBIDEF int stbi_convert_wchar_to_utf8(char *buffer, size_t bufferlen, const wchar_t* input);
+#endif
+
+////////////////////////////////////
+//
+// 16-bits-per-channel interface
+//
+
+STBIDEF stbi_us *stbi_load_16_from_memory (stbi_uc const *buffer, int len, int *x, int *y, int *channels_in_file, int desired_channels);
+STBIDEF stbi_us *stbi_load_16_from_callbacks(stbi_io_callbacks const *clbk, void *user, int *x, int *y, int *channels_in_file, int desired_channels);
+
+#ifndef STBI_NO_STDIO
+STBIDEF stbi_us *stbi_load_16 (char const *filename, int *x, int *y, int *channels_in_file, int desired_channels);
+STBIDEF stbi_us *stbi_load_from_file_16(FILE *f, int *x, int *y, int *channels_in_file, int desired_channels);
+#endif
+
+////////////////////////////////////
+//
+// float-per-channel interface
+//
#ifndef STBI_NO_LINEAR
- STBIDEF float *stbi_loadf (char const *filename, int *x, int *y, int *comp, int req_comp);
- STBIDEF float *stbi_loadf_from_memory (stbi_uc const *buffer, int len, int *x, int *y, int *comp, int req_comp);
- STBIDEF float *stbi_loadf_from_callbacks (stbi_io_callbacks const *clbk, void *user, int *x, int *y, int *comp, int req_comp);
+ STBIDEF float *stbi_loadf_from_memory (stbi_uc const *buffer, int len, int *x, int *y, int *channels_in_file, int desired_channels);
+ STBIDEF float *stbi_loadf_from_callbacks (stbi_io_callbacks const *clbk, void *user, int *x, int *y, int *channels_in_file, int desired_channels);
#ifndef STBI_NO_STDIO
- STBIDEF float *stbi_loadf_from_file (FILE *f, int *x, int *y, int *comp, int req_comp);
+ STBIDEF float *stbi_loadf (char const *filename, int *x, int *y, int *channels_in_file, int desired_channels);
+ STBIDEF float *stbi_loadf_from_file (FILE *f, int *x, int *y, int *channels_in_file, int desired_channels);
#endif
#endif
#ifndef STBI_NO_HDR
STBIDEF void stbi_hdr_to_ldr_gamma(float gamma);
STBIDEF void stbi_hdr_to_ldr_scale(float scale);
-#endif
+#endif // STBI_NO_HDR
#ifndef STBI_NO_LINEAR
STBIDEF void stbi_ldr_to_hdr_gamma(float gamma);
STBIDEF void stbi_ldr_to_hdr_scale(float scale);
-#endif // STBI_NO_HDR
+#endif // STBI_NO_LINEAR
// stbi_is_hdr is always defined, but always returns false if STBI_NO_HDR
STBIDEF int stbi_is_hdr_from_callbacks(stbi_io_callbacks const *clbk, void *user);
@@ -468,7 +484,7 @@ STBIDEF int stbi_is_hdr_from_file(FILE *f);
// get a VERY brief reason for failure
-// NOT THREADSAFE
+// on most compilers (and ALL modern mainstream compilers) this is threadsafe
STBIDEF const char *stbi_failure_reason (void);
// free the loaded image -- this is just free()
@@ -477,11 +493,14 @@ STBIDEF void stbi_image_free (void *retval_from_stbi_load);
// get image dimensions & components without fully decoding
STBIDEF int stbi_info_from_memory(stbi_uc const *buffer, int len, int *x, int *y, int *comp);
STBIDEF int stbi_info_from_callbacks(stbi_io_callbacks const *clbk, void *user, int *x, int *y, int *comp);
+STBIDEF int stbi_is_16_bit_from_memory(stbi_uc const *buffer, int len);
+STBIDEF int stbi_is_16_bit_from_callbacks(stbi_io_callbacks const *clbk, void *user);
#ifndef STBI_NO_STDIO
-STBIDEF int stbi_info (char const *filename, int *x, int *y, int *comp);
-STBIDEF int stbi_info_from_file (FILE *f, int *x, int *y, int *comp);
-
+STBIDEF int stbi_info (char const *filename, int *x, int *y, int *comp);
+STBIDEF int stbi_info_from_file (FILE *f, int *x, int *y, int *comp);
+STBIDEF int stbi_is_16_bit (char const *filename);
+STBIDEF int stbi_is_16_bit_from_file(FILE *f);
#endif
@@ -498,6 +517,13 @@ STBIDEF void stbi_convert_iphone_png_to_rgb(int flag_true_if_should_convert);
// flip the image vertically, so the first pixel in the output array is the bottom left
STBIDEF void stbi_set_flip_vertically_on_load(int flag_true_if_should_flip);
+// as above, but only applies to images loaded on the thread that calls the function
+// this function is only available if your compiler supports thread-local variables;
+// calling it will fail to link if your compiler doesn't
+STBIDEF void stbi_set_unpremultiply_on_load_thread(int flag_true_if_should_unpremultiply);
+STBIDEF void stbi_convert_iphone_png_to_rgb_thread(int flag_true_if_should_convert);
+STBIDEF void stbi_set_flip_vertically_on_load_thread(int flag_true_if_should_flip);
+
// ZLIB client - used by PNG, available for other purposes
STBIDEF char *stbi_zlib_decode_malloc_guesssize(const char *buffer, int len, int initial_size, int *outlen);
@@ -562,9 +588,10 @@ STBIDEF int stbi_zlib_decode_noheader_buffer(char *obuffer, int olen, const ch
#include // ptrdiff_t on osx
#include
#include
+#include
#if !defined(STBI_NO_LINEAR) || !defined(STBI_NO_HDR)
-#include // ldexp
+#include // ldexp, pow
#endif
#ifndef STBI_NO_STDIO
@@ -576,6 +603,12 @@ STBIDEF int stbi_zlib_decode_noheader_buffer(char *obuffer, int olen, const ch
#define STBI_ASSERT(x) assert(x)
#endif
+#ifdef __cplusplus
+#define STBI_EXTERN extern "C"
+#else
+#define STBI_EXTERN extern
+#endif
+
#ifndef _MSC_VER
#ifdef __cplusplus
@@ -587,8 +620,25 @@ STBIDEF int stbi_zlib_decode_noheader_buffer(char *obuffer, int olen, const ch
#define stbi_inline __forceinline
#endif
+#ifndef STBI_NO_THREAD_LOCALS
+ #if defined(__cplusplus) && __cplusplus >= 201103L
+ #define STBI_THREAD_LOCAL thread_local
+ #elif defined(__GNUC__) && __GNUC__ < 5
+ #define STBI_THREAD_LOCAL __thread
+ #elif defined(_MSC_VER)
+ #define STBI_THREAD_LOCAL __declspec(thread)
+ #elif defined (__STDC_VERSION__) && __STDC_VERSION__ >= 201112L && !defined(__STDC_NO_THREADS__)
+ #define STBI_THREAD_LOCAL _Thread_local
+ #endif
-#ifdef _MSC_VER
+ #ifndef STBI_THREAD_LOCAL
+ #if defined(__GNUC__)
+ #define STBI_THREAD_LOCAL __thread
+ #endif
+ #endif
+#endif
+
+#if defined(_MSC_VER) || defined(__SYMBIAN32__)
typedef unsigned short stbi__uint16;
typedef signed short stbi__int16;
typedef unsigned int stbi__uint32;
@@ -617,21 +667,25 @@ typedef unsigned char validate_uint32[sizeof(stbi__uint32)==4 ? 1 : -1];
#ifdef STBI_HAS_LROTL
#define stbi_lrot(x,y) _lrotl(x,y)
#else
- #define stbi_lrot(x,y) (((x) << (y)) | ((x) >> (32 - (y))))
+ #define stbi_lrot(x,y) (((x) << (y)) | ((x) >> (-(y) & 31)))
#endif
-#if defined(STBI_MALLOC) && defined(STBI_FREE) && defined(STBI_REALLOC)
+#if defined(STBI_MALLOC) && defined(STBI_FREE) && (defined(STBI_REALLOC) || defined(STBI_REALLOC_SIZED))
// ok
-#elif !defined(STBI_MALLOC) && !defined(STBI_FREE) && !defined(STBI_REALLOC)
+#elif !defined(STBI_MALLOC) && !defined(STBI_FREE) && !defined(STBI_REALLOC) && !defined(STBI_REALLOC_SIZED)
// ok
#else
-#error "Must define all or none of STBI_MALLOC, STBI_FREE, and STBI_REALLOC."
+#error "Must define all or none of STBI_MALLOC, STBI_FREE, and STBI_REALLOC (or STBI_REALLOC_SIZED)."
#endif
#ifndef STBI_MALLOC
-#define STBI_MALLOC(sz) malloc(sz)
-#define STBI_REALLOC(p,sz) realloc(p,sz)
-#define STBI_FREE(p) free(p)
+#define STBI_MALLOC(sz) malloc(sz)
+#define STBI_REALLOC(p,newsz) realloc(p,newsz)
+#define STBI_FREE(p) free(p)
+#endif
+
+#ifndef STBI_REALLOC_SIZED
+#define STBI_REALLOC_SIZED(p,oldsz,newsz) STBI_REALLOC(p,newsz)
#endif
// x86/x64 detection
@@ -641,12 +695,14 @@ typedef unsigned char validate_uint32[sizeof(stbi__uint32)==4 ? 1 : -1];
#define STBI__X86_TARGET
#endif
-#if defined(__GNUC__) && (defined(STBI__X86_TARGET) || defined(STBI__X64_TARGET)) && !defined(__SSE2__) && !defined(STBI_NO_SIMD)
-// NOTE: not clear do we actually need this for the 64-bit path?
+#if defined(__GNUC__) && defined(STBI__X86_TARGET) && !defined(__SSE2__) && !defined(STBI_NO_SIMD)
// gcc doesn't support sse2 intrinsics unless you compile with -msse2,
-// (but compiling with -msse2 allows the compiler to use SSE2 everywhere;
-// this is just broken and gcc are jerks for not fixing it properly
-// http://www.virtualdub.org/blog/pivot/entry.php?id=363 )
+// which in turn means it gets to use SSE2 everywhere. This is unfortunate,
+// but previous attempts to provide the SSE2 functions with runtime
+// detection caused numerous issues. The way architecture extensions are
+// exposed in GCC/Clang is, sadly, not really suited for one-file libs.
+// New behavior: if compiled with -msse2, we use SSE2 without any
+// detection; if not, we don't use it at all.
#define STBI_NO_SIMD
#endif
@@ -665,7 +721,7 @@ typedef unsigned char validate_uint32[sizeof(stbi__uint32)==4 ? 1 : -1];
#define STBI_NO_SIMD
#endif
-#if !defined(STBI_NO_SIMD) && defined(STBI__X86_TARGET)
+#if !defined(STBI_NO_SIMD) && (defined(STBI__X86_TARGET) || defined(STBI__X64_TARGET))
#define STBI_SSE2
#include
@@ -694,25 +750,27 @@ static int stbi__cpuid3(void)
#define STBI_SIMD_ALIGN(type, name) __declspec(align(16)) type name
-static int stbi__sse2_available()
+#if !defined(STBI_NO_JPEG) && defined(STBI_SSE2)
+static int stbi__sse2_available(void)
{
int info3 = stbi__cpuid3();
return ((info3 >> 26) & 1) != 0;
}
+#endif
+
#else // assume GCC-style if not VC++
#define STBI_SIMD_ALIGN(type, name) type name __attribute__((aligned(16)))
-static int stbi__sse2_available()
+#if !defined(STBI_NO_JPEG) && defined(STBI_SSE2)
+static int stbi__sse2_available(void)
{
-#if defined(__GNUC__) && (__GNUC__ * 100 + __GNUC_MINOR__) >= 408 // GCC 4.8 or later
- // GCC 4.8+ has a nice way to do this
- return __builtin_cpu_supports("sse2");
-#else
- // portable way to do this, preferably without using GCC inline ASM?
- // just bail for now.
- return 0;
-#endif
+ // If we're even attempting to compile this on GCC/Clang, that means
+ // -msse2 is on, which means the compiler is allowed to use SSE2
+ // instructions at will, and so are we.
+ return 1;
}
+#endif
+
#endif
#endif
@@ -723,14 +781,21 @@ static int stbi__sse2_available()
#ifdef STBI_NEON
#include
-// assume GCC or Clang on ARM targets
+#ifdef _MSC_VER
+#define STBI_SIMD_ALIGN(type, name) __declspec(align(16)) type name
+#else
#define STBI_SIMD_ALIGN(type, name) type name __attribute__((aligned(16)))
#endif
+#endif
#ifndef STBI_SIMD_ALIGN
#define STBI_SIMD_ALIGN(type, name) type name
#endif
+#ifndef STBI_MAX_DIMENSIONS
+#define STBI_MAX_DIMENSIONS (1 << 24)
+#endif
+
///////////////////////////////////////////////
//
// stbi__context struct and start_xxx functions
@@ -748,9 +813,10 @@ typedef struct
int read_from_callbacks;
int buflen;
stbi_uc buffer_start[128];
+ int callback_already_read;
stbi_uc *img_buffer, *img_buffer_end;
- stbi_uc *img_buffer_original;
+ stbi_uc *img_buffer_original, *img_buffer_original_end;
} stbi__context;
@@ -761,8 +827,9 @@ static void stbi__start_mem(stbi__context *s, stbi_uc const *buffer, int len)
{
s->io.read = NULL;
s->read_from_callbacks = 0;
+ s->callback_already_read = 0;
s->img_buffer = s->img_buffer_original = (stbi_uc *) buffer;
- s->img_buffer_end = (stbi_uc *) buffer+len;
+ s->img_buffer_end = s->img_buffer_original_end = (stbi_uc *) buffer+len;
}
// initialize a callback-based context
@@ -772,8 +839,10 @@ static void stbi__start_callbacks(stbi__context *s, stbi_io_callbacks *c, void *
s->io_user_data = user;
s->buflen = sizeof(s->buffer_start);
s->read_from_callbacks = 1;
- s->img_buffer_original = s->buffer_start;
+ s->callback_already_read = 0;
+ s->img_buffer = s->img_buffer_original = s->buffer_start;
stbi__refill_buffer(s);
+ s->img_buffer_original_end = s->img_buffer_end;
}
#ifndef STBI_NO_STDIO
@@ -785,12 +854,17 @@ static int stbi__stdio_read(void *user, char *data, int size)
static void stbi__stdio_skip(void *user, int n)
{
+ int ch;
fseek((FILE*) user, n, SEEK_CUR);
+ ch = fgetc((FILE*) user); /* have to read a byte to reset feof()'s flag */
+ if (ch != EOF) {
+ ungetc(ch, (FILE *) user); /* push byte back onto stream if valid. */
+ }
}
static int stbi__stdio_eof(void *user)
{
- return feof((FILE*) user);
+ return feof((FILE*) user) || ferror((FILE *) user);
}
static stbi_io_callbacks stbi__stdio_callbacks =
@@ -815,81 +889,200 @@ static void stbi__rewind(stbi__context *s)
// but we just rewind to the beginning of the initial buffer, because
// we only use it after doing 'test', which only ever looks at at most 92 bytes
s->img_buffer = s->img_buffer_original;
+ s->img_buffer_end = s->img_buffer_original_end;
}
+enum
+{
+ STBI_ORDER_RGB,
+ STBI_ORDER_BGR
+};
+
+typedef struct
+{
+ int bits_per_channel;
+ int num_channels;
+ int channel_order;
+} stbi__result_info;
+
#ifndef STBI_NO_JPEG
static int stbi__jpeg_test(stbi__context *s);
-static stbi_uc *stbi__jpeg_load(stbi__context *s, int *x, int *y, int *comp, int req_comp);
+static void *stbi__jpeg_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri);
static int stbi__jpeg_info(stbi__context *s, int *x, int *y, int *comp);
#endif
#ifndef STBI_NO_PNG
static int stbi__png_test(stbi__context *s);
-static stbi_uc *stbi__png_load(stbi__context *s, int *x, int *y, int *comp, int req_comp);
+static void *stbi__png_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri);
static int stbi__png_info(stbi__context *s, int *x, int *y, int *comp);
+static int stbi__png_is16(stbi__context *s);
#endif
#ifndef STBI_NO_BMP
static int stbi__bmp_test(stbi__context *s);
-static stbi_uc *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int req_comp);
+static void *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri);
static int stbi__bmp_info(stbi__context *s, int *x, int *y, int *comp);
#endif
#ifndef STBI_NO_TGA
static int stbi__tga_test(stbi__context *s);
-static stbi_uc *stbi__tga_load(stbi__context *s, int *x, int *y, int *comp, int req_comp);
+static void *stbi__tga_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri);
static int stbi__tga_info(stbi__context *s, int *x, int *y, int *comp);
#endif
#ifndef STBI_NO_PSD
static int stbi__psd_test(stbi__context *s);
-static stbi_uc *stbi__psd_load(stbi__context *s, int *x, int *y, int *comp, int req_comp);
+static void *stbi__psd_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri, int bpc);
static int stbi__psd_info(stbi__context *s, int *x, int *y, int *comp);
+static int stbi__psd_is16(stbi__context *s);
#endif
#ifndef STBI_NO_HDR
static int stbi__hdr_test(stbi__context *s);
-static float *stbi__hdr_load(stbi__context *s, int *x, int *y, int *comp, int req_comp);
+static float *stbi__hdr_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri);
static int stbi__hdr_info(stbi__context *s, int *x, int *y, int *comp);
#endif
#ifndef STBI_NO_PIC
static int stbi__pic_test(stbi__context *s);
-static stbi_uc *stbi__pic_load(stbi__context *s, int *x, int *y, int *comp, int req_comp);
+static void *stbi__pic_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri);
static int stbi__pic_info(stbi__context *s, int *x, int *y, int *comp);
#endif
#ifndef STBI_NO_GIF
static int stbi__gif_test(stbi__context *s);
-static stbi_uc *stbi__gif_load(stbi__context *s, int *x, int *y, int *comp, int req_comp);
+static void *stbi__gif_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri);
+static void *stbi__load_gif_main(stbi__context *s, int **delays, int *x, int *y, int *z, int *comp, int req_comp);
static int stbi__gif_info(stbi__context *s, int *x, int *y, int *comp);
#endif
#ifndef STBI_NO_PNM
static int stbi__pnm_test(stbi__context *s);
-static stbi_uc *stbi__pnm_load(stbi__context *s, int *x, int *y, int *comp, int req_comp);
+static void *stbi__pnm_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri);
static int stbi__pnm_info(stbi__context *s, int *x, int *y, int *comp);
+static int stbi__pnm_is16(stbi__context *s);
#endif
-// this is not threadsafe
-static const char *stbi__g_failure_reason;
+static
+#ifdef STBI_THREAD_LOCAL
+STBI_THREAD_LOCAL
+#endif
+const char *stbi__g_failure_reason;
STBIDEF const char *stbi_failure_reason(void)
{
return stbi__g_failure_reason;
}
+#ifndef STBI_NO_FAILURE_STRINGS
static int stbi__err(const char *str)
{
stbi__g_failure_reason = str;
return 0;
}
+#endif
static void *stbi__malloc(size_t size)
{
return STBI_MALLOC(size);
}
+// stb_image uses ints pervasively, including for offset calculations.
+// therefore the largest decoded image size we can support with the
+// current code, even on 64-bit targets, is INT_MAX. this is not a
+// significant limitation for the intended use case.
+//
+// we do, however, need to make sure our size calculations don't
+// overflow. hence a few helper functions for size calculations that
+// multiply integers together, making sure that they're non-negative
+// and no overflow occurs.
+
+// return 1 if the sum is valid, 0 on overflow.
+// negative terms are considered invalid.
+static int stbi__addsizes_valid(int a, int b)
+{
+ if (b < 0) return 0;
+ // now 0 <= b <= INT_MAX, hence also
+ // 0 <= INT_MAX - b <= INTMAX.
+ // And "a + b <= INT_MAX" (which might overflow) is the
+ // same as a <= INT_MAX - b (no overflow)
+ return a <= INT_MAX - b;
+}
+
+// returns 1 if the product is valid, 0 on overflow.
+// negative factors are considered invalid.
+static int stbi__mul2sizes_valid(int a, int b)
+{
+ if (a < 0 || b < 0) return 0;
+ if (b == 0) return 1; // mul-by-0 is always safe
+ // portable way to check for no overflows in a*b
+ return a <= INT_MAX/b;
+}
+
+#if !defined(STBI_NO_JPEG) || !defined(STBI_NO_PNG) || !defined(STBI_NO_TGA) || !defined(STBI_NO_HDR)
+// returns 1 if "a*b + add" has no negative terms/factors and doesn't overflow
+static int stbi__mad2sizes_valid(int a, int b, int add)
+{
+ return stbi__mul2sizes_valid(a, b) && stbi__addsizes_valid(a*b, add);
+}
+#endif
+
+// returns 1 if "a*b*c + add" has no negative terms/factors and doesn't overflow
+static int stbi__mad3sizes_valid(int a, int b, int c, int add)
+{
+ return stbi__mul2sizes_valid(a, b) && stbi__mul2sizes_valid(a*b, c) &&
+ stbi__addsizes_valid(a*b*c, add);
+}
+
+// returns 1 if "a*b*c*d + add" has no negative terms/factors and doesn't overflow
+#if !defined(STBI_NO_LINEAR) || !defined(STBI_NO_HDR) || !defined(STBI_NO_PNM)
+static int stbi__mad4sizes_valid(int a, int b, int c, int d, int add)
+{
+ return stbi__mul2sizes_valid(a, b) && stbi__mul2sizes_valid(a*b, c) &&
+ stbi__mul2sizes_valid(a*b*c, d) && stbi__addsizes_valid(a*b*c*d, add);
+}
+#endif
+
+#if !defined(STBI_NO_JPEG) || !defined(STBI_NO_PNG) || !defined(STBI_NO_TGA) || !defined(STBI_NO_HDR)
+// mallocs with size overflow checking
+static void *stbi__malloc_mad2(int a, int b, int add)
+{
+ if (!stbi__mad2sizes_valid(a, b, add)) return NULL;
+ return stbi__malloc(a*b + add);
+}
+#endif
+
+static void *stbi__malloc_mad3(int a, int b, int c, int add)
+{
+ if (!stbi__mad3sizes_valid(a, b, c, add)) return NULL;
+ return stbi__malloc(a*b*c + add);
+}
+
+#if !defined(STBI_NO_LINEAR) || !defined(STBI_NO_HDR) || !defined(STBI_NO_PNM)
+static void *stbi__malloc_mad4(int a, int b, int c, int d, int add)
+{
+ if (!stbi__mad4sizes_valid(a, b, c, d, add)) return NULL;
+ return stbi__malloc(a*b*c*d + add);
+}
+#endif
+
+// returns 1 if the sum of two signed ints is valid (between -2^31 and 2^31-1 inclusive), 0 on overflow.
+static int stbi__addints_valid(int a, int b)
+{
+ if ((a >= 0) != (b >= 0)) return 1; // a and b have different signs, so no overflow
+ if (a < 0 && b < 0) return a >= INT_MIN - b; // same as a + b >= INT_MIN; INT_MIN - b cannot overflow since b < 0.
+ return a <= INT_MAX - b;
+}
+
+// returns 1 if the product of two ints fits in a signed short, 0 on overflow.
+static int stbi__mul2shorts_valid(int a, int b)
+{
+ if (b == 0 || b == -1) return 1; // multiplication by 0 is always 0; check for -1 so SHRT_MIN/b doesn't overflow
+ if ((a >= 0) == (b >= 0)) return a <= SHRT_MAX/b; // product is positive, so similar to mul2sizes_valid
+ if (b < 0) return a <= SHRT_MIN / b; // same as a * b >= SHRT_MIN
+ return a >= SHRT_MIN / b;
+}
+
// stbi__err - error
// stbi__errpf - error returning pointer to float
// stbi__errpuc - error returning pointer to unsigned char
@@ -902,8 +1095,8 @@ static void *stbi__malloc(size_t size)
#define stbi__err(x,y) stbi__err(x)
#endif
-#define stbi__errpf(x,y) ((float *) (stbi__err(x,y)?NULL:NULL))
-#define stbi__errpuc(x,y) ((unsigned char *) (stbi__err(x,y)?NULL:NULL))
+#define stbi__errpf(x,y) ((float *)(size_t) (stbi__err(x,y)?NULL:NULL))
+#define stbi__errpuc(x,y) ((unsigned char *)(size_t) (stbi__err(x,y)?NULL:NULL))
STBIDEF void stbi_image_free(void *retval_from_stbi_load)
{
@@ -918,40 +1111,69 @@ static float *stbi__ldr_to_hdr(stbi_uc *data, int x, int y, int comp);
static stbi_uc *stbi__hdr_to_ldr(float *data, int x, int y, int comp);
#endif
-static int stbi__vertically_flip_on_load = 0;
+static int stbi__vertically_flip_on_load_global = 0;
STBIDEF void stbi_set_flip_vertically_on_load(int flag_true_if_should_flip)
{
- stbi__vertically_flip_on_load = flag_true_if_should_flip;
+ stbi__vertically_flip_on_load_global = flag_true_if_should_flip;
}
-static unsigned char *stbi__load_main(stbi__context *s, int *x, int *y, int *comp, int req_comp)
+#ifndef STBI_THREAD_LOCAL
+#define stbi__vertically_flip_on_load stbi__vertically_flip_on_load_global
+#else
+static STBI_THREAD_LOCAL int stbi__vertically_flip_on_load_local, stbi__vertically_flip_on_load_set;
+
+STBIDEF void stbi_set_flip_vertically_on_load_thread(int flag_true_if_should_flip)
{
- #ifndef STBI_NO_JPEG
- if (stbi__jpeg_test(s)) return stbi__jpeg_load(s,x,y,comp,req_comp);
- #endif
+ stbi__vertically_flip_on_load_local = flag_true_if_should_flip;
+ stbi__vertically_flip_on_load_set = 1;
+}
+
+#define stbi__vertically_flip_on_load (stbi__vertically_flip_on_load_set \
+ ? stbi__vertically_flip_on_load_local \
+ : stbi__vertically_flip_on_load_global)
+#endif // STBI_THREAD_LOCAL
+
+static void *stbi__load_main(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri, int bpc)
+{
+ memset(ri, 0, sizeof(*ri)); // make sure it's initialized if we add new fields
+ ri->bits_per_channel = 8; // default is 8 so most paths don't have to be changed
+ ri->channel_order = STBI_ORDER_RGB; // all current input & output are this, but this is here so we can add BGR order
+ ri->num_channels = 0;
+
+ // test the formats with a very explicit header first (at least a FOURCC
+ // or distinctive magic number first)
#ifndef STBI_NO_PNG
- if (stbi__png_test(s)) return stbi__png_load(s,x,y,comp,req_comp);
+ if (stbi__png_test(s)) return stbi__png_load(s,x,y,comp,req_comp, ri);
#endif
#ifndef STBI_NO_BMP
- if (stbi__bmp_test(s)) return stbi__bmp_load(s,x,y,comp,req_comp);
+ if (stbi__bmp_test(s)) return stbi__bmp_load(s,x,y,comp,req_comp, ri);
#endif
#ifndef STBI_NO_GIF
- if (stbi__gif_test(s)) return stbi__gif_load(s,x,y,comp,req_comp);
+ if (stbi__gif_test(s)) return stbi__gif_load(s,x,y,comp,req_comp, ri);
#endif
#ifndef STBI_NO_PSD
- if (stbi__psd_test(s)) return stbi__psd_load(s,x,y,comp,req_comp);
+ if (stbi__psd_test(s)) return stbi__psd_load(s,x,y,comp,req_comp, ri, bpc);
+ #else
+ STBI_NOTUSED(bpc);
#endif
#ifndef STBI_NO_PIC
- if (stbi__pic_test(s)) return stbi__pic_load(s,x,y,comp,req_comp);
+ if (stbi__pic_test(s)) return stbi__pic_load(s,x,y,comp,req_comp, ri);
+ #endif
+
+ // then the formats that can end up attempting to load with just 1 or 2
+ // bytes matching expectations; these are prone to false positives, so
+ // try them later
+ #ifndef STBI_NO_JPEG
+ if (stbi__jpeg_test(s)) return stbi__jpeg_load(s,x,y,comp,req_comp, ri);
#endif
#ifndef STBI_NO_PNM
- if (stbi__pnm_test(s)) return stbi__pnm_load(s,x,y,comp,req_comp);
+ if (stbi__pnm_test(s)) return stbi__pnm_load(s,x,y,comp,req_comp, ri);
#endif
#ifndef STBI_NO_HDR
if (stbi__hdr_test(s)) {
- float *hdr = stbi__hdr_load(s, x,y,comp,req_comp);
+ float *hdr = stbi__hdr_load(s, x,y,comp,req_comp, ri);
return stbi__hdr_to_ldr(hdr, *x, *y, req_comp ? req_comp : *comp);
}
#endif
@@ -959,65 +1181,179 @@ static unsigned char *stbi__load_main(stbi__context *s, int *x, int *y, int *com
#ifndef STBI_NO_TGA
// test tga last because it's a crappy test!
if (stbi__tga_test(s))
- return stbi__tga_load(s,x,y,comp,req_comp);
+ return stbi__tga_load(s,x,y,comp,req_comp, ri);
#endif
return stbi__errpuc("unknown image type", "Image not of any known type, or corrupt");
}
-static unsigned char *stbi__load_flip(stbi__context *s, int *x, int *y, int *comp, int req_comp)
+static stbi_uc *stbi__convert_16_to_8(stbi__uint16 *orig, int w, int h, int channels)
{
- unsigned char *result = stbi__load_main(s, x, y, comp, req_comp);
+ int i;
+ int img_len = w * h * channels;
+ stbi_uc *reduced;
- if (stbi__vertically_flip_on_load && result != NULL) {
- int w = *x, h = *y;
- int depth = req_comp ? req_comp : *comp;
- int row,col,z;
- stbi_uc temp;
-
- // @OPTIMIZE: use a bigger temp buffer and memcpy multiple pixels at once
- for (row = 0; row < (h>>1); row++) {
- for (col = 0; col < w; col++) {
- for (z = 0; z < depth; z++) {
- temp = result[(row * w + col) * depth + z];
- result[(row * w + col) * depth + z] = result[((h - row - 1) * w + col) * depth + z];
- result[((h - row - 1) * w + col) * depth + z] = temp;
- }
- }
+ reduced = (stbi_uc *) stbi__malloc(img_len);
+ if (reduced == NULL) return stbi__errpuc("outofmem", "Out of memory");
+
+ for (i = 0; i < img_len; ++i)
+ reduced[i] = (stbi_uc)((orig[i] >> 8) & 0xFF); // top half of each byte is sufficient approx of 16->8 bit scaling
+
+ STBI_FREE(orig);
+ return reduced;
+}
+
+static stbi__uint16 *stbi__convert_8_to_16(stbi_uc *orig, int w, int h, int channels)
+{
+ int i;
+ int img_len = w * h * channels;
+ stbi__uint16 *enlarged;
+
+ enlarged = (stbi__uint16 *) stbi__malloc(img_len*2);
+ if (enlarged == NULL) return (stbi__uint16 *) stbi__errpuc("outofmem", "Out of memory");
+
+ for (i = 0; i < img_len; ++i)
+ enlarged[i] = (stbi__uint16)((orig[i] << 8) + orig[i]); // replicate to high and low byte, maps 0->0, 255->0xffff
+
+ STBI_FREE(orig);
+ return enlarged;
+}
+
+static void stbi__vertical_flip(void *image, int w, int h, int bytes_per_pixel)
+{
+ int row;
+ size_t bytes_per_row = (size_t)w * bytes_per_pixel;
+ stbi_uc temp[2048];
+ stbi_uc *bytes = (stbi_uc *)image;
+
+ for (row = 0; row < (h>>1); row++) {
+ stbi_uc *row0 = bytes + row*bytes_per_row;
+ stbi_uc *row1 = bytes + (h - row - 1)*bytes_per_row;
+ // swap row0 with row1
+ size_t bytes_left = bytes_per_row;
+ while (bytes_left) {
+ size_t bytes_copy = (bytes_left < sizeof(temp)) ? bytes_left : sizeof(temp);
+ memcpy(temp, row0, bytes_copy);
+ memcpy(row0, row1, bytes_copy);
+ memcpy(row1, temp, bytes_copy);
+ row0 += bytes_copy;
+ row1 += bytes_copy;
+ bytes_left -= bytes_copy;
}
}
+}
- return result;
+#ifndef STBI_NO_GIF
+static void stbi__vertical_flip_slices(void *image, int w, int h, int z, int bytes_per_pixel)
+{
+ int slice;
+ int slice_size = w * h * bytes_per_pixel;
+
+ stbi_uc *bytes = (stbi_uc *)image;
+ for (slice = 0; slice < z; ++slice) {
+ stbi__vertical_flip(bytes, w, h, bytes_per_pixel);
+ bytes += slice_size;
+ }
}
+#endif
+
+static unsigned char *stbi__load_and_postprocess_8bit(stbi__context *s, int *x, int *y, int *comp, int req_comp)
+{
+ stbi__result_info ri;
+ void *result = stbi__load_main(s, x, y, comp, req_comp, &ri, 8);
+
+ if (result == NULL)
+ return NULL;
+
+ // it is the responsibility of the loaders to make sure we get either 8 or 16 bit.
+ STBI_ASSERT(ri.bits_per_channel == 8 || ri.bits_per_channel == 16);
+
+ if (ri.bits_per_channel != 8) {
+ result = stbi__convert_16_to_8((stbi__uint16 *) result, *x, *y, req_comp == 0 ? *comp : req_comp);
+ ri.bits_per_channel = 8;
+ }
+
+ // @TODO: move stbi__convert_format to here
+
+ if (stbi__vertically_flip_on_load) {
+ int channels = req_comp ? req_comp : *comp;
+ stbi__vertical_flip(result, *x, *y, channels * sizeof(stbi_uc));
+ }
+
+ return (unsigned char *) result;
+}
+
+static stbi__uint16 *stbi__load_and_postprocess_16bit(stbi__context *s, int *x, int *y, int *comp, int req_comp)
+{
+ stbi__result_info ri;
+ void *result = stbi__load_main(s, x, y, comp, req_comp, &ri, 16);
+
+ if (result == NULL)
+ return NULL;
+
+ // it is the responsibility of the loaders to make sure we get either 8 or 16 bit.
+ STBI_ASSERT(ri.bits_per_channel == 8 || ri.bits_per_channel == 16);
+
+ if (ri.bits_per_channel != 16) {
+ result = stbi__convert_8_to_16((stbi_uc *) result, *x, *y, req_comp == 0 ? *comp : req_comp);
+ ri.bits_per_channel = 16;
+ }
+ // @TODO: move stbi__convert_format16 to here
+ // @TODO: special case RGB-to-Y (and RGBA-to-YA) for 8-bit-to-16-bit case to keep more precision
+
+ if (stbi__vertically_flip_on_load) {
+ int channels = req_comp ? req_comp : *comp;
+ stbi__vertical_flip(result, *x, *y, channels * sizeof(stbi__uint16));
+ }
+
+ return (stbi__uint16 *) result;
+}
+
+#if !defined(STBI_NO_HDR) && !defined(STBI_NO_LINEAR)
static void stbi__float_postprocess(float *result, int *x, int *y, int *comp, int req_comp)
{
if (stbi__vertically_flip_on_load && result != NULL) {
- int w = *x, h = *y;
- int depth = req_comp ? req_comp : *comp;
- int row,col,z;
- float temp;
-
- // @OPTIMIZE: use a bigger temp buffer and memcpy multiple pixels at once
- for (row = 0; row < (h>>1); row++) {
- for (col = 0; col < w; col++) {
- for (z = 0; z < depth; z++) {
- temp = result[(row * w + col) * depth + z];
- result[(row * w + col) * depth + z] = result[((h - row - 1) * w + col) * depth + z];
- result[((h - row - 1) * w + col) * depth + z] = temp;
- }
- }
- }
+ int channels = req_comp ? req_comp : *comp;
+ stbi__vertical_flip(result, *x, *y, channels * sizeof(float));
}
}
-
+#endif
#ifndef STBI_NO_STDIO
+#if defined(_WIN32) && defined(STBI_WINDOWS_UTF8)
+STBI_EXTERN __declspec(dllimport) int __stdcall MultiByteToWideChar(unsigned int cp, unsigned long flags, const char *str, int cbmb, wchar_t *widestr, int cchwide);
+STBI_EXTERN __declspec(dllimport) int __stdcall WideCharToMultiByte(unsigned int cp, unsigned long flags, const wchar_t *widestr, int cchwide, char *str, int cbmb, const char *defchar, int *used_default);
+#endif
+
+#if defined(_WIN32) && defined(STBI_WINDOWS_UTF8)
+STBIDEF int stbi_convert_wchar_to_utf8(char *buffer, size_t bufferlen, const wchar_t* input)
+{
+ return WideCharToMultiByte(65001 /* UTF8 */, 0, input, -1, buffer, (int) bufferlen, NULL, NULL);
+}
+#endif
+
static FILE *stbi__fopen(char const *filename, char const *mode)
{
FILE *f;
+#if defined(_WIN32) && defined(STBI_WINDOWS_UTF8)
+ wchar_t wMode[64];
+ wchar_t wFilename[1024];
+ if (0 == MultiByteToWideChar(65001 /* UTF8 */, 0, filename, -1, wFilename, sizeof(wFilename)/sizeof(*wFilename)))
+ return 0;
+
+ if (0 == MultiByteToWideChar(65001 /* UTF8 */, 0, mode, -1, wMode, sizeof(wMode)/sizeof(*wMode)))
+ return 0;
+
#if defined(_MSC_VER) && _MSC_VER >= 1400
+ if (0 != _wfopen_s(&f, wFilename, wMode))
+ f = 0;
+#else
+ f = _wfopen(wFilename, wMode);
+#endif
+
+#elif defined(_MSC_VER) && _MSC_VER >= 1400
if (0 != fopen_s(&f, filename, mode))
f=0;
#else
@@ -1042,28 +1378,83 @@ STBIDEF stbi_uc *stbi_load_from_file(FILE *f, int *x, int *y, int *comp, int req
unsigned char *result;
stbi__context s;
stbi__start_file(&s,f);
- result = stbi__load_flip(&s,x,y,comp,req_comp);
+ result = stbi__load_and_postprocess_8bit(&s,x,y,comp,req_comp);
+ if (result) {
+ // need to 'unget' all the characters in the IO buffer
+ fseek(f, - (int) (s.img_buffer_end - s.img_buffer), SEEK_CUR);
+ }
+ return result;
+}
+
+STBIDEF stbi__uint16 *stbi_load_from_file_16(FILE *f, int *x, int *y, int *comp, int req_comp)
+{
+ stbi__uint16 *result;
+ stbi__context s;
+ stbi__start_file(&s,f);
+ result = stbi__load_and_postprocess_16bit(&s,x,y,comp,req_comp);
if (result) {
// need to 'unget' all the characters in the IO buffer
fseek(f, - (int) (s.img_buffer_end - s.img_buffer), SEEK_CUR);
}
return result;
}
+
+STBIDEF stbi_us *stbi_load_16(char const *filename, int *x, int *y, int *comp, int req_comp)
+{
+ FILE *f = stbi__fopen(filename, "rb");
+ stbi__uint16 *result;
+ if (!f) return (stbi_us *) stbi__errpuc("can't fopen", "Unable to open file");
+ result = stbi_load_from_file_16(f,x,y,comp,req_comp);
+ fclose(f);
+ return result;
+}
+
+
#endif //!STBI_NO_STDIO
+STBIDEF stbi_us *stbi_load_16_from_memory(stbi_uc const *buffer, int len, int *x, int *y, int *channels_in_file, int desired_channels)
+{
+ stbi__context s;
+ stbi__start_mem(&s,buffer,len);
+ return stbi__load_and_postprocess_16bit(&s,x,y,channels_in_file,desired_channels);
+}
+
+STBIDEF stbi_us *stbi_load_16_from_callbacks(stbi_io_callbacks const *clbk, void *user, int *x, int *y, int *channels_in_file, int desired_channels)
+{
+ stbi__context s;
+ stbi__start_callbacks(&s, (stbi_io_callbacks *)clbk, user);
+ return stbi__load_and_postprocess_16bit(&s,x,y,channels_in_file,desired_channels);
+}
+
STBIDEF stbi_uc *stbi_load_from_memory(stbi_uc const *buffer, int len, int *x, int *y, int *comp, int req_comp)
{
stbi__context s;
stbi__start_mem(&s,buffer,len);
- return stbi__load_flip(&s,x,y,comp,req_comp);
+ return stbi__load_and_postprocess_8bit(&s,x,y,comp,req_comp);
}
STBIDEF stbi_uc *stbi_load_from_callbacks(stbi_io_callbacks const *clbk, void *user, int *x, int *y, int *comp, int req_comp)
{
stbi__context s;
stbi__start_callbacks(&s, (stbi_io_callbacks *) clbk, user);
- return stbi__load_flip(&s,x,y,comp,req_comp);
+ return stbi__load_and_postprocess_8bit(&s,x,y,comp,req_comp);
+}
+
+#ifndef STBI_NO_GIF
+STBIDEF stbi_uc *stbi_load_gif_from_memory(stbi_uc const *buffer, int len, int **delays, int *x, int *y, int *z, int *comp, int req_comp)
+{
+ unsigned char *result;
+ stbi__context s;
+ stbi__start_mem(&s,buffer,len);
+
+ result = (unsigned char*) stbi__load_gif_main(&s, delays, x, y, z, comp, req_comp);
+ if (stbi__vertically_flip_on_load) {
+ stbi__vertical_flip_slices( result, *x, *y, *z, *comp );
+ }
+
+ return result;
}
+#endif
#ifndef STBI_NO_LINEAR
static float *stbi__loadf_main(stbi__context *s, int *x, int *y, int *comp, int req_comp)
@@ -1071,13 +1462,14 @@ static float *stbi__loadf_main(stbi__context *s, int *x, int *y, int *comp, int
unsigned char *data;
#ifndef STBI_NO_HDR
if (stbi__hdr_test(s)) {
- float *hdr_data = stbi__hdr_load(s,x,y,comp,req_comp);
+ stbi__result_info ri;
+ float *hdr_data = stbi__hdr_load(s,x,y,comp,req_comp, &ri);
if (hdr_data)
stbi__float_postprocess(hdr_data,x,y,comp,req_comp);
return hdr_data;
}
#endif
- data = stbi__load_flip(s, x, y, comp, req_comp);
+ data = stbi__load_and_postprocess_8bit(s, x, y, comp, req_comp);
if (data)
return stbi__ldr_to_hdr(data, *x, *y, req_comp ? req_comp : *comp);
return stbi__errpf("unknown image type", "Image not of any known type, or corrupt");
@@ -1147,13 +1539,18 @@ STBIDEF int stbi_is_hdr (char const *filename)
return result;
}
-STBIDEF int stbi_is_hdr_from_file(FILE *f)
+STBIDEF int stbi_is_hdr_from_file(FILE *f)
{
#ifndef STBI_NO_HDR
+ long pos = ftell(f);
+ int res;
stbi__context s;
stbi__start_file(&s,f);
- return stbi__hdr_test(&s);
+ res = stbi__hdr_test(&s);
+ fseek(f, pos, SEEK_SET);
+ return res;
#else
+ STBI_NOTUSED(f);
return 0;
#endif
}
@@ -1166,18 +1563,21 @@ STBIDEF int stbi_is_hdr_from_callbacks(stbi_io_callbacks const *clbk, void
stbi__start_callbacks(&s, (stbi_io_callbacks *) clbk, user);
return stbi__hdr_test(&s);
#else
+ STBI_NOTUSED(clbk);
+ STBI_NOTUSED(user);
return 0;
#endif
}
-static float stbi__h2l_gamma_i=1.0f/2.2f, stbi__h2l_scale_i=1.0f;
+#ifndef STBI_NO_LINEAR
static float stbi__l2h_gamma=2.2f, stbi__l2h_scale=1.0f;
-#ifndef STBI_NO_LINEAR
STBIDEF void stbi_ldr_to_hdr_gamma(float gamma) { stbi__l2h_gamma = gamma; }
STBIDEF void stbi_ldr_to_hdr_scale(float scale) { stbi__l2h_scale = scale; }
#endif
+static float stbi__h2l_gamma_i=1.0f/2.2f, stbi__h2l_scale_i=1.0f;
+
STBIDEF void stbi_hdr_to_ldr_gamma(float gamma) { stbi__h2l_gamma_i = 1/gamma; }
STBIDEF void stbi_hdr_to_ldr_scale(float scale) { stbi__h2l_scale_i = 1/scale; }
@@ -1197,6 +1597,7 @@ enum
static void stbi__refill_buffer(stbi__context *s)
{
int n = (s->io.read)(s->io_user_data,(char*)s->buffer_start,s->buflen);
+ s->callback_already_read += (int) (s->img_buffer - s->img_buffer_original);
if (n == 0) {
// at end of file, treat same as if from memory, but need to handle case
// where s->img_buffer isn't pointing to safe memory, e.g. 0-byte file
@@ -1221,6 +1622,9 @@ stbi_inline static stbi_uc stbi__get8(stbi__context *s)
return 0;
}
+#if defined(STBI_NO_JPEG) && defined(STBI_NO_HDR) && defined(STBI_NO_PIC) && defined(STBI_NO_PNM)
+// nothing
+#else
stbi_inline static int stbi__at_eof(stbi__context *s)
{
if (s->io.read) {
@@ -1232,9 +1636,14 @@ stbi_inline static int stbi__at_eof(stbi__context *s)
return s->img_buffer >= s->img_buffer_end;
}
+#endif
+#if defined(STBI_NO_JPEG) && defined(STBI_NO_PNG) && defined(STBI_NO_BMP) && defined(STBI_NO_PSD) && defined(STBI_NO_TGA) && defined(STBI_NO_GIF) && defined(STBI_NO_PIC)
+// nothing
+#else
static void stbi__skip(stbi__context *s, int n)
{
+ if (n == 0) return; // already there!
if (n < 0) {
s->img_buffer = s->img_buffer_end;
return;
@@ -1249,7 +1658,11 @@ static void stbi__skip(stbi__context *s, int n)
}
s->img_buffer += n;
}
+#endif
+#if defined(STBI_NO_PNG) && defined(STBI_NO_TGA) && defined(STBI_NO_HDR) && defined(STBI_NO_PNM)
+// nothing
+#else
static int stbi__getn(stbi__context *s, stbi_uc *buffer, int n)
{
if (s->io.read) {
@@ -1273,34 +1686,52 @@ static int stbi__getn(stbi__context *s, stbi_uc *buffer, int n)
} else
return 0;
}
+#endif
+#if defined(STBI_NO_JPEG) && defined(STBI_NO_PNG) && defined(STBI_NO_PSD) && defined(STBI_NO_PIC)
+// nothing
+#else
static int stbi__get16be(stbi__context *s)
{
int z = stbi__get8(s);
return (z << 8) + stbi__get8(s);
}
+#endif
+#if defined(STBI_NO_PNG) && defined(STBI_NO_PSD) && defined(STBI_NO_PIC)
+// nothing
+#else
static stbi__uint32 stbi__get32be(stbi__context *s)
{
stbi__uint32 z = stbi__get16be(s);
return (z << 16) + stbi__get16be(s);
}
+#endif
+#if defined(STBI_NO_BMP) && defined(STBI_NO_TGA) && defined(STBI_NO_GIF)
+// nothing
+#else
static int stbi__get16le(stbi__context *s)
{
int z = stbi__get8(s);
return z + (stbi__get8(s) << 8);
}
+#endif
+#ifndef STBI_NO_BMP
static stbi__uint32 stbi__get32le(stbi__context *s)
{
stbi__uint32 z = stbi__get16le(s);
- return z + (stbi__get16le(s) << 16);
+ z += (stbi__uint32)stbi__get16le(s) << 16;
+ return z;
}
+#endif
#define STBI__BYTECAST(x) ((stbi_uc) ((x) & 255)) // truncate int to byte without warnings
-
+#if defined(STBI_NO_JPEG) && defined(STBI_NO_PNG) && defined(STBI_NO_BMP) && defined(STBI_NO_PSD) && defined(STBI_NO_TGA) && defined(STBI_NO_GIF) && defined(STBI_NO_PIC) && defined(STBI_NO_PNM)
+// nothing
+#else
//////////////////////////////////////////////////////////////////////////////
//
// generic converter from built-in img_n to req_comp
@@ -1316,7 +1747,11 @@ static stbi_uc stbi__compute_y(int r, int g, int b)
{
return (stbi_uc) (((r*77) + (g*150) + (29*b)) >> 8);
}
+#endif
+#if defined(STBI_NO_PNG) && defined(STBI_NO_BMP) && defined(STBI_NO_PSD) && defined(STBI_NO_TGA) && defined(STBI_NO_GIF) && defined(STBI_NO_PIC) && defined(STBI_NO_PNM)
+// nothing
+#else
static unsigned char *stbi__convert_format(unsigned char *data, int img_n, int req_comp, unsigned int x, unsigned int y)
{
int i,j;
@@ -1325,7 +1760,7 @@ static unsigned char *stbi__convert_format(unsigned char *data, int img_n, int r
if (req_comp == img_n) return data;
STBI_ASSERT(req_comp >= 1 && req_comp <= 4);
- good = (unsigned char *) stbi__malloc(req_comp * x * y);
+ good = (unsigned char *) stbi__malloc_mad3(req_comp, x, y, 0);
if (good == NULL) {
STBI_FREE(data);
return stbi__errpuc("outofmem", "Out of memory");
@@ -1335,37 +1770,97 @@ static unsigned char *stbi__convert_format(unsigned char *data, int img_n, int r
unsigned char *src = data + j * x * img_n ;
unsigned char *dest = good + j * x * req_comp;
- #define COMBO(a,b) ((a)*8+(b))
- #define CASE(a,b) case COMBO(a,b): for(i=x-1; i >= 0; --i, src += a, dest += b)
+ #define STBI__COMBO(a,b) ((a)*8+(b))
+ #define STBI__CASE(a,b) case STBI__COMBO(a,b): for(i=x-1; i >= 0; --i, src += a, dest += b)
// convert source image with img_n components to one with req_comp components;
// avoid switch per pixel, so use switch per scanline and massive macros
- switch (COMBO(img_n, req_comp)) {
- CASE(1,2) dest[0]=src[0], dest[1]=255; break;
- CASE(1,3) dest[0]=dest[1]=dest[2]=src[0]; break;
- CASE(1,4) dest[0]=dest[1]=dest[2]=src[0], dest[3]=255; break;
- CASE(2,1) dest[0]=src[0]; break;
- CASE(2,3) dest[0]=dest[1]=dest[2]=src[0]; break;
- CASE(2,4) dest[0]=dest[1]=dest[2]=src[0], dest[3]=src[1]; break;
- CASE(3,4) dest[0]=src[0],dest[1]=src[1],dest[2]=src[2],dest[3]=255; break;
- CASE(3,1) dest[0]=stbi__compute_y(src[0],src[1],src[2]); break;
- CASE(3,2) dest[0]=stbi__compute_y(src[0],src[1],src[2]), dest[1] = 255; break;
- CASE(4,1) dest[0]=stbi__compute_y(src[0],src[1],src[2]); break;
- CASE(4,2) dest[0]=stbi__compute_y(src[0],src[1],src[2]), dest[1] = src[3]; break;
- CASE(4,3) dest[0]=src[0],dest[1]=src[1],dest[2]=src[2]; break;
- default: STBI_ASSERT(0);
+ switch (STBI__COMBO(img_n, req_comp)) {
+ STBI__CASE(1,2) { dest[0]=src[0]; dest[1]=255; } break;
+ STBI__CASE(1,3) { dest[0]=dest[1]=dest[2]=src[0]; } break;
+ STBI__CASE(1,4) { dest[0]=dest[1]=dest[2]=src[0]; dest[3]=255; } break;
+ STBI__CASE(2,1) { dest[0]=src[0]; } break;
+ STBI__CASE(2,3) { dest[0]=dest[1]=dest[2]=src[0]; } break;
+ STBI__CASE(2,4) { dest[0]=dest[1]=dest[2]=src[0]; dest[3]=src[1]; } break;
+ STBI__CASE(3,4) { dest[0]=src[0];dest[1]=src[1];dest[2]=src[2];dest[3]=255; } break;
+ STBI__CASE(3,1) { dest[0]=stbi__compute_y(src[0],src[1],src[2]); } break;
+ STBI__CASE(3,2) { dest[0]=stbi__compute_y(src[0],src[1],src[2]); dest[1] = 255; } break;
+ STBI__CASE(4,1) { dest[0]=stbi__compute_y(src[0],src[1],src[2]); } break;
+ STBI__CASE(4,2) { dest[0]=stbi__compute_y(src[0],src[1],src[2]); dest[1] = src[3]; } break;
+ STBI__CASE(4,3) { dest[0]=src[0];dest[1]=src[1];dest[2]=src[2]; } break;
+ default: STBI_ASSERT(0); STBI_FREE(data); STBI_FREE(good); return stbi__errpuc("unsupported", "Unsupported format conversion");
}
- #undef CASE
+ #undef STBI__CASE
}
STBI_FREE(data);
return good;
}
+#endif
+
+#if defined(STBI_NO_PNG) && defined(STBI_NO_PSD)
+// nothing
+#else
+static stbi__uint16 stbi__compute_y_16(int r, int g, int b)
+{
+ return (stbi__uint16) (((r*77) + (g*150) + (29*b)) >> 8);
+}
+#endif
+
+#if defined(STBI_NO_PNG) && defined(STBI_NO_PSD)
+// nothing
+#else
+static stbi__uint16 *stbi__convert_format16(stbi__uint16 *data, int img_n, int req_comp, unsigned int x, unsigned int y)
+{
+ int i,j;
+ stbi__uint16 *good;
+
+ if (req_comp == img_n) return data;
+ STBI_ASSERT(req_comp >= 1 && req_comp <= 4);
+
+ good = (stbi__uint16 *) stbi__malloc(req_comp * x * y * 2);
+ if (good == NULL) {
+ STBI_FREE(data);
+ return (stbi__uint16 *) stbi__errpuc("outofmem", "Out of memory");
+ }
+
+ for (j=0; j < (int) y; ++j) {
+ stbi__uint16 *src = data + j * x * img_n ;
+ stbi__uint16 *dest = good + j * x * req_comp;
+
+ #define STBI__COMBO(a,b) ((a)*8+(b))
+ #define STBI__CASE(a,b) case STBI__COMBO(a,b): for(i=x-1; i >= 0; --i, src += a, dest += b)
+ // convert source image with img_n components to one with req_comp components;
+ // avoid switch per pixel, so use switch per scanline and massive macros
+ switch (STBI__COMBO(img_n, req_comp)) {
+ STBI__CASE(1,2) { dest[0]=src[0]; dest[1]=0xffff; } break;
+ STBI__CASE(1,3) { dest[0]=dest[1]=dest[2]=src[0]; } break;
+ STBI__CASE(1,4) { dest[0]=dest[1]=dest[2]=src[0]; dest[3]=0xffff; } break;
+ STBI__CASE(2,1) { dest[0]=src[0]; } break;
+ STBI__CASE(2,3) { dest[0]=dest[1]=dest[2]=src[0]; } break;
+ STBI__CASE(2,4) { dest[0]=dest[1]=dest[2]=src[0]; dest[3]=src[1]; } break;
+ STBI__CASE(3,4) { dest[0]=src[0];dest[1]=src[1];dest[2]=src[2];dest[3]=0xffff; } break;
+ STBI__CASE(3,1) { dest[0]=stbi__compute_y_16(src[0],src[1],src[2]); } break;
+ STBI__CASE(3,2) { dest[0]=stbi__compute_y_16(src[0],src[1],src[2]); dest[1] = 0xffff; } break;
+ STBI__CASE(4,1) { dest[0]=stbi__compute_y_16(src[0],src[1],src[2]); } break;
+ STBI__CASE(4,2) { dest[0]=stbi__compute_y_16(src[0],src[1],src[2]); dest[1] = src[3]; } break;
+ STBI__CASE(4,3) { dest[0]=src[0];dest[1]=src[1];dest[2]=src[2]; } break;
+ default: STBI_ASSERT(0); STBI_FREE(data); STBI_FREE(good); return (stbi__uint16*) stbi__errpuc("unsupported", "Unsupported format conversion");
+ }
+ #undef STBI__CASE
+ }
+
+ STBI_FREE(data);
+ return good;
+}
+#endif
#ifndef STBI_NO_LINEAR
static float *stbi__ldr_to_hdr(stbi_uc *data, int x, int y, int comp)
{
int i,k,n;
- float *output = (float *) stbi__malloc(x * y * comp * sizeof(float));
+ float *output;
+ if (!data) return NULL;
+ output = (float *) stbi__malloc_mad4(x, y, comp, sizeof(float), 0);
if (output == NULL) { STBI_FREE(data); return stbi__errpf("outofmem", "Out of memory"); }
// compute number of non-alpha components
if (comp & 1) n = comp; else n = comp-1;
@@ -1373,7 +1868,11 @@ static float *stbi__ldr_to_hdr(stbi_uc *data, int x, int y, int comp)
for (k=0; k < n; ++k) {
output[i*comp + k] = (float) (pow(data[i*comp+k]/255.0f, stbi__l2h_gamma) * stbi__l2h_scale);
}
- if (k < comp) output[i*comp + k] = data[i*comp+k]/255.0f;
+ }
+ if (n < comp) {
+ for (i=0; i < x*y; ++i) {
+ output[i*comp + n] = data[i*comp + n]/255.0f;
+ }
}
STBI_FREE(data);
return output;
@@ -1385,7 +1884,9 @@ static float *stbi__ldr_to_hdr(stbi_uc *data, int x, int y, int comp)
static stbi_uc *stbi__hdr_to_ldr(float *data, int x, int y, int comp)
{
int i,k,n;
- stbi_uc *output = (stbi_uc *) stbi__malloc(x * y * comp);
+ stbi_uc *output;
+ if (!data) return NULL;
+ output = (stbi_uc *) stbi__malloc_mad3(x, y, comp, 0);
if (output == NULL) { STBI_FREE(data); return stbi__errpuc("outofmem", "Out of memory"); }
// compute number of non-alpha components
if (comp & 1) n = comp; else n = comp-1;
@@ -1450,7 +1951,7 @@ typedef struct
stbi__context *s;
stbi__huffman huff_dc[4];
stbi__huffman huff_ac[4];
- stbi_uc dequant[4][64];
+ stbi__uint16 dequant[4][64];
stbi__int16 fast_ac[4][1 << FAST_BITS];
// sizes for components, interleaved MCUs
@@ -1486,6 +1987,9 @@ typedef struct
int succ_high;
int succ_low;
int eob_run;
+ int jfif;
+ int app14_color_transform; // Adobe APP14 tag
+ int rgb;
int scan_n, order[4];
int restart_interval, todo;
@@ -1498,11 +2002,15 @@ typedef struct
static int stbi__build_huffman(stbi__huffman *h, int *count)
{
- int i,j,k=0,code;
+ int i,j,k=0;
+ unsigned int code;
// build size list for each symbol (from JPEG spec)
- for (i=0; i < 16; ++i)
- for (j=0; j < count[i]; ++j)
+ for (i=0; i < 16; ++i) {
+ for (j=0; j < count[i]; ++j) {
h->size[k++] = (stbi_uc) (i+1);
+ if(k >= 257) return stbi__err("bad size list","Corrupt JPEG");
+ }
+ }
h->size[k] = 0;
// compute actual symbols (from jpeg spec)
@@ -1514,7 +2022,7 @@ static int stbi__build_huffman(stbi__huffman *h, int *count)
if (h->size[k] == j) {
while (h->size[k] == j)
h->code[k++] = (stbi__uint16) (code++);
- if (code-1 >= (1 << j)) return stbi__err("bad code lengths","Corrupt JPEG");
+ if (code-1 >= (1u << j)) return stbi__err("bad code lengths","Corrupt JPEG");
}
// compute largest code + 1 for this size, preshifted as needed later
h->maxcode[j] = code << (16-j);
@@ -1555,10 +2063,10 @@ static void stbi__build_fast_ac(stbi__int16 *fast_ac, stbi__huffman *h)
// magnitude code followed by receive_extend code
int k = ((i << len) & ((1 << FAST_BITS) - 1)) >> (FAST_BITS - magbits);
int m = 1 << (magbits - 1);
- if (k < m) k += (-1 << magbits) + 1;
+ if (k < m) k += (~0U << magbits) + 1;
// if the result is small enough, we can fit it in fast_ac table
if (k >= -128 && k <= 127)
- fast_ac[i] = (stbi__int16) ((k << 8) + (run << 4) + (len + magbits));
+ fast_ac[i] = (stbi__int16) ((k * 256) + (run * 16) + (len + magbits));
}
}
}
@@ -1567,9 +2075,10 @@ static void stbi__build_fast_ac(stbi__int16 *fast_ac, stbi__huffman *h)
static void stbi__grow_buffer_unsafe(stbi__jpeg *j)
{
do {
- int b = j->nomore ? 0 : stbi__get8(j->s);
+ unsigned int b = j->nomore ? 0 : stbi__get8(j->s);
if (b == 0xff) {
int c = stbi__get8(j->s);
+ while (c == 0xff) c = stbi__get8(j->s); // consume fill bytes
if (c != 0) {
j->marker = (unsigned char) c;
j->nomore = 1;
@@ -1582,7 +2091,7 @@ static void stbi__grow_buffer_unsafe(stbi__jpeg *j)
}
// (1 << n) - 1
-static stbi__uint32 stbi__bmask[17]={0,1,3,7,15,31,63,127,255,511,1023,2047,4095,8191,16383,32767,65535};
+static const stbi__uint32 stbi__bmask[17]={0,1,3,7,15,31,63,127,255,511,1023,2047,4095,8191,16383,32767,65535};
// decode a jpeg huffman value from the bitstream
stbi_inline static int stbi__jpeg_huff_decode(stbi__jpeg *j, stbi__huffman *h)
@@ -1626,6 +2135,8 @@ stbi_inline static int stbi__jpeg_huff_decode(stbi__jpeg *j, stbi__huffman *h)
// convert the huffman code to the symbol id
c = ((j->code_buffer >> (32 - k)) & stbi__bmask[k]) + h->delta[k];
+ if(c < 0 || c >= 256) // symbol id out of bounds!
+ return -1;
STBI_ASSERT((((j->code_buffer) >> (32 - h->size[c])) & stbi__bmask[h->size[c]]) == h->code[c]);
// convert the id to a symbol
@@ -1635,7 +2146,7 @@ stbi_inline static int stbi__jpeg_huff_decode(stbi__jpeg *j, stbi__huffman *h)
}
// bias[n] = (-1<code_bits < n) stbi__grow_buffer_unsafe(j);
+ if (j->code_bits < n) return 0; // ran out of bits from stream, return 0s intead of continuing
- sgn = (stbi__int32)j->code_buffer >> 31; // sign bit is always in MSB
+ sgn = j->code_buffer >> 31; // sign bit always in MSB; 0 if MSB clear (positive), 1 if MSB set (negative)
k = stbi_lrot(j->code_buffer, n);
- STBI_ASSERT(n >= 0 && n < (int) (sizeof(stbi__bmask)/sizeof(*stbi__bmask)));
j->code_buffer = k & ~stbi__bmask[n];
k &= stbi__bmask[n];
j->code_bits -= n;
- return k + (stbi__jbias[n] & ~sgn);
+ return k + (stbi__jbias[n] & (sgn - 1));
}
// get some unsigned bits
@@ -1659,6 +2170,7 @@ stbi_inline static int stbi__jpeg_get_bits(stbi__jpeg *j, int n)
{
unsigned int k;
if (j->code_bits < n) stbi__grow_buffer_unsafe(j);
+ if (j->code_bits < n) return 0; // ran out of bits from stream, return 0s intead of continuing
k = stbi_lrot(j->code_buffer, n);
j->code_buffer = k & ~stbi__bmask[n];
k &= stbi__bmask[n];
@@ -1670,6 +2182,7 @@ stbi_inline static int stbi__jpeg_get_bit(stbi__jpeg *j)
{
unsigned int k;
if (j->code_bits < 1) stbi__grow_buffer_unsafe(j);
+ if (j->code_bits < 1) return 0; // ran out of bits from stream, return 0s intead of continuing
k = j->code_buffer;
j->code_buffer <<= 1;
--j->code_bits;
@@ -1678,7 +2191,7 @@ stbi_inline static int stbi__jpeg_get_bit(stbi__jpeg *j)
// given a value that's at position X in the zigzag stream,
// where does it appear in the 8x8 matrix coded as row-major?
-static stbi_uc stbi__jpeg_dezigzag[64+15] =
+static const stbi_uc stbi__jpeg_dezigzag[64+15] =
{
0, 1, 8, 16, 9, 2, 3, 10,
17, 24, 32, 25, 18, 11, 4, 5,
@@ -1694,21 +2207,23 @@ static stbi_uc stbi__jpeg_dezigzag[64+15] =
};
// decode one 64-entry block--
-static int stbi__jpeg_decode_block(stbi__jpeg *j, short data[64], stbi__huffman *hdc, stbi__huffman *hac, stbi__int16 *fac, int b, stbi_uc *dequant)
+static int stbi__jpeg_decode_block(stbi__jpeg *j, short data[64], stbi__huffman *hdc, stbi__huffman *hac, stbi__int16 *fac, int b, stbi__uint16 *dequant)
{
int diff,dc,k;
int t;
if (j->code_bits < 16) stbi__grow_buffer_unsafe(j);
t = stbi__jpeg_huff_decode(j, hdc);
- if (t < 0) return stbi__err("bad huffman code","Corrupt JPEG");
+ if (t < 0 || t > 15) return stbi__err("bad huffman code","Corrupt JPEG");
// 0 all the ac values now so we can do it 32-bits at a time
memset(data,0,64*sizeof(data[0]));
diff = t ? stbi__extend_receive(j, t) : 0;
+ if (!stbi__addints_valid(j->img_comp[b].dc_pred, diff)) return stbi__err("bad delta","Corrupt JPEG");
dc = j->img_comp[b].dc_pred + diff;
j->img_comp[b].dc_pred = dc;
+ if (!stbi__mul2shorts_valid(dc, dequant[0])) return stbi__err("can't merge dc and ac", "Corrupt JPEG");
data[0] = (short) (dc * dequant[0]);
// decode AC components, see JPEG spec
@@ -1722,6 +2237,7 @@ static int stbi__jpeg_decode_block(stbi__jpeg *j, short data[64], stbi__huffman
if (r) { // fast-AC path
k += (r >> 4) & 15; // run
s = r & 15; // combined length
+ if (s > j->code_bits) return stbi__err("bad huffman code", "Combined length longer than code bits available");
j->code_buffer <<= s;
j->code_bits -= s;
// decode into unzigzag'd location
@@ -1758,11 +2274,14 @@ static int stbi__jpeg_decode_block_prog_dc(stbi__jpeg *j, short data[64], stbi__
// first scan for DC coefficient, must be first
memset(data,0,64*sizeof(data[0])); // 0 all the ac values now
t = stbi__jpeg_huff_decode(j, hdc);
+ if (t < 0 || t > 15) return stbi__err("can't merge dc and ac", "Corrupt JPEG");
diff = t ? stbi__extend_receive(j, t) : 0;
+ if (!stbi__addints_valid(j->img_comp[b].dc_pred, diff)) return stbi__err("bad delta", "Corrupt JPEG");
dc = j->img_comp[b].dc_pred + diff;
j->img_comp[b].dc_pred = dc;
- data[0] = (short) (dc << j->succ_low);
+ if (!stbi__mul2shorts_valid(dc, 1 << j->succ_low)) return stbi__err("can't merge dc and ac", "Corrupt JPEG");
+ data[0] = (short) (dc * (1 << j->succ_low));
} else {
// refinement scan for DC coefficient
if (stbi__jpeg_get_bit(j))
@@ -1796,10 +2315,11 @@ static int stbi__jpeg_decode_block_prog_ac(stbi__jpeg *j, short data[64], stbi__
if (r) { // fast-AC path
k += (r >> 4) & 15; // run
s = r & 15; // combined length
+ if (s > j->code_bits) return stbi__err("bad huffman code", "Combined length longer than code bits available");
j->code_buffer <<= s;
j->code_bits -= s;
zig = stbi__jpeg_dezigzag[k++];
- data[zig] = (short) ((r >> 8) << shift);
+ data[zig] = (short) ((r >> 8) * (1 << shift));
} else {
int rs = stbi__jpeg_huff_decode(j, hac);
if (rs < 0) return stbi__err("bad huffman code","Corrupt JPEG");
@@ -1817,7 +2337,7 @@ static int stbi__jpeg_decode_block_prog_ac(stbi__jpeg *j, short data[64], stbi__
} else {
k += r;
zig = stbi__jpeg_dezigzag[k++];
- data[zig] = (short) (stbi__extend_receive(j,s) << shift);
+ data[zig] = (short) (stbi__extend_receive(j,s) * (1 << shift));
}
}
} while (k <= j->spec_end);
@@ -1904,7 +2424,7 @@ stbi_inline static stbi_uc stbi__clamp(int x)
}
#define stbi__f2f(x) ((int) (((x) * 4096 + 0.5)))
-#define stbi__fsh(x) ((x) << 12)
+#define stbi__fsh(x) ((x) * 4096)
// derived from jidctint -- DCT_ISLOW
#define STBI__IDCT_1D(s0,s1,s2,s3,s4,s5,s6,s7) \
@@ -1959,7 +2479,7 @@ static void stbi__idct_block(stbi_uc *out, int out_stride, short data[64])
// (1|2|3|4|5|6|7)==0 0 seconds
// all separate -0.047 seconds
// 1 && 2|3 && 4|5 && 6|7: -0.047 seconds
- int dcterm = d[0] << 2;
+ int dcterm = d[0]*4;
v[0] = v[8] = v[16] = v[24] = v[32] = v[40] = v[48] = v[56] = dcterm;
} else {
STBI__IDCT_1D(d[ 0],d[ 8],d[16],d[24],d[32],d[40],d[48],d[56])
@@ -2403,7 +2923,7 @@ static stbi_uc stbi__get_marker(stbi__jpeg *j)
x = stbi__get8(j->s);
if (x != 0xff) return STBI__MARKER_none;
while (x == 0xff)
- x = stbi__get8(j->s);
+ x = stbi__get8(j->s); // consume repeated 0xff fill bytes
return x;
}
@@ -2418,7 +2938,7 @@ static void stbi__jpeg_reset(stbi__jpeg *j)
j->code_bits = 0;
j->code_buffer = 0;
j->nomore = 0;
- j->img_comp[0].dc_pred = j->img_comp[1].dc_pred = j->img_comp[2].dc_pred = 0;
+ j->img_comp[0].dc_pred = j->img_comp[1].dc_pred = j->img_comp[2].dc_pred = j->img_comp[3].dc_pred = 0;
j->marker = STBI__MARKER_none;
j->todo = j->restart_interval ? j->restart_interval : 0x7fffffff;
j->eob_run = 0;
@@ -2550,7 +3070,7 @@ static int stbi__parse_entropy_coded_data(stbi__jpeg *z)
}
}
-static void stbi__jpeg_dequantize(short *data, stbi_uc *dequant)
+static void stbi__jpeg_dequantize(short *data, stbi__uint16 *dequant)
{
int i;
for (i=0; i < 64; ++i)
@@ -2592,13 +3112,14 @@ static int stbi__process_marker(stbi__jpeg *z, int m)
L = stbi__get16be(z->s)-2;
while (L > 0) {
int q = stbi__get8(z->s);
- int p = q >> 4;
+ int p = q >> 4, sixteen = (p != 0);
int t = q & 15,i;
- if (p != 0) return stbi__err("bad DQT type","Corrupt JPEG");
+ if (p != 0 && p != 1) return stbi__err("bad DQT type","Corrupt JPEG");
if (t > 3) return stbi__err("bad DQT table","Corrupt JPEG");
+
for (i=0; i < 64; ++i)
- z->dequant[t][stbi__jpeg_dezigzag[i]] = stbi__get8(z->s);
- L -= 65;
+ z->dequant[t][stbi__jpeg_dezigzag[i]] = (stbi__uint16)(sixteen ? stbi__get16be(z->s) : stbi__get8(z->s));
+ L -= (sixteen ? 129 : 65);
}
return L==0;
@@ -2615,6 +3136,7 @@ static int stbi__process_marker(stbi__jpeg *z, int m)
sizes[i] = stbi__get8(z->s);
n += sizes[i];
}
+ if(n > 256) return stbi__err("bad DHT header","Corrupt JPEG"); // Loop over i < n would write past end of values!
L -= 17;
if (tc == 0) {
if (!stbi__build_huffman(z->huff_dc+th, sizes)) return 0;
@@ -2631,12 +3153,50 @@ static int stbi__process_marker(stbi__jpeg *z, int m)
}
return L==0;
}
+
// check for comment block or APP blocks
if ((m >= 0xE0 && m <= 0xEF) || m == 0xFE) {
- stbi__skip(z->s, stbi__get16be(z->s)-2);
+ L = stbi__get16be(z->s);
+ if (L < 2) {
+ if (m == 0xFE)
+ return stbi__err("bad COM len","Corrupt JPEG");
+ else
+ return stbi__err("bad APP len","Corrupt JPEG");
+ }
+ L -= 2;
+
+ if (m == 0xE0 && L >= 5) { // JFIF APP0 segment
+ static const unsigned char tag[5] = {'J','F','I','F','\0'};
+ int ok = 1;
+ int i;
+ for (i=0; i < 5; ++i)
+ if (stbi__get8(z->s) != tag[i])
+ ok = 0;
+ L -= 5;
+ if (ok)
+ z->jfif = 1;
+ } else if (m == 0xEE && L >= 12) { // Adobe APP14 segment
+ static const unsigned char tag[6] = {'A','d','o','b','e','\0'};
+ int ok = 1;
+ int i;
+ for (i=0; i < 6; ++i)
+ if (stbi__get8(z->s) != tag[i])
+ ok = 0;
+ L -= 6;
+ if (ok) {
+ stbi__get8(z->s); // version
+ stbi__get16be(z->s); // flags0
+ stbi__get16be(z->s); // flags1
+ z->app14_color_transform = stbi__get8(z->s); // color transform
+ L -= 6;
+ }
+ }
+
+ stbi__skip(z->s, L);
return 1;
}
- return 0;
+
+ return stbi__err("unknown marker","Corrupt JPEG");
}
// after we see SOS
@@ -2675,8 +3235,30 @@ static int stbi__process_scan_header(stbi__jpeg *z)
z->spec_end = 63;
}
}
-
- return 1;
+
+ return 1;
+}
+
+static int stbi__free_jpeg_components(stbi__jpeg *z, int ncomp, int why)
+{
+ int i;
+ for (i=0; i < ncomp; ++i) {
+ if (z->img_comp[i].raw_data) {
+ STBI_FREE(z->img_comp[i].raw_data);
+ z->img_comp[i].raw_data = NULL;
+ z->img_comp[i].data = NULL;
+ }
+ if (z->img_comp[i].raw_coeff) {
+ STBI_FREE(z->img_comp[i].raw_coeff);
+ z->img_comp[i].raw_coeff = 0;
+ z->img_comp[i].coeff = 0;
+ }
+ if (z->img_comp[i].linebuf) {
+ STBI_FREE(z->img_comp[i].linebuf);
+ z->img_comp[i].linebuf = NULL;
+ }
+ }
+ return why;
}
static int stbi__process_frame_header(stbi__jpeg *z, int scan)
@@ -2687,8 +3269,10 @@ static int stbi__process_frame_header(stbi__jpeg *z, int scan)
p = stbi__get8(s); if (p != 8) return stbi__err("only 8-bit","JPEG format not supported: 8-bit only"); // JPEG baseline
s->img_y = stbi__get16be(s); if (s->img_y == 0) return stbi__err("no header height", "JPEG format not supported: delayed height"); // Legal, but we don't handle it--but neither does IJG
s->img_x = stbi__get16be(s); if (s->img_x == 0) return stbi__err("0 width","Corrupt JPEG"); // JPEG requires
+ if (s->img_y > STBI_MAX_DIMENSIONS) return stbi__err("too large","Very large image (corrupt?)");
+ if (s->img_x > STBI_MAX_DIMENSIONS) return stbi__err("too large","Very large image (corrupt?)");
c = stbi__get8(s);
- if (c != 3 && c != 1) return stbi__err("bad component count","Corrupt JPEG"); // JFIF requires
+ if (c != 3 && c != 1 && c != 4) return stbi__err("bad component count","Corrupt JPEG");
s->img_n = c;
for (i=0; i < c; ++i) {
z->img_comp[i].data = NULL;
@@ -2697,11 +3281,12 @@ static int stbi__process_frame_header(stbi__jpeg *z, int scan)
if (Lf != 8+3*s->img_n) return stbi__err("bad SOF len","Corrupt JPEG");
+ z->rgb = 0;
for (i=0; i < s->img_n; ++i) {
+ static const unsigned char rgb[3] = { 'R', 'G', 'B' };
z->img_comp[i].id = stbi__get8(s);
- if (z->img_comp[i].id != i+1) // JFIF requires
- if (z->img_comp[i].id != i) // some version of jpegtran outputs non-JFIF-compliant files!
- return stbi__err("bad component ID","Corrupt JPEG");
+ if (s->img_n == 3 && z->img_comp[i].id == rgb[i])
+ ++z->rgb;
q = stbi__get8(s);
z->img_comp[i].h = (q >> 4); if (!z->img_comp[i].h || z->img_comp[i].h > 4) return stbi__err("bad H","Corrupt JPEG");
z->img_comp[i].v = q & 15; if (!z->img_comp[i].v || z->img_comp[i].v > 4) return stbi__err("bad V","Corrupt JPEG");
@@ -2710,18 +3295,26 @@ static int stbi__process_frame_header(stbi__jpeg *z, int scan)
if (scan != STBI__SCAN_load) return 1;
- if ((1 << 30) / s->img_x / s->img_n < s->img_y) return stbi__err("too large", "Image too large to decode");
+ if (!stbi__mad3sizes_valid(s->img_x, s->img_y, s->img_n, 0)) return stbi__err("too large", "Image too large to decode");
for (i=0; i < s->img_n; ++i) {
if (z->img_comp[i].h > h_max) h_max = z->img_comp[i].h;
if (z->img_comp[i].v > v_max) v_max = z->img_comp[i].v;
}
+ // check that plane subsampling factors are integer ratios; our resamplers can't deal with fractional ratios
+ // and I've never seen a non-corrupted JPEG file actually use them
+ for (i=0; i < s->img_n; ++i) {
+ if (h_max % z->img_comp[i].h != 0) return stbi__err("bad H","Corrupt JPEG");
+ if (v_max % z->img_comp[i].v != 0) return stbi__err("bad V","Corrupt JPEG");
+ }
+
// compute interleaved mcu info
z->img_h_max = h_max;
z->img_v_max = v_max;
z->img_mcu_w = h_max * 8;
z->img_mcu_h = v_max * 8;
+ // these sizes can't be more than 17 bits
z->img_mcu_x = (s->img_x + z->img_mcu_w-1) / z->img_mcu_w;
z->img_mcu_y = (s->img_y + z->img_mcu_h-1) / z->img_mcu_h;
@@ -2733,28 +3326,27 @@ static int stbi__process_frame_header(stbi__jpeg *z, int scan)
// the bogus oversized data from using interleaved MCUs and their
// big blocks (e.g. a 16x16 iMCU on an image of width 33); we won't
// discard the extra data until colorspace conversion
+ //
+ // img_mcu_x, img_mcu_y: <=17 bits; comp[i].h and .v are <=4 (checked earlier)
+ // so these muls can't overflow with 32-bit ints (which we require)
z->img_comp[i].w2 = z->img_mcu_x * z->img_comp[i].h * 8;
z->img_comp[i].h2 = z->img_mcu_y * z->img_comp[i].v * 8;
- z->img_comp[i].raw_data = stbi__malloc(z->img_comp[i].w2 * z->img_comp[i].h2+15);
-
- if (z->img_comp[i].raw_data == NULL) {
- for(--i; i >= 0; --i) {
- STBI_FREE(z->img_comp[i].raw_data);
- z->img_comp[i].data = NULL;
- }
- return stbi__err("outofmem", "Out of memory");
- }
+ z->img_comp[i].coeff = 0;
+ z->img_comp[i].raw_coeff = 0;
+ z->img_comp[i].linebuf = NULL;
+ z->img_comp[i].raw_data = stbi__malloc_mad2(z->img_comp[i].w2, z->img_comp[i].h2, 15);
+ if (z->img_comp[i].raw_data == NULL)
+ return stbi__free_jpeg_components(z, i+1, stbi__err("outofmem", "Out of memory"));
// align blocks for idct using mmx/sse
z->img_comp[i].data = (stbi_uc*) (((size_t) z->img_comp[i].raw_data + 15) & ~15);
- z->img_comp[i].linebuf = NULL;
if (z->progressive) {
- z->img_comp[i].coeff_w = (z->img_comp[i].w2 + 7) >> 3;
- z->img_comp[i].coeff_h = (z->img_comp[i].h2 + 7) >> 3;
- z->img_comp[i].raw_coeff = STBI_MALLOC(z->img_comp[i].coeff_w * z->img_comp[i].coeff_h * 64 * sizeof(short) + 15);
+ // w2, h2 are multiples of 8 (see above)
+ z->img_comp[i].coeff_w = z->img_comp[i].w2 / 8;
+ z->img_comp[i].coeff_h = z->img_comp[i].h2 / 8;
+ z->img_comp[i].raw_coeff = stbi__malloc_mad3(z->img_comp[i].w2, z->img_comp[i].h2, sizeof(short), 15);
+ if (z->img_comp[i].raw_coeff == NULL)
+ return stbi__free_jpeg_components(z, i+1, stbi__err("outofmem", "Out of memory"));
z->img_comp[i].coeff = (short*) (((size_t) z->img_comp[i].raw_coeff + 15) & ~15);
- } else {
- z->img_comp[i].coeff = 0;
- z->img_comp[i].raw_coeff = 0;
}
}
@@ -2773,6 +3365,8 @@ static int stbi__process_frame_header(stbi__jpeg *z, int scan)
static int stbi__decode_jpeg_header(stbi__jpeg *z, int scan)
{
int m;
+ z->jfif = 0;
+ z->app14_color_transform = -1; // valid values are 0,1,2
z->marker = STBI__MARKER_none; // initialize cached marker to empty
m = stbi__get_marker(z);
if (!stbi__SOI(m)) return stbi__err("no SOI","Corrupt JPEG");
@@ -2792,6 +3386,28 @@ static int stbi__decode_jpeg_header(stbi__jpeg *z, int scan)
return 1;
}
+static stbi_uc stbi__skip_jpeg_junk_at_end(stbi__jpeg *j)
+{
+ // some JPEGs have junk at end, skip over it but if we find what looks
+ // like a valid marker, resume there
+ while (!stbi__at_eof(j->s)) {
+ stbi_uc x = stbi__get8(j->s);
+ while (x == 0xff) { // might be a marker
+ if (stbi__at_eof(j->s)) return STBI__MARKER_none;
+ x = stbi__get8(j->s);
+ if (x != 0x00 && x != 0xff) {
+ // not a stuffed zero or lead-in to another marker, looks
+ // like an actual marker, return it
+ return x;
+ }
+ // stuffed zero has x=0 now which ends the loop, meaning we go
+ // back to regular scan loop.
+ // repeated 0xff keeps trying to read the next byte of the marker.
+ }
+ }
+ return STBI__MARKER_none;
+}
+
// decode image to YCbCr format
static int stbi__decode_jpeg_image(stbi__jpeg *j)
{
@@ -2808,22 +3424,22 @@ static int stbi__decode_jpeg_image(stbi__jpeg *j)
if (!stbi__process_scan_header(j)) return 0;
if (!stbi__parse_entropy_coded_data(j)) return 0;
if (j->marker == STBI__MARKER_none ) {
- // handle 0s at the end of image data from IP Kamera 9060
- while (!stbi__at_eof(j->s)) {
- int x = stbi__get8(j->s);
- if (x == 255) {
- j->marker = stbi__get8(j->s);
- break;
- } else if (x != 0) {
- return stbi__err("junk before marker", "Corrupt JPEG");
- }
- }
+ j->marker = stbi__skip_jpeg_junk_at_end(j);
// if we reach eof without hitting a marker, stbi__get_marker() below will fail and we'll eventually return 0
}
+ m = stbi__get_marker(j);
+ if (STBI__RESTART(m))
+ m = stbi__get_marker(j);
+ } else if (stbi__DNL(m)) {
+ int Ld = stbi__get16be(j->s);
+ stbi__uint32 NL = stbi__get16be(j->s);
+ if (Ld != 4) return stbi__err("bad DNL len", "Corrupt JPEG");
+ if (NL != j->s->img_y) return stbi__err("bad DNL height", "Corrupt JPEG");
+ m = stbi__get_marker(j);
} else {
- if (!stbi__process_marker(j, m)) return 0;
+ if (!stbi__process_marker(j, m)) return 1;
+ m = stbi__get_marker(j);
}
- m = stbi__get_marker(j);
}
if (j->progressive)
stbi__jpeg_finish(j);
@@ -3038,38 +3654,9 @@ static stbi_uc *stbi__resample_row_generic(stbi_uc *out, stbi_uc *in_near, stbi_
return out;
}
-#ifdef STBI_JPEG_OLD
-// this is the same YCbCr-to-RGB calculation that stb_image has used
-// historically before the algorithm changes in 1.49
-#define float2fixed(x) ((int) ((x) * 65536 + 0.5))
-static void stbi__YCbCr_to_RGB_row(stbi_uc *out, const stbi_uc *y, const stbi_uc *pcb, const stbi_uc *pcr, int count, int step)
-{
- int i;
- for (i=0; i < count; ++i) {
- int y_fixed = (y[i] << 16) + 32768; // rounding
- int r,g,b;
- int cr = pcr[i] - 128;
- int cb = pcb[i] - 128;
- r = y_fixed + cr*float2fixed(1.40200f);
- g = y_fixed - cr*float2fixed(0.71414f) - cb*float2fixed(0.34414f);
- b = y_fixed + cb*float2fixed(1.77200f);
- r >>= 16;
- g >>= 16;
- b >>= 16;
- if ((unsigned) r > 255) { if (r < 0) r = 0; else r = 255; }
- if ((unsigned) g > 255) { if (g < 0) g = 0; else g = 255; }
- if ((unsigned) b > 255) { if (b < 0) b = 0; else b = 255; }
- out[0] = (stbi_uc)r;
- out[1] = (stbi_uc)g;
- out[2] = (stbi_uc)b;
- out[3] = 255;
- out += step;
- }
-}
-#else
// this is a reduced-precision calculation of YCbCr-to-RGB introduced
// to make sure the code produces the same results in both SIMD and scalar
-#define float2fixed(x) (((int) ((x) * 4096.0f + 0.5f)) << 8)
+#define stbi__float2fixed(x) (((int) ((x) * 4096.0f + 0.5f)) << 8)
static void stbi__YCbCr_to_RGB_row(stbi_uc *out, const stbi_uc *y, const stbi_uc *pcb, const stbi_uc *pcr, int count, int step)
{
int i;
@@ -3078,9 +3665,9 @@ static void stbi__YCbCr_to_RGB_row(stbi_uc *out, const stbi_uc *y, const stbi_uc
int r,g,b;
int cr = pcr[i] - 128;
int cb = pcb[i] - 128;
- r = y_fixed + cr* float2fixed(1.40200f);
- g = y_fixed + (cr*-float2fixed(0.71414f)) + ((cb*-float2fixed(0.34414f)) & 0xffff0000);
- b = y_fixed + cb* float2fixed(1.77200f);
+ r = y_fixed + cr* stbi__float2fixed(1.40200f);
+ g = y_fixed + (cr*-stbi__float2fixed(0.71414f)) + ((cb*-stbi__float2fixed(0.34414f)) & 0xffff0000);
+ b = y_fixed + cb* stbi__float2fixed(1.77200f);
r >>= 20;
g >>= 20;
b >>= 20;
@@ -3094,7 +3681,6 @@ static void stbi__YCbCr_to_RGB_row(stbi_uc *out, const stbi_uc *y, const stbi_uc
out += step;
}
}
-#endif
#if defined(STBI_SSE2) || defined(STBI_NEON)
static void stbi__YCbCr_to_RGB_simd(stbi_uc *out, stbi_uc const *y, stbi_uc const *pcb, stbi_uc const *pcr, int count, int step)
@@ -3213,9 +3799,9 @@ static void stbi__YCbCr_to_RGB_simd(stbi_uc *out, stbi_uc const *y, stbi_uc cons
int r,g,b;
int cr = pcr[i] - 128;
int cb = pcb[i] - 128;
- r = y_fixed + cr* float2fixed(1.40200f);
- g = y_fixed + cr*-float2fixed(0.71414f) + ((cb*-float2fixed(0.34414f)) & 0xffff0000);
- b = y_fixed + cb* float2fixed(1.77200f);
+ r = y_fixed + cr* stbi__float2fixed(1.40200f);
+ g = y_fixed + cr*-stbi__float2fixed(0.71414f) + ((cb*-stbi__float2fixed(0.34414f)) & 0xffff0000);
+ b = y_fixed + cb* stbi__float2fixed(1.77200f);
r >>= 20;
g >>= 20;
b >>= 20;
@@ -3241,18 +3827,14 @@ static void stbi__setup_jpeg(stbi__jpeg *j)
#ifdef STBI_SSE2
if (stbi__sse2_available()) {
j->idct_block_kernel = stbi__idct_simd;
- #ifndef STBI_JPEG_OLD
j->YCbCr_to_RGB_kernel = stbi__YCbCr_to_RGB_simd;
- #endif
j->resample_row_hv_2_kernel = stbi__resample_row_hv_2_simd;
}
#endif
#ifdef STBI_NEON
j->idct_block_kernel = stbi__idct_simd;
- #ifndef STBI_JPEG_OLD
j->YCbCr_to_RGB_kernel = stbi__YCbCr_to_RGB_simd;
- #endif
j->resample_row_hv_2_kernel = stbi__resample_row_hv_2_simd;
#endif
}
@@ -3260,23 +3842,7 @@ static void stbi__setup_jpeg(stbi__jpeg *j)
// clean up the temporary component buffers
static void stbi__cleanup_jpeg(stbi__jpeg *j)
{
- int i;
- for (i=0; i < j->s->img_n; ++i) {
- if (j->img_comp[i].raw_data) {
- STBI_FREE(j->img_comp[i].raw_data);
- j->img_comp[i].raw_data = NULL;
- j->img_comp[i].data = NULL;
- }
- if (j->img_comp[i].raw_coeff) {
- STBI_FREE(j->img_comp[i].raw_coeff);
- j->img_comp[i].raw_coeff = 0;
- j->img_comp[i].coeff = 0;
- }
- if (j->img_comp[i].linebuf) {
- STBI_FREE(j->img_comp[i].linebuf);
- j->img_comp[i].linebuf = NULL;
- }
- }
+ stbi__free_jpeg_components(j, j->s->img_n, 0);
}
typedef struct
@@ -3289,9 +3855,16 @@ typedef struct
int ypos; // which pre-expansion row we're on
} stbi__resample;
+// fast 0..255 * 0..255 => 0..255 rounded multiplication
+static stbi_uc stbi__blinn_8x8(stbi_uc x, stbi_uc y)
+{
+ unsigned int t = x*y + 128;
+ return (stbi_uc) ((t + (t >>8)) >> 8);
+}
+
static stbi_uc *load_jpeg_image(stbi__jpeg *z, int *out_x, int *out_y, int *comp, int req_comp)
{
- int n, decode_n;
+ int n, decode_n, is_rgb;
z->s->img_n = 0; // make stbi__cleanup_jpeg safe
// validate req_comp
@@ -3301,19 +3874,25 @@ static stbi_uc *load_jpeg_image(stbi__jpeg *z, int *out_x, int *out_y, int *comp
if (!stbi__decode_jpeg_image(z)) { stbi__cleanup_jpeg(z); return NULL; }
// determine actual number of components to generate
- n = req_comp ? req_comp : z->s->img_n;
+ n = req_comp ? req_comp : z->s->img_n >= 3 ? 3 : 1;
+
+ is_rgb = z->s->img_n == 3 && (z->rgb == 3 || (z->app14_color_transform == 0 && !z->jfif));
- if (z->s->img_n == 3 && n < 3)
+ if (z->s->img_n == 3 && n < 3 && !is_rgb)
decode_n = 1;
else
decode_n = z->s->img_n;
+ // nothing to do if no components requested; check this now to avoid
+ // accessing uninitialized coutput[0] later
+ if (decode_n <= 0) { stbi__cleanup_jpeg(z); return NULL; }
+
// resample and color-convert
{
int k;
unsigned int i,j;
stbi_uc *output;
- stbi_uc *coutput[4];
+ stbi_uc *coutput[4] = { NULL, NULL, NULL, NULL };
stbi__resample res_comp[4];
@@ -3340,7 +3919,7 @@ static stbi_uc *load_jpeg_image(stbi__jpeg *z, int *out_x, int *out_y, int *comp
}
// can't error after this so, this is safe
- output = (stbi_uc *) stbi__malloc(n * z->s->img_x * z->s->img_y + 1);
+ output = (stbi_uc *) stbi__malloc_mad3(n, z->s->img_x, z->s->img_y, 1);
if (!output) { stbi__cleanup_jpeg(z); return stbi__errpuc("outofmem", "Out of memory"); }
// now go ahead and resample
@@ -3363,7 +3942,39 @@ static stbi_uc *load_jpeg_image(stbi__jpeg *z, int *out_x, int *out_y, int *comp
if (n >= 3) {
stbi_uc *y = coutput[0];
if (z->s->img_n == 3) {
- z->YCbCr_to_RGB_kernel(out, y, coutput[1], coutput[2], z->s->img_x, n);
+ if (is_rgb) {
+ for (i=0; i < z->s->img_x; ++i) {
+ out[0] = y[i];
+ out[1] = coutput[1][i];
+ out[2] = coutput[2][i];
+ out[3] = 255;
+ out += n;
+ }
+ } else {
+ z->YCbCr_to_RGB_kernel(out, y, coutput[1], coutput[2], z->s->img_x, n);
+ }
+ } else if (z->s->img_n == 4) {
+ if (z->app14_color_transform == 0) { // CMYK
+ for (i=0; i < z->s->img_x; ++i) {
+ stbi_uc m = coutput[3][i];
+ out[0] = stbi__blinn_8x8(coutput[0][i], m);
+ out[1] = stbi__blinn_8x8(coutput[1][i], m);
+ out[2] = stbi__blinn_8x8(coutput[2][i], m);
+ out[3] = 255;
+ out += n;
+ }
+ } else if (z->app14_color_transform == 2) { // YCCK
+ z->YCbCr_to_RGB_kernel(out, y, coutput[1], coutput[2], z->s->img_x, n);
+ for (i=0; i < z->s->img_x; ++i) {
+ stbi_uc m = coutput[3][i];
+ out[0] = stbi__blinn_8x8(255 - out[0], m);
+ out[1] = stbi__blinn_8x8(255 - out[1], m);
+ out[2] = stbi__blinn_8x8(255 - out[2], m);
+ out += n;
+ }
+ } else { // YCbCr + alpha? Ignore the fourth channel for now
+ z->YCbCr_to_RGB_kernel(out, y, coutput[1], coutput[2], z->s->img_x, n);
+ }
} else
for (i=0; i < z->s->img_x; ++i) {
out[0] = out[1] = out[2] = y[i];
@@ -3371,37 +3982,74 @@ static stbi_uc *load_jpeg_image(stbi__jpeg *z, int *out_x, int *out_y, int *comp
out += n;
}
} else {
- stbi_uc *y = coutput[0];
- if (n == 1)
- for (i=0; i < z->s->img_x; ++i) out[i] = y[i];
- else
- for (i=0; i < z->s->img_x; ++i) *out++ = y[i], *out++ = 255;
+ if (is_rgb) {
+ if (n == 1)
+ for (i=0; i < z->s->img_x; ++i)
+ *out++ = stbi__compute_y(coutput[0][i], coutput[1][i], coutput[2][i]);
+ else {
+ for (i=0; i < z->s->img_x; ++i, out += 2) {
+ out[0] = stbi__compute_y(coutput[0][i], coutput[1][i], coutput[2][i]);
+ out[1] = 255;
+ }
+ }
+ } else if (z->s->img_n == 4 && z->app14_color_transform == 0) {
+ for (i=0; i < z->s->img_x; ++i) {
+ stbi_uc m = coutput[3][i];
+ stbi_uc r = stbi__blinn_8x8(coutput[0][i], m);
+ stbi_uc g = stbi__blinn_8x8(coutput[1][i], m);
+ stbi_uc b = stbi__blinn_8x8(coutput[2][i], m);
+ out[0] = stbi__compute_y(r, g, b);
+ out[1] = 255;
+ out += n;
+ }
+ } else if (z->s->img_n == 4 && z->app14_color_transform == 2) {
+ for (i=0; i < z->s->img_x; ++i) {
+ out[0] = stbi__blinn_8x8(255 - coutput[0][i], coutput[3][i]);
+ out[1] = 255;
+ out += n;
+ }
+ } else {
+ stbi_uc *y = coutput[0];
+ if (n == 1)
+ for (i=0; i < z->s->img_x; ++i) out[i] = y[i];
+ else
+ for (i=0; i < z->s->img_x; ++i) { *out++ = y[i]; *out++ = 255; }
+ }
}
}
stbi__cleanup_jpeg(z);
*out_x = z->s->img_x;
*out_y = z->s->img_y;
- if (comp) *comp = z->s->img_n; // report original components, not output
+ if (comp) *comp = z->s->img_n >= 3 ? 3 : 1; // report original components, not output
return output;
}
}
-static unsigned char *stbi__jpeg_load(stbi__context *s, int *x, int *y, int *comp, int req_comp)
+static void *stbi__jpeg_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri)
{
- stbi__jpeg j;
- j.s = s;
- stbi__setup_jpeg(&j);
- return load_jpeg_image(&j, x,y,comp,req_comp);
+ unsigned char* result;
+ stbi__jpeg* j = (stbi__jpeg*) stbi__malloc(sizeof(stbi__jpeg));
+ if (!j) return stbi__errpuc("outofmem", "Out of memory");
+ memset(j, 0, sizeof(stbi__jpeg));
+ STBI_NOTUSED(ri);
+ j->s = s;
+ stbi__setup_jpeg(j);
+ result = load_jpeg_image(j, x,y,comp,req_comp);
+ STBI_FREE(j);
+ return result;
}
static int stbi__jpeg_test(stbi__context *s)
{
int r;
- stbi__jpeg j;
- j.s = s;
- stbi__setup_jpeg(&j);
- r = stbi__decode_jpeg_header(&j, STBI__SCAN_type);
+ stbi__jpeg* j = (stbi__jpeg*)stbi__malloc(sizeof(stbi__jpeg));
+ if (!j) return stbi__err("outofmem", "Out of memory");
+ memset(j, 0, sizeof(stbi__jpeg));
+ j->s = s;
+ stbi__setup_jpeg(j);
+ r = stbi__decode_jpeg_header(j, STBI__SCAN_type);
stbi__rewind(s);
+ STBI_FREE(j);
return r;
}
@@ -3413,15 +4061,20 @@ static int stbi__jpeg_info_raw(stbi__jpeg *j, int *x, int *y, int *comp)
}
if (x) *x = j->s->img_x;
if (y) *y = j->s->img_y;
- if (comp) *comp = j->s->img_n;
+ if (comp) *comp = j->s->img_n >= 3 ? 3 : 1;
return 1;
}
static int stbi__jpeg_info(stbi__context *s, int *x, int *y, int *comp)
{
- stbi__jpeg j;
- j.s = s;
- return stbi__jpeg_info_raw(&j, x, y, comp);
+ int result;
+ stbi__jpeg* j = (stbi__jpeg*) (stbi__malloc(sizeof(stbi__jpeg)));
+ if (!j) return stbi__err("outofmem", "Out of memory");
+ memset(j, 0, sizeof(stbi__jpeg));
+ j->s = s;
+ result = stbi__jpeg_info_raw(j, x, y, comp);
+ STBI_FREE(j);
+ return result;
}
#endif
@@ -3437,6 +4090,7 @@ static int stbi__jpeg_info(stbi__context *s, int *x, int *y, int *comp)
// fast-way is faster to check than jpeg huffman, but slow way is slower
#define STBI__ZFAST_BITS 9 // accelerate all cases in default tables
#define STBI__ZFAST_MASK ((1 << STBI__ZFAST_BITS) - 1)
+#define STBI__ZNSYMS 288 // number of symbols in literal/length alphabet
// zlib-style huffman encoding
// (jpegs packs from left, zlib from right, so can't share code)
@@ -3446,8 +4100,8 @@ typedef struct
stbi__uint16 firstcode[16];
int maxcode[17];
stbi__uint16 firstsymbol[16];
- stbi_uc size[288];
- stbi__uint16 value[288];
+ stbi_uc size[STBI__ZNSYMS];
+ stbi__uint16 value[STBI__ZNSYMS];
} stbi__zhuffman;
stbi_inline static int stbi__bitreverse16(int n)
@@ -3467,7 +4121,7 @@ stbi_inline static int stbi__bit_reverse(int v, int bits)
return stbi__bitreverse16(v) >> (16-bits);
}
-static int stbi__zbuild_huffman(stbi__zhuffman *z, stbi_uc *sizelist, int num)
+static int stbi__zbuild_huffman(stbi__zhuffman *z, const stbi_uc *sizelist, int num)
{
int i,k=0;
int code, next_code[16], sizes[17];
@@ -3502,10 +4156,10 @@ static int stbi__zbuild_huffman(stbi__zhuffman *z, stbi_uc *sizelist, int num)
z->size [c] = (stbi_uc ) s;
z->value[c] = (stbi__uint16) i;
if (s <= STBI__ZFAST_BITS) {
- int k = stbi__bit_reverse(next_code[s],s);
- while (k < (1 << STBI__ZFAST_BITS)) {
- z->fast[k] = fastv;
- k += (1 << s);
+ int j = stbi__bit_reverse(next_code[s],s);
+ while (j < (1 << STBI__ZFAST_BITS)) {
+ z->fast[j] = fastv;
+ j += (1 << s);
}
}
++next_code[s];
@@ -3524,6 +4178,7 @@ typedef struct
{
stbi_uc *zbuffer, *zbuffer_end;
int num_bits;
+ int hit_zeof_once;
stbi__uint32 code_buffer;
char *zout;
@@ -3534,16 +4189,23 @@ typedef struct
stbi__zhuffman z_length, z_distance;
} stbi__zbuf;
+stbi_inline static int stbi__zeof(stbi__zbuf *z)
+{
+ return (z->zbuffer >= z->zbuffer_end);
+}
+
stbi_inline static stbi_uc stbi__zget8(stbi__zbuf *z)
{
- if (z->zbuffer >= z->zbuffer_end) return 0;
- return *z->zbuffer++;
+ return stbi__zeof(z) ? 0 : *z->zbuffer++;
}
static void stbi__fill_bits(stbi__zbuf *z)
{
do {
- STBI_ASSERT(z->code_buffer < (1U << z->num_bits));
+ if (z->code_buffer >= (1U << z->num_bits)) {
+ z->zbuffer = z->zbuffer_end; /* treat this as EOF so we fail. */
+ return;
+ }
z->code_buffer |= (unsigned int) stbi__zget8(z) << z->num_bits;
z->num_bits += 8;
} while (z->num_bits <= 24);
@@ -3568,10 +4230,11 @@ static int stbi__zhuffman_decode_slowpath(stbi__zbuf *a, stbi__zhuffman *z)
for (s=STBI__ZFAST_BITS+1; ; ++s)
if (k < z->maxcode[s])
break;
- if (s == 16) return -1; // invalid code!
+ if (s >= 16) return -1; // invalid code!
// code size is s, so:
b = (k >> (16-s)) - z->firstcode[s] + z->firstsymbol[s];
- STBI_ASSERT(z->size[b] == s);
+ if (b >= STBI__ZNSYMS) return -1; // some data was corrupt somewhere!
+ if (z->size[b] != s) return -1; // was originally an assert, but report failure instead.
a->code_buffer >>= s;
a->num_bits -= s;
return z->value[b];
@@ -3580,7 +4243,23 @@ static int stbi__zhuffman_decode_slowpath(stbi__zbuf *a, stbi__zhuffman *z)
stbi_inline static int stbi__zhuffman_decode(stbi__zbuf *a, stbi__zhuffman *z)
{
int b,s;
- if (a->num_bits < 16) stbi__fill_bits(a);
+ if (a->num_bits < 16) {
+ if (stbi__zeof(a)) {
+ if (!a->hit_zeof_once) {
+ // This is the first time we hit eof, insert 16 extra padding btis
+ // to allow us to keep going; if we actually consume any of them
+ // though, that is invalid data. This is caught later.
+ a->hit_zeof_once = 1;
+ a->num_bits += 16; // add 16 implicit zero bits
+ } else {
+ // We already inserted our extra 16 padding bits and are again
+ // out, this stream is actually prematurely terminated.
+ return -1;
+ }
+ } else {
+ stbi__fill_bits(a);
+ }
+ }
b = z->fast[a->code_buffer & STBI__ZFAST_MASK];
if (b) {
s = b >> 9;
@@ -3594,14 +4273,18 @@ stbi_inline static int stbi__zhuffman_decode(stbi__zbuf *a, stbi__zhuffman *z)
static int stbi__zexpand(stbi__zbuf *z, char *zout, int n) // need to make room for n bytes
{
char *q;
- int cur, limit;
+ unsigned int cur, limit, old_limit;
z->zout = zout;
if (!z->z_expandable) return stbi__err("output buffer limit","Corrupt PNG");
- cur = (int) (z->zout - z->zout_start);
- limit = (int) (z->zout_end - z->zout_start);
- while (cur + n > limit)
+ cur = (unsigned int) (z->zout - z->zout_start);
+ limit = old_limit = (unsigned) (z->zout_end - z->zout_start);
+ if (UINT_MAX - cur < (unsigned) n) return stbi__err("outofmem", "Out of memory");
+ while (cur + n > limit) {
+ if(limit > UINT_MAX / 2) return stbi__err("outofmem", "Out of memory");
limit *= 2;
- q = (char *) STBI_REALLOC(z->zout_start, limit);
+ }
+ q = (char *) STBI_REALLOC_SIZED(z->zout_start, old_limit, limit);
+ STBI_NOTUSED(old_limit);
if (q == NULL) return stbi__err("outofmem", "Out of memory");
z->zout_start = q;
z->zout = q + cur;
@@ -3609,18 +4292,18 @@ static int stbi__zexpand(stbi__zbuf *z, char *zout, int n) // need to make room
return 1;
}
-static int stbi__zlength_base[31] = {
+static const int stbi__zlength_base[31] = {
3,4,5,6,7,8,9,10,11,13,
15,17,19,23,27,31,35,43,51,59,
67,83,99,115,131,163,195,227,258,0,0 };
-static int stbi__zlength_extra[31]=
+static const int stbi__zlength_extra[31]=
{ 0,0,0,0,0,0,0,0,1,1,1,1,2,2,2,2,3,3,3,3,4,4,4,4,5,5,5,5,0,0,0 };
-static int stbi__zdist_base[32] = { 1,2,3,4,5,7,9,13,17,25,33,49,65,97,129,193,
+static const int stbi__zdist_base[32] = { 1,2,3,4,5,7,9,13,17,25,33,49,65,97,129,193,
257,385,513,769,1025,1537,2049,3073,4097,6145,8193,12289,16385,24577,0,0};
-static int stbi__zdist_extra[32] =
+static const int stbi__zdist_extra[32] =
{ 0,0,0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,10,10,11,11,12,12,13,13};
static int stbi__parse_huffman_block(stbi__zbuf *a)
@@ -3640,17 +4323,25 @@ static int stbi__parse_huffman_block(stbi__zbuf *a)
int len,dist;
if (z == 256) {
a->zout = zout;
+ if (a->hit_zeof_once && a->num_bits < 16) {
+ // The first time we hit zeof, we inserted 16 extra zero bits into our bit
+ // buffer so the decoder can just do its speculative decoding. But if we
+ // actually consumed any of those bits (which is the case when num_bits < 16),
+ // the stream actually read past the end so it is malformed.
+ return stbi__err("unexpected end","Corrupt PNG");
+ }
return 1;
}
+ if (z >= 286) return stbi__err("bad huffman code","Corrupt PNG"); // per DEFLATE, length codes 286 and 287 must not appear in compressed data
z -= 257;
len = stbi__zlength_base[z];
if (stbi__zlength_extra[z]) len += stbi__zreceive(a, stbi__zlength_extra[z]);
z = stbi__zhuffman_decode(a, &a->z_distance);
- if (z < 0) return stbi__err("bad huffman code","Corrupt PNG");
+ if (z < 0 || z >= 30) return stbi__err("bad huffman code","Corrupt PNG"); // per DEFLATE, distance codes 30 and 31 must not appear in compressed data
dist = stbi__zdist_base[z];
if (stbi__zdist_extra[z]) dist += stbi__zreceive(a, stbi__zdist_extra[z]);
if (zout - a->zout_start < dist) return stbi__err("bad dist","Corrupt PNG");
- if (zout + len > a->zout_end) {
+ if (len > a->zout_end - zout) {
if (!stbi__zexpand(a, zout, len)) return 0;
zout = a->zout;
}
@@ -3667,7 +4358,7 @@ static int stbi__parse_huffman_block(stbi__zbuf *a)
static int stbi__compute_huffman_codes(stbi__zbuf *a)
{
- static stbi_uc length_dezigzag[19] = { 16,17,18,0,8,7,9,6,10,5,11,4,12,3,13,2,14,1,15 };
+ static const stbi_uc length_dezigzag[19] = { 16,17,18,0,8,7,9,6,10,5,11,4,12,3,13,2,14,1,15 };
stbi__zhuffman z_codelength;
stbi_uc lencodes[286+32+137];//padding for maximum single op
stbi_uc codelength_sizes[19];
@@ -3676,6 +4367,7 @@ static int stbi__compute_huffman_codes(stbi__zbuf *a)
int hlit = stbi__zreceive(a,5) + 257;
int hdist = stbi__zreceive(a,5) + 1;
int hclen = stbi__zreceive(a,4) + 4;
+ int ntot = hlit + hdist;
memset(codelength_sizes, 0, sizeof(codelength_sizes));
for (i=0; i < hclen; ++i) {
@@ -3685,33 +4377,36 @@ static int stbi__compute_huffman_codes(stbi__zbuf *a)
if (!stbi__zbuild_huffman(&z_codelength, codelength_sizes, 19)) return 0;
n = 0;
- while (n < hlit + hdist) {
+ while (n < ntot) {
int c = stbi__zhuffman_decode(a, &z_codelength);
if (c < 0 || c >= 19) return stbi__err("bad codelengths", "Corrupt PNG");
if (c < 16)
lencodes[n++] = (stbi_uc) c;
- else if (c == 16) {
- c = stbi__zreceive(a,2)+3;
- memset(lencodes+n, lencodes[n-1], c);
- n += c;
- } else if (c == 17) {
- c = stbi__zreceive(a,3)+3;
- memset(lencodes+n, 0, c);
- n += c;
- } else {
- STBI_ASSERT(c == 18);
- c = stbi__zreceive(a,7)+11;
- memset(lencodes+n, 0, c);
+ else {
+ stbi_uc fill = 0;
+ if (c == 16) {
+ c = stbi__zreceive(a,2)+3;
+ if (n == 0) return stbi__err("bad codelengths", "Corrupt PNG");
+ fill = lencodes[n-1];
+ } else if (c == 17) {
+ c = stbi__zreceive(a,3)+3;
+ } else if (c == 18) {
+ c = stbi__zreceive(a,7)+11;
+ } else {
+ return stbi__err("bad codelengths", "Corrupt PNG");
+ }
+ if (ntot - n < c) return stbi__err("bad codelengths", "Corrupt PNG");
+ memset(lencodes+n, fill, c);
n += c;
}
}
- if (n != hlit+hdist) return stbi__err("bad codelengths","Corrupt PNG");
+ if (n != ntot) return stbi__err("bad codelengths","Corrupt PNG");
if (!stbi__zbuild_huffman(&a->z_length, lencodes, hlit)) return 0;
if (!stbi__zbuild_huffman(&a->z_distance, lencodes+hlit, hdist)) return 0;
return 1;
}
-static int stbi__parse_uncomperssed_block(stbi__zbuf *a)
+static int stbi__parse_uncompressed_block(stbi__zbuf *a)
{
stbi_uc header[4];
int len,nlen,k;
@@ -3724,7 +4419,7 @@ static int stbi__parse_uncomperssed_block(stbi__zbuf *a)
a->code_buffer >>= 8;
a->num_bits -= 8;
}
- STBI_ASSERT(a->num_bits == 0);
+ if (a->num_bits < 0) return stbi__err("zlib corrupt","Corrupt PNG");
// now fill header the normal way
while (k < 4)
header[k++] = stbi__zget8(a);
@@ -3746,6 +4441,7 @@ static int stbi__parse_zlib_header(stbi__zbuf *a)
int cm = cmf & 15;
/* int cinfo = cmf >> 4; */
int flg = stbi__zget8(a);
+ if (stbi__zeof(a)) return stbi__err("bad zlib header","Corrupt PNG"); // zlib spec
if ((cmf*256+flg) % 31 != 0) return stbi__err("bad zlib header","Corrupt PNG"); // zlib spec
if (flg & 32) return stbi__err("no preset dict","Corrupt PNG"); // preset dictionary not allowed in png
if (cm != 8) return stbi__err("bad compression","Corrupt PNG"); // DEFLATE required for png
@@ -3753,9 +4449,24 @@ static int stbi__parse_zlib_header(stbi__zbuf *a)
return 1;
}
-// @TODO: should statically initialize these for optimal thread safety
-static stbi_uc stbi__zdefault_length[288], stbi__zdefault_distance[32];
-static void stbi__init_zdefaults(void)
+static const stbi_uc stbi__zdefault_length[STBI__ZNSYMS] =
+{
+ 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8, 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,
+ 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8, 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,
+ 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8, 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,
+ 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8, 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,
+ 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8, 9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,
+ 9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9, 9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,
+ 9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9, 9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,
+ 9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9, 9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,9,
+ 7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7, 7,7,7,7,7,7,7,7,8,8,8,8,8,8,8,8
+};
+static const stbi_uc stbi__zdefault_distance[32] =
+{
+ 5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5
+};
+/*
+Init algorithm:
{
int i; // use <= to match clearly with spec
for (i=0; i <= 143; ++i) stbi__zdefault_length[i] = 8;
@@ -3765,6 +4476,7 @@ static void stbi__init_zdefaults(void)
for (i=0; i <= 31; ++i) stbi__zdefault_distance[i] = 5;
}
+*/
static int stbi__parse_zlib(stbi__zbuf *a, int parse_header)
{
@@ -3773,18 +4485,18 @@ static int stbi__parse_zlib(stbi__zbuf *a, int parse_header)
if (!stbi__parse_zlib_header(a)) return 0;
a->num_bits = 0;
a->code_buffer = 0;
+ a->hit_zeof_once = 0;
do {
final = stbi__zreceive(a,1);
type = stbi__zreceive(a,2);
if (type == 0) {
- if (!stbi__parse_uncomperssed_block(a)) return 0;
+ if (!stbi__parse_uncompressed_block(a)) return 0;
} else if (type == 3) {
return 0;
} else {
if (type == 1) {
// use fixed code lengths
- if (!stbi__zdefault_distance[31]) stbi__init_zdefaults();
- if (!stbi__zbuild_huffman(&a->z_length , stbi__zdefault_length , 288)) return 0;
+ if (!stbi__zbuild_huffman(&a->z_length , stbi__zdefault_length , STBI__ZNSYMS)) return 0;
if (!stbi__zbuild_huffman(&a->z_distance, stbi__zdefault_distance, 32)) return 0;
} else {
if (!stbi__compute_huffman_codes(a)) return 0;
@@ -3908,7 +4620,7 @@ static stbi__pngchunk stbi__get_chunk_header(stbi__context *s)
static int stbi__check_png_header(stbi__context *s)
{
- static stbi_uc png_sig[8] = { 137,80,78,71,13,10,26,10 };
+ static const stbi_uc png_sig[8] = { 137,80,78,71,13,10,26,10 };
int i;
for (i=0; i < 8; ++i)
if (stbi__get8(s) != png_sig[i]) return stbi__err("bad png sig","Not a PNG");
@@ -3919,6 +4631,7 @@ typedef struct
{
stbi__context *s;
stbi_uc *idata, *expanded, *out;
+ int depth;
} stbi__png;
@@ -3928,9 +4641,8 @@ enum {
STBI__F_up=2,
STBI__F_avg=3,
STBI__F_paeth=4,
- // synthetic filters used for first scanline to avoid needing a dummy row of 0s
- STBI__F_avg_first,
- STBI__F_paeth_first
+ // synthetic filter used for first scanline to avoid needing a dummy row of 0s
+ STBI__F_avg_first
};
static stbi_uc first_row_filter[5] =
@@ -3939,210 +4651,225 @@ static stbi_uc first_row_filter[5] =
STBI__F_sub,
STBI__F_none,
STBI__F_avg_first,
- STBI__F_paeth_first
+ STBI__F_sub // Paeth with b=c=0 turns out to be equivalent to sub
};
static int stbi__paeth(int a, int b, int c)
{
- int p = a + b - c;
- int pa = abs(p-a);
- int pb = abs(p-b);
- int pc = abs(p-c);
- if (pa <= pb && pa <= pc) return a;
- if (pb <= pc) return b;
- return c;
+ // This formulation looks very different from the reference in the PNG spec, but is
+ // actually equivalent and has favorable data dependencies and admits straightforward
+ // generation of branch-free code, which helps performance significantly.
+ int thresh = c*3 - (a + b);
+ int lo = a < b ? a : b;
+ int hi = a < b ? b : a;
+ int t0 = (hi <= thresh) ? lo : c;
+ int t1 = (thresh <= lo) ? hi : t0;
+ return t1;
}
-static stbi_uc stbi__depth_scale_table[9] = { 0, 0xff, 0x55, 0, 0x11, 0,0,0, 0x01 };
+static const stbi_uc stbi__depth_scale_table[9] = { 0, 0xff, 0x55, 0, 0x11, 0,0,0, 0x01 };
+
+// adds an extra all-255 alpha channel
+// dest == src is legal
+// img_n must be 1 or 3
+static void stbi__create_png_alpha_expand8(stbi_uc *dest, stbi_uc *src, stbi__uint32 x, int img_n)
+{
+ int i;
+ // must process data backwards since we allow dest==src
+ if (img_n == 1) {
+ for (i=x-1; i >= 0; --i) {
+ dest[i*2+1] = 255;
+ dest[i*2+0] = src[i];
+ }
+ } else {
+ STBI_ASSERT(img_n == 3);
+ for (i=x-1; i >= 0; --i) {
+ dest[i*4+3] = 255;
+ dest[i*4+2] = src[i*3+2];
+ dest[i*4+1] = src[i*3+1];
+ dest[i*4+0] = src[i*3+0];
+ }
+ }
+}
// create the png data from post-deflated data
static int stbi__create_png_image_raw(stbi__png *a, stbi_uc *raw, stbi__uint32 raw_len, int out_n, stbi__uint32 x, stbi__uint32 y, int depth, int color)
{
+ int bytes = (depth == 16 ? 2 : 1);
stbi__context *s = a->s;
- stbi__uint32 i,j,stride = x*out_n;
+ stbi__uint32 i,j,stride = x*out_n*bytes;
stbi__uint32 img_len, img_width_bytes;
+ stbi_uc *filter_buf;
+ int all_ok = 1;
int k;
int img_n = s->img_n; // copy it into a local for later
+ int output_bytes = out_n*bytes;
+ int filter_bytes = img_n*bytes;
+ int width = x;
+
STBI_ASSERT(out_n == s->img_n || out_n == s->img_n+1);
- a->out = (stbi_uc *) stbi__malloc(x * y * out_n); // extra bytes to write off the end into
+ a->out = (stbi_uc *) stbi__malloc_mad3(x, y, output_bytes, 0); // extra bytes to write off the end into
if (!a->out) return stbi__err("outofmem", "Out of memory");
+ // note: error exits here don't need to clean up a->out individually,
+ // stbi__do_png always does on error.
+ if (!stbi__mad3sizes_valid(img_n, x, depth, 7)) return stbi__err("too large", "Corrupt PNG");
img_width_bytes = (((img_n * x * depth) + 7) >> 3);
+ if (!stbi__mad2sizes_valid(img_width_bytes, y, img_width_bytes)) return stbi__err("too large", "Corrupt PNG");
img_len = (img_width_bytes + 1) * y;
- if (s->img_x == x && s->img_y == y) {
- if (raw_len != img_len) return stbi__err("not enough pixels","Corrupt PNG");
- } else { // interlaced:
- if (raw_len < img_len) return stbi__err("not enough pixels","Corrupt PNG");
+
+ // we used to check for exact match between raw_len and img_len on non-interlaced PNGs,
+ // but issue #276 reported a PNG in the wild that had extra data at the end (all zeros),
+ // so just check for raw_len < img_len always.
+ if (raw_len < img_len) return stbi__err("not enough pixels","Corrupt PNG");
+
+ // Allocate two scan lines worth of filter workspace buffer.
+ filter_buf = (stbi_uc *) stbi__malloc_mad2(img_width_bytes, 2, 0);
+ if (!filter_buf) return stbi__err("outofmem", "Out of memory");
+
+ // Filtering for low-bit-depth images
+ if (depth < 8) {
+ filter_bytes = 1;
+ width = img_width_bytes;
}
for (j=0; j < y; ++j) {
- stbi_uc *cur = a->out + stride*j;
- stbi_uc *prior = cur - stride;
+ // cur/prior filter buffers alternate
+ stbi_uc *cur = filter_buf + (j & 1)*img_width_bytes;
+ stbi_uc *prior = filter_buf + (~j & 1)*img_width_bytes;
+ stbi_uc *dest = a->out + stride*j;
+ int nk = width * filter_bytes;
int filter = *raw++;
- int filter_bytes = img_n;
- int width = x;
- if (filter > 4)
- return stbi__err("invalid filter","Corrupt PNG");
- if (depth < 8) {
- STBI_ASSERT(img_width_bytes <= x);
- cur += x*out_n - img_width_bytes; // store output to the rightmost img_len bytes, so we can decode in place
- filter_bytes = 1;
- width = img_width_bytes;
+ // check filter type
+ if (filter > 4) {
+ all_ok = stbi__err("invalid filter","Corrupt PNG");
+ break;
}
// if first row, use special filter that doesn't sample previous row
if (j == 0) filter = first_row_filter[filter];
- // handle first byte explicitly
- for (k=0; k < filter_bytes; ++k) {
- switch (filter) {
- case STBI__F_none : cur[k] = raw[k]; break;
- case STBI__F_sub : cur[k] = raw[k]; break;
- case STBI__F_up : cur[k] = STBI__BYTECAST(raw[k] + prior[k]); break;
- case STBI__F_avg : cur[k] = STBI__BYTECAST(raw[k] + (prior[k]>>1)); break;
- case STBI__F_paeth : cur[k] = STBI__BYTECAST(raw[k] + stbi__paeth(0,prior[k],0)); break;
- case STBI__F_avg_first : cur[k] = raw[k]; break;
- case STBI__F_paeth_first: cur[k] = raw[k]; break;
- }
- }
-
- if (depth == 8) {
- if (img_n != out_n)
- cur[img_n] = 255; // first pixel
- raw += img_n;
- cur += out_n;
- prior += out_n;
- } else {
- raw += 1;
- cur += 1;
- prior += 1;
+ // perform actual filtering
+ switch (filter) {
+ case STBI__F_none:
+ memcpy(cur, raw, nk);
+ break;
+ case STBI__F_sub:
+ memcpy(cur, raw, filter_bytes);
+ for (k = filter_bytes; k < nk; ++k)
+ cur[k] = STBI__BYTECAST(raw[k] + cur[k-filter_bytes]);
+ break;
+ case STBI__F_up:
+ for (k = 0; k < nk; ++k)
+ cur[k] = STBI__BYTECAST(raw[k] + prior[k]);
+ break;
+ case STBI__F_avg:
+ for (k = 0; k < filter_bytes; ++k)
+ cur[k] = STBI__BYTECAST(raw[k] + (prior[k]>>1));
+ for (k = filter_bytes; k < nk; ++k)
+ cur[k] = STBI__BYTECAST(raw[k] + ((prior[k] + cur[k-filter_bytes])>>1));
+ break;
+ case STBI__F_paeth:
+ for (k = 0; k < filter_bytes; ++k)
+ cur[k] = STBI__BYTECAST(raw[k] + prior[k]); // prior[k] == stbi__paeth(0,prior[k],0)
+ for (k = filter_bytes; k < nk; ++k)
+ cur[k] = STBI__BYTECAST(raw[k] + stbi__paeth(cur[k-filter_bytes], prior[k], prior[k-filter_bytes]));
+ break;
+ case STBI__F_avg_first:
+ memcpy(cur, raw, filter_bytes);
+ for (k = filter_bytes; k < nk; ++k)
+ cur[k] = STBI__BYTECAST(raw[k] + (cur[k-filter_bytes] >> 1));
+ break;
}
- // this is a little gross, so that we don't switch per-pixel or per-component
- if (depth < 8 || img_n == out_n) {
- int nk = (width - 1)*img_n;
- #define CASE(f) \
- case f: \
- for (k=0; k < nk; ++k)
- switch (filter) {
- // "none" filter turns into a memcpy here; make that explicit.
- case STBI__F_none: memcpy(cur, raw, nk); break;
- CASE(STBI__F_sub) cur[k] = STBI__BYTECAST(raw[k] + cur[k-filter_bytes]); break;
- CASE(STBI__F_up) cur[k] = STBI__BYTECAST(raw[k] + prior[k]); break;
- CASE(STBI__F_avg) cur[k] = STBI__BYTECAST(raw[k] + ((prior[k] + cur[k-filter_bytes])>>1)); break;
- CASE(STBI__F_paeth) cur[k] = STBI__BYTECAST(raw[k] + stbi__paeth(cur[k-filter_bytes],prior[k],prior[k-filter_bytes])); break;
- CASE(STBI__F_avg_first) cur[k] = STBI__BYTECAST(raw[k] + (cur[k-filter_bytes] >> 1)); break;
- CASE(STBI__F_paeth_first) cur[k] = STBI__BYTECAST(raw[k] + stbi__paeth(cur[k-filter_bytes],0,0)); break;
- }
- #undef CASE
- raw += nk;
- } else {
- STBI_ASSERT(img_n+1 == out_n);
- #define CASE(f) \
- case f: \
- for (i=x-1; i >= 1; --i, cur[img_n]=255,raw+=img_n,cur+=out_n,prior+=out_n) \
- for (k=0; k < img_n; ++k)
- switch (filter) {
- CASE(STBI__F_none) cur[k] = raw[k]; break;
- CASE(STBI__F_sub) cur[k] = STBI__BYTECAST(raw[k] + cur[k-out_n]); break;
- CASE(STBI__F_up) cur[k] = STBI__BYTECAST(raw[k] + prior[k]); break;
- CASE(STBI__F_avg) cur[k] = STBI__BYTECAST(raw[k] + ((prior[k] + cur[k-out_n])>>1)); break;
- CASE(STBI__F_paeth) cur[k] = STBI__BYTECAST(raw[k] + stbi__paeth(cur[k-out_n],prior[k],prior[k-out_n])); break;
- CASE(STBI__F_avg_first) cur[k] = STBI__BYTECAST(raw[k] + (cur[k-out_n] >> 1)); break;
- CASE(STBI__F_paeth_first) cur[k] = STBI__BYTECAST(raw[k] + stbi__paeth(cur[k-out_n],0,0)); break;
- }
- #undef CASE
- }
- }
+ raw += nk;
- // we make a separate pass to expand bits to pixels; for performance,
- // this could run two scanlines behind the above code, so it won't
- // intefere with filtering but will still be in the cache.
- if (depth < 8) {
- for (j=0; j < y; ++j) {
- stbi_uc *cur = a->out + stride*j;
- stbi_uc *in = a->out + stride*j + x*out_n - img_width_bytes;
- // unpack 1/2/4-bit into a 8-bit buffer. allows us to keep the common 8-bit path optimal at minimal cost for 1/2/4-bit
- // png guarante byte alignment, if width is not multiple of 8/4/2 we'll decode dummy trailing data that will be skipped in the later loop
+ // expand decoded bits in cur to dest, also adding an extra alpha channel if desired
+ if (depth < 8) {
stbi_uc scale = (color == 0) ? stbi__depth_scale_table[depth] : 1; // scale grayscale values to 0..255 range
+ stbi_uc *in = cur;
+ stbi_uc *out = dest;
+ stbi_uc inb = 0;
+ stbi__uint32 nsmp = x*img_n;
- // note that the final byte might overshoot and write more data than desired.
- // we can allocate enough data that this never writes out of memory, but it
- // could also overwrite the next scanline. can it overwrite non-empty data
- // on the next scanline? yes, consider 1-pixel-wide scanlines with 1-bit-per-pixel.
- // so we need to explicitly clamp the final ones
-
+ // expand bits to bytes first
if (depth == 4) {
- for (k=x*img_n; k >= 2; k-=2, ++in) {
- *cur++ = scale * ((*in >> 4) );
- *cur++ = scale * ((*in ) & 0x0f);
+ for (i=0; i < nsmp; ++i) {
+ if ((i & 1) == 0) inb = *in++;
+ *out++ = scale * (inb >> 4);
+ inb <<= 4;
}
- if (k > 0) *cur++ = scale * ((*in >> 4) );
} else if (depth == 2) {
- for (k=x*img_n; k >= 4; k-=4, ++in) {
- *cur++ = scale * ((*in >> 6) );
- *cur++ = scale * ((*in >> 4) & 0x03);
- *cur++ = scale * ((*in >> 2) & 0x03);
- *cur++ = scale * ((*in ) & 0x03);
+ for (i=0; i < nsmp; ++i) {
+ if ((i & 3) == 0) inb = *in++;
+ *out++ = scale * (inb >> 6);
+ inb <<= 2;
}
- if (k > 0) *cur++ = scale * ((*in >> 6) );
- if (k > 1) *cur++ = scale * ((*in >> 4) & 0x03);
- if (k > 2) *cur++ = scale * ((*in >> 2) & 0x03);
- } else if (depth == 1) {
- for (k=x*img_n; k >= 8; k-=8, ++in) {
- *cur++ = scale * ((*in >> 7) );
- *cur++ = scale * ((*in >> 6) & 0x01);
- *cur++ = scale * ((*in >> 5) & 0x01);
- *cur++ = scale * ((*in >> 4) & 0x01);
- *cur++ = scale * ((*in >> 3) & 0x01);
- *cur++ = scale * ((*in >> 2) & 0x01);
- *cur++ = scale * ((*in >> 1) & 0x01);
- *cur++ = scale * ((*in ) & 0x01);
+ } else {
+ STBI_ASSERT(depth == 1);
+ for (i=0; i < nsmp; ++i) {
+ if ((i & 7) == 0) inb = *in++;
+ *out++ = scale * (inb >> 7);
+ inb <<= 1;
}
- if (k > 0) *cur++ = scale * ((*in >> 7) );
- if (k > 1) *cur++ = scale * ((*in >> 6) & 0x01);
- if (k > 2) *cur++ = scale * ((*in >> 5) & 0x01);
- if (k > 3) *cur++ = scale * ((*in >> 4) & 0x01);
- if (k > 4) *cur++ = scale * ((*in >> 3) & 0x01);
- if (k > 5) *cur++ = scale * ((*in >> 2) & 0x01);
- if (k > 6) *cur++ = scale * ((*in >> 1) & 0x01);
}
- if (img_n != out_n) {
- // insert alpha = 255
- stbi_uc *cur = a->out + stride*j;
- int i;
+
+ // insert alpha=255 values if desired
+ if (img_n != out_n)
+ stbi__create_png_alpha_expand8(dest, dest, x, img_n);
+ } else if (depth == 8) {
+ if (img_n == out_n)
+ memcpy(dest, cur, x*img_n);
+ else
+ stbi__create_png_alpha_expand8(dest, cur, x, img_n);
+ } else if (depth == 16) {
+ // convert the image data from big-endian to platform-native
+ stbi__uint16 *dest16 = (stbi__uint16*)dest;
+ stbi__uint32 nsmp = x*img_n;
+
+ if (img_n == out_n) {
+ for (i = 0; i < nsmp; ++i, ++dest16, cur += 2)
+ *dest16 = (cur[0] << 8) | cur[1];
+ } else {
+ STBI_ASSERT(img_n+1 == out_n);
if (img_n == 1) {
- for (i=x-1; i >= 0; --i) {
- cur[i*2+1] = 255;
- cur[i*2+0] = cur[i];
+ for (i = 0; i < x; ++i, dest16 += 2, cur += 2) {
+ dest16[0] = (cur[0] << 8) | cur[1];
+ dest16[1] = 0xffff;
}
} else {
STBI_ASSERT(img_n == 3);
- for (i=x-1; i >= 0; --i) {
- cur[i*4+3] = 255;
- cur[i*4+2] = cur[i*3+2];
- cur[i*4+1] = cur[i*3+1];
- cur[i*4+0] = cur[i*3+0];
+ for (i = 0; i < x; ++i, dest16 += 4, cur += 6) {
+ dest16[0] = (cur[0] << 8) | cur[1];
+ dest16[1] = (cur[2] << 8) | cur[3];
+ dest16[2] = (cur[4] << 8) | cur[5];
+ dest16[3] = 0xffff;
}
}
}
}
}
+ STBI_FREE(filter_buf);
+ if (!all_ok) return 0;
+
return 1;
}
static int stbi__create_png_image(stbi__png *a, stbi_uc *image_data, stbi__uint32 image_data_len, int out_n, int depth, int color, int interlaced)
{
+ int bytes = (depth == 16 ? 2 : 1);
+ int out_bytes = out_n * bytes;
stbi_uc *final;
int p;
if (!interlaced)
return stbi__create_png_image_raw(a, image_data, image_data_len, out_n, a->s->img_x, a->s->img_y, depth, color);
// de-interlacing
- final = (stbi_uc *) stbi__malloc(a->s->img_x * a->s->img_y * out_n);
+ final = (stbi_uc *) stbi__malloc_mad3(a->s->img_x, a->s->img_y, out_bytes, 0);
+ if (!final) return stbi__err("outofmem", "Out of memory");
for (p=0; p < 7; ++p) {
int xorig[] = { 0,4,0,2,0,1,0 };
int yorig[] = { 0,0,4,0,2,0,1 };
@@ -4162,8 +4889,8 @@ static int stbi__create_png_image(stbi__png *a, stbi_uc *image_data, stbi__uint3
for (i=0; i < x; ++i) {
int out_y = j*yspc[p]+yorig[p];
int out_x = i*xspc[p]+xorig[p];
- memcpy(final + out_y*a->s->img_x*out_n + out_x*out_n,
- a->out + (j*x+i)*out_n, out_n);
+ memcpy(final + out_y*a->s->img_x*out_bytes + out_x*out_bytes,
+ a->out + (j*x+i)*out_bytes, out_bytes);
}
}
STBI_FREE(a->out);
@@ -4201,12 +4928,37 @@ static int stbi__compute_transparency(stbi__png *z, stbi_uc tc[3], int out_n)
return 1;
}
+static int stbi__compute_transparency16(stbi__png *z, stbi__uint16 tc[3], int out_n)
+{
+ stbi__context *s = z->s;
+ stbi__uint32 i, pixel_count = s->img_x * s->img_y;
+ stbi__uint16 *p = (stbi__uint16*) z->out;
+
+ // compute color-based transparency, assuming we've
+ // already got 65535 as the alpha value in the output
+ STBI_ASSERT(out_n == 2 || out_n == 4);
+
+ if (out_n == 2) {
+ for (i = 0; i < pixel_count; ++i) {
+ p[1] = (p[0] == tc[0] ? 0 : 65535);
+ p += 2;
+ }
+ } else {
+ for (i = 0; i < pixel_count; ++i) {
+ if (p[0] == tc[0] && p[1] == tc[1] && p[2] == tc[2])
+ p[3] = 0;
+ p += 4;
+ }
+ }
+ return 1;
+}
+
static int stbi__expand_png_palette(stbi__png *a, stbi_uc *palette, int len, int pal_img_n)
{
stbi__uint32 i, pixel_count = a->s->img_x * a->s->img_y;
stbi_uc *p, *temp_out, *orig = a->out;
- p = (stbi_uc *) stbi__malloc(pixel_count * pal_img_n);
+ p = (stbi_uc *) stbi__malloc_mad2(pixel_count, pal_img_n, 0);
if (p == NULL) return stbi__err("outofmem", "Out of memory");
// between here and free(out) below, exitting would leak
@@ -4238,19 +4990,46 @@ static int stbi__expand_png_palette(stbi__png *a, stbi_uc *palette, int len, int
return 1;
}
-static int stbi__unpremultiply_on_load = 0;
-static int stbi__de_iphone_flag = 0;
+static int stbi__unpremultiply_on_load_global = 0;
+static int stbi__de_iphone_flag_global = 0;
STBIDEF void stbi_set_unpremultiply_on_load(int flag_true_if_should_unpremultiply)
{
- stbi__unpremultiply_on_load = flag_true_if_should_unpremultiply;
+ stbi__unpremultiply_on_load_global = flag_true_if_should_unpremultiply;
}
STBIDEF void stbi_convert_iphone_png_to_rgb(int flag_true_if_should_convert)
{
- stbi__de_iphone_flag = flag_true_if_should_convert;
+ stbi__de_iphone_flag_global = flag_true_if_should_convert;
+}
+
+#ifndef STBI_THREAD_LOCAL
+#define stbi__unpremultiply_on_load stbi__unpremultiply_on_load_global
+#define stbi__de_iphone_flag stbi__de_iphone_flag_global
+#else
+static STBI_THREAD_LOCAL int stbi__unpremultiply_on_load_local, stbi__unpremultiply_on_load_set;
+static STBI_THREAD_LOCAL int stbi__de_iphone_flag_local, stbi__de_iphone_flag_set;
+
+STBIDEF void stbi_set_unpremultiply_on_load_thread(int flag_true_if_should_unpremultiply)
+{
+ stbi__unpremultiply_on_load_local = flag_true_if_should_unpremultiply;
+ stbi__unpremultiply_on_load_set = 1;
+}
+
+STBIDEF void stbi_convert_iphone_png_to_rgb_thread(int flag_true_if_should_convert)
+{
+ stbi__de_iphone_flag_local = flag_true_if_should_convert;
+ stbi__de_iphone_flag_set = 1;
}
+#define stbi__unpremultiply_on_load (stbi__unpremultiply_on_load_set \
+ ? stbi__unpremultiply_on_load_local \
+ : stbi__unpremultiply_on_load_global)
+#define stbi__de_iphone_flag (stbi__de_iphone_flag_set \
+ ? stbi__de_iphone_flag_local \
+ : stbi__de_iphone_flag_global)
+#endif // STBI_THREAD_LOCAL
+
static void stbi__de_iphone(stbi__png *z)
{
stbi__context *s = z->s;
@@ -4272,9 +5051,10 @@ static void stbi__de_iphone(stbi__png *z)
stbi_uc a = p[3];
stbi_uc t = p[0];
if (a) {
- p[0] = p[2] * 255 / a;
- p[1] = p[1] * 255 / a;
- p[2] = t * 255 / a;
+ stbi_uc half = a / 2;
+ p[0] = (p[2] * 255 + half) / a;
+ p[1] = (p[1] * 255 + half) / a;
+ p[2] = ( t * 255 + half) / a;
} else {
p[0] = p[2];
p[2] = t;
@@ -4293,14 +5073,15 @@ static void stbi__de_iphone(stbi__png *z)
}
}
-#define STBI__PNG_TYPE(a,b,c,d) (((a) << 24) + ((b) << 16) + ((c) << 8) + (d))
+#define STBI__PNG_TYPE(a,b,c,d) (((unsigned) (a) << 24) + ((unsigned) (b) << 16) + ((unsigned) (c) << 8) + (unsigned) (d))
static int stbi__parse_png_file(stbi__png *z, int scan, int req_comp)
{
stbi_uc palette[1024], pal_img_n=0;
- stbi_uc has_trans=0, tc[3];
+ stbi_uc has_trans=0, tc[3]={0};
+ stbi__uint16 tc16[3];
stbi__uint32 ioff=0, idata_limit=0, i, pal_len=0;
- int first=1,k,interlace=0, color=0, depth=0, is_iphone=0;
+ int first=1,k,interlace=0, color=0, is_iphone=0;
stbi__context *s = z->s;
z->expanded = NULL;
@@ -4323,10 +5104,13 @@ static int stbi__parse_png_file(stbi__png *z, int scan, int req_comp)
if (!first) return stbi__err("multiple IHDR","Corrupt PNG");
first = 0;
if (c.length != 13) return stbi__err("bad IHDR len","Corrupt PNG");
- s->img_x = stbi__get32be(s); if (s->img_x > (1 << 24)) return stbi__err("too large","Very large image (corrupt?)");
- s->img_y = stbi__get32be(s); if (s->img_y > (1 << 24)) return stbi__err("too large","Very large image (corrupt?)");
- depth = stbi__get8(s); if (depth != 1 && depth != 2 && depth != 4 && depth != 8) return stbi__err("1/2/4/8-bit only","PNG not supported: 1/2/4/8-bit only");
+ s->img_x = stbi__get32be(s);
+ s->img_y = stbi__get32be(s);
+ if (s->img_y > STBI_MAX_DIMENSIONS) return stbi__err("too large","Very large image (corrupt?)");
+ if (s->img_x > STBI_MAX_DIMENSIONS) return stbi__err("too large","Very large image (corrupt?)");
+ z->depth = stbi__get8(s); if (z->depth != 1 && z->depth != 2 && z->depth != 4 && z->depth != 8 && z->depth != 16) return stbi__err("1/2/4/8/16-bit only","PNG not supported: 1/2/4/8/16-bit only");
color = stbi__get8(s); if (color > 6) return stbi__err("bad ctype","Corrupt PNG");
+ if (color == 3 && z->depth == 16) return stbi__err("bad ctype","Corrupt PNG");
if (color == 3) pal_img_n = 3; else if (color & 1) return stbi__err("bad ctype","Corrupt PNG");
comp = stbi__get8(s); if (comp) return stbi__err("bad comp method","Corrupt PNG");
filter= stbi__get8(s); if (filter) return stbi__err("bad filter method","Corrupt PNG");
@@ -4335,14 +5119,13 @@ static int stbi__parse_png_file(stbi__png *z, int scan, int req_comp)
if (!pal_img_n) {
s->img_n = (color & 2 ? 3 : 1) + (color & 4 ? 1 : 0);
if ((1 << 30) / s->img_x / s->img_n < s->img_y) return stbi__err("too large", "Image too large to decode");
- if (scan == STBI__SCAN_header) return 1;
} else {
// if paletted, then pal_n is our final components, and
// img_n is # components to decompress/filter.
s->img_n = 1;
if ((1 << 30) / s->img_x / 4 < s->img_y) return stbi__err("too large","Corrupt PNG");
- // if SCAN_header, have to scan to see if we have a tRNS
}
+ // even with SCAN_header, have to scan to see if we have a tRNS
break;
}
@@ -4374,8 +5157,15 @@ static int stbi__parse_png_file(stbi__png *z, int scan, int req_comp)
if (!(s->img_n & 1)) return stbi__err("tRNS with alpha","Corrupt PNG");
if (c.length != (stbi__uint32) s->img_n*2) return stbi__err("bad tRNS len","Corrupt PNG");
has_trans = 1;
- for (k=0; k < s->img_n; ++k)
- tc[k] = (stbi_uc) (stbi__get16be(s) & 255) * stbi__depth_scale_table[depth]; // non 8-bit images will be larger
+ // non-paletted with tRNS = constant alpha. if header-scanning, we can stop now.
+ if (scan == STBI__SCAN_header) { ++s->img_n; return 1; }
+ if (z->depth == 16) {
+ for (k = 0; k < s->img_n && k < 3; ++k) // extra loop test to suppress false GCC warning
+ tc16[k] = (stbi__uint16)stbi__get16be(s); // copy the values as-is
+ } else {
+ for (k = 0; k < s->img_n && k < 3; ++k)
+ tc[k] = (stbi_uc)(stbi__get16be(s) & 255) * stbi__depth_scale_table[z->depth]; // non 8-bit images will be larger
+ }
}
break;
}
@@ -4383,14 +5173,22 @@ static int stbi__parse_png_file(stbi__png *z, int scan, int req_comp)
case STBI__PNG_TYPE('I','D','A','T'): {
if (first) return stbi__err("first not IHDR", "Corrupt PNG");
if (pal_img_n && !pal_len) return stbi__err("no PLTE","Corrupt PNG");
- if (scan == STBI__SCAN_header) { s->img_n = pal_img_n; return 1; }
+ if (scan == STBI__SCAN_header) {
+ // header scan definitely stops at first IDAT
+ if (pal_img_n)
+ s->img_n = pal_img_n;
+ return 1;
+ }
+ if (c.length > (1u << 30)) return stbi__err("IDAT size limit", "IDAT section larger than 2^30 bytes");
if ((int)(ioff + c.length) < (int)ioff) return 0;
if (ioff + c.length > idata_limit) {
+ stbi__uint32 idata_limit_old = idata_limit;
stbi_uc *p;
if (idata_limit == 0) idata_limit = c.length > 4096 ? c.length : 4096;
while (ioff + c.length > idata_limit)
idata_limit *= 2;
- p = (stbi_uc *) STBI_REALLOC(z->idata, idata_limit); if (p == NULL) return stbi__err("outofmem", "Out of memory");
+ STBI_NOTUSED(idata_limit_old);
+ p = (stbi_uc *) STBI_REALLOC_SIZED(z->idata, idata_limit_old, idata_limit); if (p == NULL) return stbi__err("outofmem", "Out of memory");
z->idata = p;
}
if (!stbi__getn(s, z->idata+ioff,c.length)) return stbi__err("outofdata","Corrupt PNG");
@@ -4404,7 +5202,7 @@ static int stbi__parse_png_file(stbi__png *z, int scan, int req_comp)
if (scan != STBI__SCAN_load) return 1;
if (z->idata == NULL) return stbi__err("no IDAT","Corrupt PNG");
// initial guess for decoded data size to avoid unnecessary reallocs
- bpl = (s->img_x * depth + 7) / 8; // bytes per line, per component
+ bpl = (s->img_x * z->depth + 7) / 8; // bytes per line, per component
raw_len = bpl * s->img_y * s->img_n /* pixels */ + s->img_y /* filter mode per row */;
z->expanded = (stbi_uc *) stbi_zlib_decode_malloc_guesssize_headerflag((char *) z->idata, ioff, raw_len, (int *) &raw_len, !is_iphone);
if (z->expanded == NULL) return 0; // zlib should set error
@@ -4413,9 +5211,14 @@ static int stbi__parse_png_file(stbi__png *z, int scan, int req_comp)
s->img_out_n = s->img_n+1;
else
s->img_out_n = s->img_n;
- if (!stbi__create_png_image(z, z->expanded, raw_len, s->img_out_n, depth, color, interlace)) return 0;
- if (has_trans)
- if (!stbi__compute_transparency(z, tc, s->img_out_n)) return 0;
+ if (!stbi__create_png_image(z, z->expanded, raw_len, s->img_out_n, z->depth, color, interlace)) return 0;
+ if (has_trans) {
+ if (z->depth == 16) {
+ if (!stbi__compute_transparency16(z, tc16, s->img_out_n)) return 0;
+ } else {
+ if (!stbi__compute_transparency(z, tc, s->img_out_n)) return 0;
+ }
+ }
if (is_iphone && stbi__de_iphone_flag && s->img_out_n > 2)
stbi__de_iphone(z);
if (pal_img_n) {
@@ -4425,8 +5228,13 @@ static int stbi__parse_png_file(stbi__png *z, int scan, int req_comp)
if (req_comp >= 3) s->img_out_n = req_comp;
if (!stbi__expand_png_palette(z, palette, pal_len, s->img_out_n))
return 0;
+ } else if (has_trans) {
+ // non-paletted image with tRNS -> source image has (constant) alpha
+ ++s->img_n;
}
STBI_FREE(z->expanded); z->expanded = NULL;
+ // end of PNG chunk, read and skip CRC
+ stbi__get32be(s);
return 1;
}
@@ -4452,21 +5260,30 @@ static int stbi__parse_png_file(stbi__png *z, int scan, int req_comp)
}
}
-static unsigned char *stbi__do_png(stbi__png *p, int *x, int *y, int *n, int req_comp)
+static void *stbi__do_png(stbi__png *p, int *x, int *y, int *n, int req_comp, stbi__result_info *ri)
{
- unsigned char *result=NULL;
+ void *result=NULL;
if (req_comp < 0 || req_comp > 4) return stbi__errpuc("bad req_comp", "Internal error");
if (stbi__parse_png_file(p, STBI__SCAN_load, req_comp)) {
+ if (p->depth <= 8)
+ ri->bits_per_channel = 8;
+ else if (p->depth == 16)
+ ri->bits_per_channel = 16;
+ else
+ return stbi__errpuc("bad bits_per_channel", "PNG not supported: unsupported color depth");
result = p->out;
p->out = NULL;
if (req_comp && req_comp != p->s->img_out_n) {
- result = stbi__convert_format(result, p->s->img_out_n, req_comp, p->s->img_x, p->s->img_y);
+ if (ri->bits_per_channel == 8)
+ result = stbi__convert_format((unsigned char *) result, p->s->img_out_n, req_comp, p->s->img_x, p->s->img_y);
+ else
+ result = stbi__convert_format16((stbi__uint16 *) result, p->s->img_out_n, req_comp, p->s->img_x, p->s->img_y);
p->s->img_out_n = req_comp;
if (result == NULL) return result;
}
*x = p->s->img_x;
*y = p->s->img_y;
- if (n) *n = p->s->img_out_n;
+ if (n) *n = p->s->img_n;
}
STBI_FREE(p->out); p->out = NULL;
STBI_FREE(p->expanded); p->expanded = NULL;
@@ -4475,11 +5292,11 @@ static unsigned char *stbi__do_png(stbi__png *p, int *x, int *y, int *n, int req
return result;
}
-static unsigned char *stbi__png_load(stbi__context *s, int *x, int *y, int *comp, int req_comp)
+static void *stbi__png_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri)
{
stbi__png p;
p.s = s;
- return stbi__do_png(&p, x,y,comp,req_comp);
+ return stbi__do_png(&p, x,y,comp,req_comp, ri);
}
static int stbi__png_test(stbi__context *s)
@@ -4508,6 +5325,19 @@ static int stbi__png_info(stbi__context *s, int *x, int *y, int *comp)
p.s = s;
return stbi__png_info_raw(&p, x, y, comp);
}
+
+static int stbi__png_is16(stbi__context *s)
+{
+ stbi__png p;
+ p.s = s;
+ if (!stbi__png_info_raw(&p, NULL, NULL, NULL))
+ return 0;
+ if (p.depth != 16) {
+ stbi__rewind(p.s);
+ return 0;
+ }
+ return 1;
+}
#endif
// Microsoft/Windows BMP image
@@ -4541,11 +5371,11 @@ static int stbi__high_bit(unsigned int z)
{
int n=0;
if (z == 0) return -1;
- if (z >= 0x10000) n += 16, z >>= 16;
- if (z >= 0x00100) n += 8, z >>= 8;
- if (z >= 0x00010) n += 4, z >>= 4;
- if (z >= 0x00004) n += 2, z >>= 2;
- if (z >= 0x00002) n += 1, z >>= 1;
+ if (z >= 0x10000) { n += 16; z >>= 16; }
+ if (z >= 0x00100) { n += 8; z >>= 8; }
+ if (z >= 0x00010) { n += 4; z >>= 4; }
+ if (z >= 0x00004) { n += 2; z >>= 2; }
+ if (z >= 0x00002) { n += 1;/* >>= 1;*/ }
return n;
}
@@ -4559,36 +5389,76 @@ static int stbi__bitcount(unsigned int a)
return a & 0xff;
}
-static int stbi__shiftsigned(int v, int shift, int bits)
+// extract an arbitrarily-aligned N-bit value (N=bits)
+// from v, and then make it 8-bits long and fractionally
+// extend it to full full range.
+static int stbi__shiftsigned(unsigned int v, int shift, int bits)
+{
+ static unsigned int mul_table[9] = {
+ 0,
+ 0xff/*0b11111111*/, 0x55/*0b01010101*/, 0x49/*0b01001001*/, 0x11/*0b00010001*/,
+ 0x21/*0b00100001*/, 0x41/*0b01000001*/, 0x81/*0b10000001*/, 0x01/*0b00000001*/,
+ };
+ static unsigned int shift_table[9] = {
+ 0, 0,0,1,0,2,4,6,0,
+ };
+ if (shift < 0)
+ v <<= -shift;
+ else
+ v >>= shift;
+ STBI_ASSERT(v < 256);
+ v >>= (8-bits);
+ STBI_ASSERT(bits >= 0 && bits <= 8);
+ return (int) ((unsigned) v * mul_table[bits]) >> shift_table[bits];
+}
+
+typedef struct
{
- int result;
- int z=0;
+ int bpp, offset, hsz;
+ unsigned int mr,mg,mb,ma, all_a;
+ int extra_read;
+} stbi__bmp_data;
- if (shift < 0) v <<= -shift;
- else v >>= shift;
- result = v;
+static int stbi__bmp_set_mask_defaults(stbi__bmp_data *info, int compress)
+{
+ // BI_BITFIELDS specifies masks explicitly, don't override
+ if (compress == 3)
+ return 1;
- z = bits;
- while (z < 8) {
- result += v >> z;
- z += bits;
+ if (compress == 0) {
+ if (info->bpp == 16) {
+ info->mr = 31u << 10;
+ info->mg = 31u << 5;
+ info->mb = 31u << 0;
+ } else if (info->bpp == 32) {
+ info->mr = 0xffu << 16;
+ info->mg = 0xffu << 8;
+ info->mb = 0xffu << 0;
+ info->ma = 0xffu << 24;
+ info->all_a = 0; // if all_a is 0 at end, then we loaded alpha channel but it was all 0
+ } else {
+ // otherwise, use defaults, which is all-0
+ info->mr = info->mg = info->mb = info->ma = 0;
+ }
+ return 1;
}
- return result;
+ return 0; // error
}
-static stbi_uc *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int req_comp)
+static void *stbi__bmp_parse_header(stbi__context *s, stbi__bmp_data *info)
{
- stbi_uc *out;
- unsigned int mr=0,mg=0,mb=0,ma=0, fake_a=0;
- stbi_uc pal[256][4];
- int psize=0,i,j,compress=0,width;
- int bpp, flip_vertically, pad, target, offset, hsz;
+ int hsz;
if (stbi__get8(s) != 'B' || stbi__get8(s) != 'M') return stbi__errpuc("not BMP", "Corrupt BMP");
stbi__get32le(s); // discard filesize
stbi__get16le(s); // discard reserved
stbi__get16le(s); // discard reserved
- offset = stbi__get32le(s);
- hsz = stbi__get32le(s);
+ info->offset = stbi__get32le(s);
+ info->hsz = hsz = stbi__get32le(s);
+ info->mr = info->mg = info->mb = info->ma = 0;
+ info->extra_read = 14;
+
+ if (info->offset < 0) return stbi__errpuc("bad BMP", "bad BMP");
+
if (hsz != 12 && hsz != 40 && hsz != 56 && hsz != 108 && hsz != 124) return stbi__errpuc("unknown BMP", "BMP type not supported: unknown");
if (hsz == 12) {
s->img_x = stbi__get16le(s);
@@ -4598,16 +5468,12 @@ static stbi_uc *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int
s->img_y = stbi__get32le(s);
}
if (stbi__get16le(s) != 1) return stbi__errpuc("bad BMP", "bad BMP");
- bpp = stbi__get16le(s);
- if (bpp == 1) return stbi__errpuc("monochrome", "BMP type not supported: 1-bit");
- flip_vertically = ((int) s->img_y) > 0;
- s->img_y = abs((int) s->img_y);
- if (hsz == 12) {
- if (bpp < 24)
- psize = (offset - 14 - 24) / 3;
- } else {
- compress = stbi__get32le(s);
+ info->bpp = stbi__get16le(s);
+ if (hsz != 12) {
+ int compress = stbi__get32le(s);
if (compress == 1 || compress == 2) return stbi__errpuc("BMP RLE", "BMP type not supported: RLE");
+ if (compress >= 4) return stbi__errpuc("BMP JPEG/PNG", "BMP type not supported: unsupported compression"); // this includes PNG/JPEG modes
+ if (compress == 3 && info->bpp != 16 && info->bpp != 32) return stbi__errpuc("bad BMP", "bad BMP"); // bitfields requires 16 or 32 bits/pixel
stbi__get32le(s); // discard sizeof
stbi__get32le(s); // discard hres
stbi__get32le(s); // discard vres
@@ -4620,27 +5486,16 @@ static stbi_uc *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int
stbi__get32le(s);
stbi__get32le(s);
}
- if (bpp == 16 || bpp == 32) {
- mr = mg = mb = 0;
+ if (info->bpp == 16 || info->bpp == 32) {
if (compress == 0) {
- if (bpp == 32) {
- mr = 0xffu << 16;
- mg = 0xffu << 8;
- mb = 0xffu << 0;
- ma = 0xffu << 24;
- fake_a = 1; // @TODO: check for cases like alpha value is all 0 and switch it to 255
- STBI_NOTUSED(fake_a);
- } else {
- mr = 31u << 10;
- mg = 31u << 5;
- mb = 31u << 0;
- }
+ stbi__bmp_set_mask_defaults(info, compress);
} else if (compress == 3) {
- mr = stbi__get32le(s);
- mg = stbi__get32le(s);
- mb = stbi__get32le(s);
+ info->mr = stbi__get32le(s);
+ info->mg = stbi__get32le(s);
+ info->mb = stbi__get32le(s);
+ info->extra_read += 12;
// not documented, but generated by photoshop and handled by mspaint
- if (mr == mg && mg == mb) {
+ if (info->mr == info->mg && info->mg == info->mb) {
// ?!?!?
return stbi__errpuc("bad BMP", "bad BMP");
}
@@ -4648,11 +5503,16 @@ static stbi_uc *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int
return stbi__errpuc("bad BMP", "bad BMP");
}
} else {
- STBI_ASSERT(hsz == 108 || hsz == 124);
- mr = stbi__get32le(s);
- mg = stbi__get32le(s);
- mb = stbi__get32le(s);
- ma = stbi__get32le(s);
+ // V4/V5 header
+ int i;
+ if (hsz != 108 && hsz != 124)
+ return stbi__errpuc("bad BMP", "bad BMP");
+ info->mr = stbi__get32le(s);
+ info->mg = stbi__get32le(s);
+ info->mb = stbi__get32le(s);
+ info->ma = stbi__get32le(s);
+ if (compress != 3) // override mr/mg/mb unless in BI_BITFIELDS mode, as per docs
+ stbi__bmp_set_mask_defaults(info, compress);
stbi__get32le(s); // discard color space
for (i=0; i < 12; ++i)
stbi__get32le(s); // discard color space parameters
@@ -4663,63 +5523,146 @@ static stbi_uc *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int
stbi__get32le(s); // discard reserved
}
}
- if (bpp < 16)
- psize = (offset - 14 - hsz) >> 2;
}
- s->img_n = ma ? 4 : 3;
+ return (void *) 1;
+}
+
+
+static void *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri)
+{
+ stbi_uc *out;
+ unsigned int mr=0,mg=0,mb=0,ma=0, all_a;
+ stbi_uc pal[256][4];
+ int psize=0,i,j,width;
+ int flip_vertically, pad, target;
+ stbi__bmp_data info;
+ STBI_NOTUSED(ri);
+
+ info.all_a = 255;
+ if (stbi__bmp_parse_header(s, &info) == NULL)
+ return NULL; // error code already set
+
+ flip_vertically = ((int) s->img_y) > 0;
+ s->img_y = abs((int) s->img_y);
+
+ if (s->img_y > STBI_MAX_DIMENSIONS) return stbi__errpuc("too large","Very large image (corrupt?)");
+ if (s->img_x > STBI_MAX_DIMENSIONS) return stbi__errpuc("too large","Very large image (corrupt?)");
+
+ mr = info.mr;
+ mg = info.mg;
+ mb = info.mb;
+ ma = info.ma;
+ all_a = info.all_a;
+
+ if (info.hsz == 12) {
+ if (info.bpp < 24)
+ psize = (info.offset - info.extra_read - 24) / 3;
+ } else {
+ if (info.bpp < 16)
+ psize = (info.offset - info.extra_read - info.hsz) >> 2;
+ }
+ if (psize == 0) {
+ // accept some number of extra bytes after the header, but if the offset points either to before
+ // the header ends or implies a large amount of extra data, reject the file as malformed
+ int bytes_read_so_far = s->callback_already_read + (int)(s->img_buffer - s->img_buffer_original);
+ int header_limit = 1024; // max we actually read is below 256 bytes currently.
+ int extra_data_limit = 256*4; // what ordinarily goes here is a palette; 256 entries*4 bytes is its max size.
+ if (bytes_read_so_far <= 0 || bytes_read_so_far > header_limit) {
+ return stbi__errpuc("bad header", "Corrupt BMP");
+ }
+ // we established that bytes_read_so_far is positive and sensible.
+ // the first half of this test rejects offsets that are either too small positives, or
+ // negative, and guarantees that info.offset >= bytes_read_so_far > 0. this in turn
+ // ensures the number computed in the second half of the test can't overflow.
+ if (info.offset < bytes_read_so_far || info.offset - bytes_read_so_far > extra_data_limit) {
+ return stbi__errpuc("bad offset", "Corrupt BMP");
+ } else {
+ stbi__skip(s, info.offset - bytes_read_so_far);
+ }
+ }
+
+ if (info.bpp == 24 && ma == 0xff000000)
+ s->img_n = 3;
+ else
+ s->img_n = ma ? 4 : 3;
if (req_comp && req_comp >= 3) // we can directly decode 3 or 4
target = req_comp;
else
target = s->img_n; // if they want monochrome, we'll post-convert
- out = (stbi_uc *) stbi__malloc(target * s->img_x * s->img_y);
+
+ // sanity-check size
+ if (!stbi__mad3sizes_valid(target, s->img_x, s->img_y, 0))
+ return stbi__errpuc("too large", "Corrupt BMP");
+
+ out = (stbi_uc *) stbi__malloc_mad3(target, s->img_x, s->img_y, 0);
if (!out) return stbi__errpuc("outofmem", "Out of memory");
- if (bpp < 16) {
+ if (info.bpp < 16) {
int z=0;
if (psize == 0 || psize > 256) { STBI_FREE(out); return stbi__errpuc("invalid", "Corrupt BMP"); }
for (i=0; i < psize; ++i) {
pal[i][2] = stbi__get8(s);
pal[i][1] = stbi__get8(s);
pal[i][0] = stbi__get8(s);
- if (hsz != 12) stbi__get8(s);
+ if (info.hsz != 12) stbi__get8(s);
pal[i][3] = 255;
}
- stbi__skip(s, offset - 14 - hsz - psize * (hsz == 12 ? 3 : 4));
- if (bpp == 4) width = (s->img_x + 1) >> 1;
- else if (bpp == 8) width = s->img_x;
+ stbi__skip(s, info.offset - info.extra_read - info.hsz - psize * (info.hsz == 12 ? 3 : 4));
+ if (info.bpp == 1) width = (s->img_x + 7) >> 3;
+ else if (info.bpp == 4) width = (s->img_x + 1) >> 1;
+ else if (info.bpp == 8) width = s->img_x;
else { STBI_FREE(out); return stbi__errpuc("bad bpp", "Corrupt BMP"); }
pad = (-width)&3;
- for (j=0; j < (int) s->img_y; ++j) {
- for (i=0; i < (int) s->img_x; i += 2) {
- int v=stbi__get8(s),v2=0;
- if (bpp == 4) {
- v2 = v & 15;
- v >>= 4;
+ if (info.bpp == 1) {
+ for (j=0; j < (int) s->img_y; ++j) {
+ int bit_offset = 7, v = stbi__get8(s);
+ for (i=0; i < (int) s->img_x; ++i) {
+ int color = (v>>bit_offset)&0x1;
+ out[z++] = pal[color][0];
+ out[z++] = pal[color][1];
+ out[z++] = pal[color][2];
+ if (target == 4) out[z++] = 255;
+ if (i+1 == (int) s->img_x) break;
+ if((--bit_offset) < 0) {
+ bit_offset = 7;
+ v = stbi__get8(s);
+ }
}
- out[z++] = pal[v][0];
- out[z++] = pal[v][1];
- out[z++] = pal[v][2];
- if (target == 4) out[z++] = 255;
- if (i+1 == (int) s->img_x) break;
- v = (bpp == 8) ? stbi__get8(s) : v2;
- out[z++] = pal[v][0];
- out[z++] = pal[v][1];
- out[z++] = pal[v][2];
- if (target == 4) out[z++] = 255;
+ stbi__skip(s, pad);
+ }
+ } else {
+ for (j=0; j < (int) s->img_y; ++j) {
+ for (i=0; i < (int) s->img_x; i += 2) {
+ int v=stbi__get8(s),v2=0;
+ if (info.bpp == 4) {
+ v2 = v & 15;
+ v >>= 4;
+ }
+ out[z++] = pal[v][0];
+ out[z++] = pal[v][1];
+ out[z++] = pal[v][2];
+ if (target == 4) out[z++] = 255;
+ if (i+1 == (int) s->img_x) break;
+ v = (info.bpp == 8) ? stbi__get8(s) : v2;
+ out[z++] = pal[v][0];
+ out[z++] = pal[v][1];
+ out[z++] = pal[v][2];
+ if (target == 4) out[z++] = 255;
+ }
+ stbi__skip(s, pad);
}
- stbi__skip(s, pad);
}
} else {
int rshift=0,gshift=0,bshift=0,ashift=0,rcount=0,gcount=0,bcount=0,acount=0;
int z = 0;
int easy=0;
- stbi__skip(s, offset - 14 - hsz);
- if (bpp == 24) width = 3 * s->img_x;
- else if (bpp == 16) width = 2*s->img_x;
+ stbi__skip(s, info.offset - info.extra_read - info.hsz);
+ if (info.bpp == 24) width = 3 * s->img_x;
+ else if (info.bpp == 16) width = 2*s->img_x;
else /* bpp = 32 and pad = 0 */ width=0;
pad = (-width) & 3;
- if (bpp == 24) {
+ if (info.bpp == 24) {
easy = 1;
- } else if (bpp == 32) {
+ } else if (info.bpp == 32) {
if (mb == 0xff && mg == 0xff00 && mr == 0x00ff0000 && ma == 0xff000000)
easy = 2;
}
@@ -4730,6 +5673,7 @@ static stbi_uc *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int
gshift = stbi__high_bit(mg)-7; gcount = stbi__bitcount(mg);
bshift = stbi__high_bit(mb)-7; bcount = stbi__bitcount(mb);
ashift = stbi__high_bit(ma)-7; acount = stbi__bitcount(ma);
+ if (rcount > 8 || gcount > 8 || bcount > 8 || acount > 8) { STBI_FREE(out); return stbi__errpuc("bad masks", "Corrupt BMP"); }
}
for (j=0; j < (int) s->img_y; ++j) {
if (easy) {
@@ -4740,29 +5684,38 @@ static stbi_uc *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int
out[z+0] = stbi__get8(s);
z += 3;
a = (easy == 2 ? stbi__get8(s) : 255);
+ all_a |= a;
if (target == 4) out[z++] = a;
}
} else {
+ int bpp = info.bpp;
for (i=0; i < (int) s->img_x; ++i) {
stbi__uint32 v = (bpp == 16 ? (stbi__uint32) stbi__get16le(s) : stbi__get32le(s));
- int a;
+ unsigned int a;
out[z++] = STBI__BYTECAST(stbi__shiftsigned(v & mr, rshift, rcount));
out[z++] = STBI__BYTECAST(stbi__shiftsigned(v & mg, gshift, gcount));
out[z++] = STBI__BYTECAST(stbi__shiftsigned(v & mb, bshift, bcount));
a = (ma ? stbi__shiftsigned(v & ma, ashift, acount) : 255);
+ all_a |= a;
if (target == 4) out[z++] = STBI__BYTECAST(a);
}
}
stbi__skip(s, pad);
}
}
+
+ // if alpha channel is all 0s, replace with all 255s
+ if (target == 4 && all_a == 0)
+ for (i=4*s->img_x*s->img_y-1; i >= 0; i -= 4)
+ out[i] = 255;
+
if (flip_vertically) {
stbi_uc t;
for (j=0; j < (int) s->img_y>>1; ++j) {
stbi_uc *p1 = out + j *s->img_x*target;
stbi_uc *p2 = out + (s->img_y-1-j)*s->img_x*target;
for (i=0; i < (int) s->img_x*target; ++i) {
- t = p1[i], p1[i] = p2[i], p2[i] = t;
+ t = p1[i]; p1[i] = p2[i]; p2[i] = t;
}
}
}
@@ -4782,20 +5735,55 @@ static stbi_uc *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int
// Targa Truevision - TGA
// by Jonathan Dummer
#ifndef STBI_NO_TGA
+// returns STBI_rgb or whatever, 0 on error
+static int stbi__tga_get_comp(int bits_per_pixel, int is_grey, int* is_rgb16)
+{
+ // only RGB or RGBA (incl. 16bit) or grey allowed
+ if (is_rgb16) *is_rgb16 = 0;
+ switch(bits_per_pixel) {
+ case 8: return STBI_grey;
+ case 16: if(is_grey) return STBI_grey_alpha;
+ // fallthrough
+ case 15: if(is_rgb16) *is_rgb16 = 1;
+ return STBI_rgb;
+ case 24: // fallthrough
+ case 32: return bits_per_pixel/8;
+ default: return 0;
+ }
+}
+
static int stbi__tga_info(stbi__context *s, int *x, int *y, int *comp)
{
- int tga_w, tga_h, tga_comp;
- int sz;
+ int tga_w, tga_h, tga_comp, tga_image_type, tga_bits_per_pixel, tga_colormap_bpp;
+ int sz, tga_colormap_type;
stbi__get8(s); // discard Offset
- sz = stbi__get8(s); // color type
- if( sz > 1 ) {
+ tga_colormap_type = stbi__get8(s); // colormap type
+ if( tga_colormap_type > 1 ) {
stbi__rewind(s);
return 0; // only RGB or indexed allowed
}
- sz = stbi__get8(s); // image type
- // only RGB or grey allowed, +/- RLE
- if ((sz != 1) && (sz != 2) && (sz != 3) && (sz != 9) && (sz != 10) && (sz != 11)) return 0;
- stbi__skip(s,9);
+ tga_image_type = stbi__get8(s); // image type
+ if ( tga_colormap_type == 1 ) { // colormapped (paletted) image
+ if (tga_image_type != 1 && tga_image_type != 9) {
+ stbi__rewind(s);
+ return 0;
+ }
+ stbi__skip(s,4); // skip index of first colormap entry and number of entries
+ sz = stbi__get8(s); // check bits per palette color entry
+ if ( (sz != 8) && (sz != 15) && (sz != 16) && (sz != 24) && (sz != 32) ) {
+ stbi__rewind(s);
+ return 0;
+ }
+ stbi__skip(s,4); // skip image x and y origin
+ tga_colormap_bpp = sz;
+ } else { // "normal" image w/o colormap - only RGB or grey allowed, +/- RLE
+ if ( (tga_image_type != 2) && (tga_image_type != 3) && (tga_image_type != 10) && (tga_image_type != 11) ) {
+ stbi__rewind(s);
+ return 0; // only RGB or grey allowed, +/- RLE
+ }
+ stbi__skip(s,9); // skip colormap specification and image x/y origin
+ tga_colormap_bpp = 0;
+ }
tga_w = stbi__get16le(s);
if( tga_w < 1 ) {
stbi__rewind(s);
@@ -4806,45 +5794,81 @@ static int stbi__tga_info(stbi__context *s, int *x, int *y, int *comp)
stbi__rewind(s);
return 0; // test height
}
- sz = stbi__get8(s); // bits per pixel
- // only RGB or RGBA or grey allowed
- if ((sz != 8) && (sz != 16) && (sz != 24) && (sz != 32)) {
- stbi__rewind(s);
- return 0;
+ tga_bits_per_pixel = stbi__get8(s); // bits per pixel
+ stbi__get8(s); // ignore alpha bits
+ if (tga_colormap_bpp != 0) {
+ if((tga_bits_per_pixel != 8) && (tga_bits_per_pixel != 16)) {
+ // when using a colormap, tga_bits_per_pixel is the size of the indexes
+ // I don't think anything but 8 or 16bit indexes makes sense
+ stbi__rewind(s);
+ return 0;
+ }
+ tga_comp = stbi__tga_get_comp(tga_colormap_bpp, 0, NULL);
+ } else {
+ tga_comp = stbi__tga_get_comp(tga_bits_per_pixel, (tga_image_type == 3) || (tga_image_type == 11), NULL);
+ }
+ if(!tga_comp) {
+ stbi__rewind(s);
+ return 0;
}
- tga_comp = sz;
if (x) *x = tga_w;
if (y) *y = tga_h;
- if (comp) *comp = tga_comp / 8;
+ if (comp) *comp = tga_comp;
return 1; // seems to have passed everything
}
static int stbi__tga_test(stbi__context *s)
{
- int res;
- int sz;
+ int res = 0;
+ int sz, tga_color_type;
stbi__get8(s); // discard Offset
- sz = stbi__get8(s); // color type
- if ( sz > 1 ) return 0; // only RGB or indexed allowed
+ tga_color_type = stbi__get8(s); // color type
+ if ( tga_color_type > 1 ) goto errorEnd; // only RGB or indexed allowed
sz = stbi__get8(s); // image type
- if ( (sz != 1) && (sz != 2) && (sz != 3) && (sz != 9) && (sz != 10) && (sz != 11) ) return 0; // only RGB or grey allowed, +/- RLE
- stbi__get16be(s); // discard palette start
- stbi__get16be(s); // discard palette length
- stbi__get8(s); // discard bits per palette color entry
- stbi__get16be(s); // discard x origin
- stbi__get16be(s); // discard y origin
- if ( stbi__get16be(s) < 1 ) return 0; // test width
- if ( stbi__get16be(s) < 1 ) return 0; // test height
+ if ( tga_color_type == 1 ) { // colormapped (paletted) image
+ if (sz != 1 && sz != 9) goto errorEnd; // colortype 1 demands image type 1 or 9
+ stbi__skip(s,4); // skip index of first colormap entry and number of entries
+ sz = stbi__get8(s); // check bits per palette color entry
+ if ( (sz != 8) && (sz != 15) && (sz != 16) && (sz != 24) && (sz != 32) ) goto errorEnd;
+ stbi__skip(s,4); // skip image x and y origin
+ } else { // "normal" image w/o colormap
+ if ( (sz != 2) && (sz != 3) && (sz != 10) && (sz != 11) ) goto errorEnd; // only RGB or grey allowed, +/- RLE
+ stbi__skip(s,9); // skip colormap specification and image x/y origin
+ }
+ if ( stbi__get16le(s) < 1 ) goto errorEnd; // test width
+ if ( stbi__get16le(s) < 1 ) goto errorEnd; // test height
sz = stbi__get8(s); // bits per pixel
- if ( (sz != 8) && (sz != 16) && (sz != 24) && (sz != 32) )
- res = 0;
- else
- res = 1;
+ if ( (tga_color_type == 1) && (sz != 8) && (sz != 16) ) goto errorEnd; // for colormapped images, bpp is size of an index
+ if ( (sz != 8) && (sz != 15) && (sz != 16) && (sz != 24) && (sz != 32) ) goto errorEnd;
+
+ res = 1; // if we got this far, everything's good and we can return 1 instead of 0
+
+errorEnd:
stbi__rewind(s);
return res;
}
-static stbi_uc *stbi__tga_load(stbi__context *s, int *x, int *y, int *comp, int req_comp)
+// read 16bit value and convert to 24bit RGB
+static void stbi__tga_read_rgb16(stbi__context *s, stbi_uc* out)
+{
+ stbi__uint16 px = (stbi__uint16)stbi__get16le(s);
+ stbi__uint16 fiveBitMask = 31;
+ // we have 3 channels with 5bits each
+ int r = (px >> 10) & fiveBitMask;
+ int g = (px >> 5) & fiveBitMask;
+ int b = px & fiveBitMask;
+ // Note that this saves the data in RGB(A) order, so it doesn't need to be swapped later
+ out[0] = (stbi_uc)((r * 255)/31);
+ out[1] = (stbi_uc)((g * 255)/31);
+ out[2] = (stbi_uc)((b * 255)/31);
+
+ // some people claim that the most significant bit might be used for alpha
+ // (possibly if an alpha-bit is set in the "image descriptor byte")
+ // but that only made 16bit test images completely translucent..
+ // so let's treat all 15 and 16bit TGAs as RGB with no alpha.
+}
+
+static void *stbi__tga_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri)
{
// read in the TGA header stuff
int tga_offset = stbi__get8(s);
@@ -4859,16 +5883,23 @@ static stbi_uc *stbi__tga_load(stbi__context *s, int *x, int *y, int *comp, int
int tga_width = stbi__get16le(s);
int tga_height = stbi__get16le(s);
int tga_bits_per_pixel = stbi__get8(s);
- int tga_comp = tga_bits_per_pixel / 8;
+ int tga_comp, tga_rgb16=0;
int tga_inverted = stbi__get8(s);
+ // int tga_alpha_bits = tga_inverted & 15; // the 4 lowest bits - unused (useless?)
// image data
unsigned char *tga_data;
unsigned char *tga_palette = NULL;
int i, j;
- unsigned char raw_data[4];
+ unsigned char raw_data[4] = {0};
int RLE_count = 0;
int RLE_repeating = 0;
int read_next_pixel = 1;
+ STBI_NOTUSED(ri);
+ STBI_NOTUSED(tga_x_origin); // @TODO
+ STBI_NOTUSED(tga_y_origin); // @TODO
+
+ if (tga_height > STBI_MAX_DIMENSIONS) return stbi__errpuc("too large","Very large image (corrupt?)");
+ if (tga_width > STBI_MAX_DIMENSIONS) return stbi__errpuc("too large","Very large image (corrupt?)");
// do a tiny bit of precessing
if ( tga_image_type >= 8 )
@@ -4876,59 +5907,63 @@ static stbi_uc *stbi__tga_load(stbi__context *s, int *x, int *y, int *comp, int
tga_image_type -= 8;
tga_is_RLE = 1;
}
- /* int tga_alpha_bits = tga_inverted & 15; */
tga_inverted = 1 - ((tga_inverted >> 5) & 1);
- // error check
- if ( //(tga_indexed) ||
- (tga_width < 1) || (tga_height < 1) ||
- (tga_image_type < 1) || (tga_image_type > 3) ||
- ((tga_bits_per_pixel != 8) && (tga_bits_per_pixel != 16) &&
- (tga_bits_per_pixel != 24) && (tga_bits_per_pixel != 32))
- )
- {
- return NULL; // we don't report this as a bad TGA because we don't even know if it's TGA
- }
-
// If I'm paletted, then I'll use the number of bits from the palette
- if ( tga_indexed )
- {
- tga_comp = tga_palette_bits / 8;
- }
+ if ( tga_indexed ) tga_comp = stbi__tga_get_comp(tga_palette_bits, 0, &tga_rgb16);
+ else tga_comp = stbi__tga_get_comp(tga_bits_per_pixel, (tga_image_type == 3), &tga_rgb16);
+
+ if(!tga_comp) // shouldn't really happen, stbi__tga_test() should have ensured basic consistency
+ return stbi__errpuc("bad format", "Can't find out TGA pixelformat");
// tga info
*x = tga_width;
*y = tga_height;
if (comp) *comp = tga_comp;
- tga_data = (unsigned char*)stbi__malloc( (size_t)tga_width * tga_height * tga_comp );
+ if (!stbi__mad3sizes_valid(tga_width, tga_height, tga_comp, 0))
+ return stbi__errpuc("too large", "Corrupt TGA");
+
+ tga_data = (unsigned char*)stbi__malloc_mad3(tga_width, tga_height, tga_comp, 0);
if (!tga_data) return stbi__errpuc("outofmem", "Out of memory");
// skip to the data's starting position (offset usually = 0)
stbi__skip(s, tga_offset );
- if ( !tga_indexed && !tga_is_RLE) {
+ if ( !tga_indexed && !tga_is_RLE && !tga_rgb16 ) {
for (i=0; i < tga_height; ++i) {
- int y = tga_inverted ? tga_height -i - 1 : i;
- stbi_uc *tga_row = tga_data + y*tga_width*tga_comp;
+ int row = tga_inverted ? tga_height -i - 1 : i;
+ stbi_uc *tga_row = tga_data + row*tga_width*tga_comp;
stbi__getn(s, tga_row, tga_width * tga_comp);
}
} else {
// do I need to load a palette?
if ( tga_indexed)
{
+ if (tga_palette_len == 0) { /* you have to have at least one entry! */
+ STBI_FREE(tga_data);
+ return stbi__errpuc("bad palette", "Corrupt TGA");
+ }
+
// any data to skip? (offset usually = 0)
stbi__skip(s, tga_palette_start );
// load the palette
- tga_palette = (unsigned char*)stbi__malloc( tga_palette_len * tga_palette_bits / 8 );
+ tga_palette = (unsigned char*)stbi__malloc_mad2(tga_palette_len, tga_comp, 0);
if (!tga_palette) {
STBI_FREE(tga_data);
return stbi__errpuc("outofmem", "Out of memory");
}
- if (!stbi__getn(s, tga_palette, tga_palette_len * tga_palette_bits / 8 )) {
- STBI_FREE(tga_data);
- STBI_FREE(tga_palette);
- return stbi__errpuc("bad palette", "Corrupt TGA");
+ if (tga_rgb16) {
+ stbi_uc *pal_entry = tga_palette;
+ STBI_ASSERT(tga_comp == STBI_rgb);
+ for (i=0; i < tga_palette_len; ++i) {
+ stbi__tga_read_rgb16(s, pal_entry);
+ pal_entry += tga_comp;
+ }
+ } else if (!stbi__getn(s, tga_palette, tga_palette_len * tga_comp)) {
+ STBI_FREE(tga_data);
+ STBI_FREE(tga_palette);
+ return stbi__errpuc("bad palette", "Corrupt TGA");
}
}
// load the data
@@ -4958,23 +5993,22 @@ static stbi_uc *stbi__tga_load(stbi__context *s, int *x, int *y, int *comp, int
// load however much data we did have
if ( tga_indexed )
{
- // read in 1 byte, then perform the lookup
- int pal_idx = stbi__get8(s);
- if ( pal_idx >= tga_palette_len )
- {
- // invalid index
+ // read in index, then perform the lookup
+ int pal_idx = (tga_bits_per_pixel == 8) ? stbi__get8(s) : stbi__get16le(s);
+ if ( pal_idx >= tga_palette_len ) {
+ // invalid index
pal_idx = 0;
}
- pal_idx *= tga_bits_per_pixel / 8;
- for (j = 0; j*8 < tga_bits_per_pixel; ++j)
- {
+ pal_idx *= tga_comp;
+ for (j = 0; j < tga_comp; ++j) {
raw_data[j] = tga_palette[pal_idx+j];
}
- } else
- {
+ } else if(tga_rgb16) {
+ STBI_ASSERT(tga_comp == STBI_rgb);
+ stbi__tga_read_rgb16(s, raw_data);
+ } else {
// read in the data raw
- for (j = 0; j*8 < tga_bits_per_pixel; ++j)
- {
+ for (j = 0; j < tga_comp; ++j) {
raw_data[j] = stbi__get8(s);
}
}
@@ -5013,8 +6047,8 @@ static stbi_uc *stbi__tga_load(stbi__context *s, int *x, int *y, int *comp, int
}
}
- // swap RGB
- if (tga_comp >= 3)
+ // swap RGB - if the source data was RGB16, it already is in the right order
+ if (tga_comp >= 3 && !tga_rgb16)
{
unsigned char* tga_pixel = tga_data;
for (i=0; i < tga_width * tga_height; ++i)
@@ -5034,6 +6068,7 @@ static stbi_uc *stbi__tga_load(stbi__context *s, int *x, int *y, int *comp, int
// Microsoft's C compilers happy... [8^(
tga_palette_start = tga_palette_len = tga_palette_bits =
tga_x_origin = tga_y_origin = 0;
+ STBI_NOTUSED(tga_palette_start);
// OK, done
return tga_data;
}
@@ -5050,13 +6085,53 @@ static int stbi__psd_test(stbi__context *s)
return r;
}
-static stbi_uc *stbi__psd_load(stbi__context *s, int *x, int *y, int *comp, int req_comp)
+static int stbi__psd_decode_rle(stbi__context *s, stbi_uc *p, int pixelCount)
+{
+ int count, nleft, len;
+
+ count = 0;
+ while ((nleft = pixelCount - count) > 0) {
+ len = stbi__get8(s);
+ if (len == 128) {
+ // No-op.
+ } else if (len < 128) {
+ // Copy next len+1 bytes literally.
+ len++;
+ if (len > nleft) return 0; // corrupt data
+ count += len;
+ while (len) {
+ *p = stbi__get8(s);
+ p += 4;
+ len--;
+ }
+ } else if (len > 128) {
+ stbi_uc val;
+ // Next -len+1 bytes in the dest are replicated from next source byte.
+ // (Interpret len as a negative 8-bit int.)
+ len = 257 - len;
+ if (len > nleft) return 0; // corrupt data
+ val = stbi__get8(s);
+ count += len;
+ while (len) {
+ *p = val;
+ p += 4;
+ len--;
+ }
+ }
+ }
+
+ return 1;
+}
+
+static void *stbi__psd_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri, int bpc)
{
- int pixelCount;
+ int pixelCount;
int channelCount, compression;
- int channel, i, count, len;
+ int channel, i;
+ int bitdepth;
int w,h;
stbi_uc *out;
+ STBI_NOTUSED(ri);
// Check identifier
if (stbi__get32be(s) != 0x38425053) // "8BPS"
@@ -5078,9 +6153,13 @@ static stbi_uc *stbi__psd_load(stbi__context *s, int *x, int *y, int *comp, int
h = stbi__get32be(s);
w = stbi__get32be(s);
+ if (h > STBI_MAX_DIMENSIONS) return stbi__errpuc("too large","Very large image (corrupt?)");
+ if (w > STBI_MAX_DIMENSIONS) return stbi__errpuc("too large","Very large image (corrupt?)");
+
// Make sure the depth is 8 bits.
- if (stbi__get16be(s) != 8)
- return stbi__errpuc("unsupported bit depth", "PSD bit depth is not 8 bit");
+ bitdepth = stbi__get16be(s);
+ if (bitdepth != 8 && bitdepth != 16)
+ return stbi__errpuc("unsupported bit depth", "PSD bit depth is not 8 or 16 bit");
// Make sure the color mode is RGB.
// Valid options are:
@@ -5112,8 +6191,18 @@ static stbi_uc *stbi__psd_load(stbi__context *s, int *x, int *y, int *comp, int
if (compression > 1)
return stbi__errpuc("bad compression", "PSD has an unknown compression format");
+ // Check size
+ if (!stbi__mad3sizes_valid(4, w, h, 0))
+ return stbi__errpuc("too large", "Corrupt PSD");
+
// Create the destination image.
- out = (stbi_uc *) stbi__malloc(4 * w*h);
+
+ if (!compression && bitdepth == 16 && bpc == 16) {
+ out = (stbi_uc *) stbi__malloc_mad3(8, w, h, 0);
+ ri->bits_per_channel = 16;
+ } else
+ out = (stbi_uc *) stbi__malloc(4 * w*h);
+
if (!out) return stbi__errpuc("outofmem", "Out of memory");
pixelCount = w*h;
@@ -5130,7 +6219,7 @@ static stbi_uc *stbi__psd_load(stbi__context *s, int *x, int *y, int *comp, int
// Else if n is 128, noop.
// Endloop
- // The RLE-compressed data is preceeded by a 2-byte data count for each row in the data,
+ // The RLE-compressed data is preceded by a 2-byte data count for each row in the data,
// which we're going to just skip.
stbi__skip(s, h * channelCount * 2 );
@@ -5145,61 +6234,86 @@ static stbi_uc *stbi__psd_load(stbi__context *s, int *x, int *y, int *comp, int
*p = (channel == 3 ? 255 : 0);
} else {
// Read the RLE data.
- count = 0;
- while (count < pixelCount) {
- len = stbi__get8(s);
- if (len == 128) {
- // No-op.
- } else if (len < 128) {
- // Copy next len+1 bytes literally.
- len++;
- count += len;
- while (len) {
- *p = stbi__get8(s);
- p += 4;
- len--;
- }
- } else if (len > 128) {
- stbi_uc val;
- // Next -len+1 bytes in the dest are replicated from next source byte.
- // (Interpret len as a negative 8-bit int.)
- len ^= 0x0FF;
- len += 2;
- val = stbi__get8(s);
- count += len;
- while (len) {
- *p = val;
- p += 4;
- len--;
- }
- }
+ if (!stbi__psd_decode_rle(s, p, pixelCount)) {
+ STBI_FREE(out);
+ return stbi__errpuc("corrupt", "bad RLE data");
}
}
}
} else {
// We're at the raw image data. It's each channel in order (Red, Green, Blue, Alpha, ...)
- // where each channel consists of an 8-bit value for each pixel in the image.
+ // where each channel consists of an 8-bit (or 16-bit) value for each pixel in the image.
// Read the data by channel.
for (channel = 0; channel < 4; channel++) {
- stbi_uc *p;
-
- p = out + channel;
- if (channel > channelCount) {
+ if (channel >= channelCount) {
// Fill this channel with default data.
- for (i = 0; i < pixelCount; i++, p += 4)
- *p = channel == 3 ? 255 : 0;
+ if (bitdepth == 16 && bpc == 16) {
+ stbi__uint16 *q = ((stbi__uint16 *) out) + channel;
+ stbi__uint16 val = channel == 3 ? 65535 : 0;
+ for (i = 0; i < pixelCount; i++, q += 4)
+ *q = val;
+ } else {
+ stbi_uc *p = out+channel;
+ stbi_uc val = channel == 3 ? 255 : 0;
+ for (i = 0; i < pixelCount; i++, p += 4)
+ *p = val;
+ }
} else {
- // Read the data.
- for (i = 0; i < pixelCount; i++, p += 4)
- *p = stbi__get8(s);
+ if (ri->bits_per_channel == 16) { // output bpc
+ stbi__uint16 *q = ((stbi__uint16 *) out) + channel;
+ for (i = 0; i < pixelCount; i++, q += 4)
+ *q = (stbi__uint16) stbi__get16be(s);
+ } else {
+ stbi_uc *p = out+channel;
+ if (bitdepth == 16) { // input bpc
+ for (i = 0; i < pixelCount; i++, p += 4)
+ *p = (stbi_uc) (stbi__get16be(s) >> 8);
+ } else {
+ for (i = 0; i < pixelCount; i++, p += 4)
+ *p = stbi__get8(s);
+ }
+ }
+ }
+ }
+ }
+
+ // remove weird white matte from PSD
+ if (channelCount >= 4) {
+ if (ri->bits_per_channel == 16) {
+ for (i=0; i < w*h; ++i) {
+ stbi__uint16 *pixel = (stbi__uint16 *) out + 4*i;
+ if (pixel[3] != 0 && pixel[3] != 65535) {
+ float a = pixel[3] / 65535.0f;
+ float ra = 1.0f / a;
+ float inv_a = 65535.0f * (1 - ra);
+ pixel[0] = (stbi__uint16) (pixel[0]*ra + inv_a);
+ pixel[1] = (stbi__uint16) (pixel[1]*ra + inv_a);
+ pixel[2] = (stbi__uint16) (pixel[2]*ra + inv_a);
+ }
+ }
+ } else {
+ for (i=0; i < w*h; ++i) {
+ unsigned char *pixel = out + 4*i;
+ if (pixel[3] != 0 && pixel[3] != 255) {
+ float a = pixel[3] / 255.0f;
+ float ra = 1.0f / a;
+ float inv_a = 255.0f * (1 - ra);
+ pixel[0] = (unsigned char) (pixel[0]*ra + inv_a);
+ pixel[1] = (unsigned char) (pixel[1]*ra + inv_a);
+ pixel[2] = (unsigned char) (pixel[2]*ra + inv_a);
+ }
}
}
}
+ // convert to desired output format
if (req_comp && req_comp != 4) {
- out = stbi__convert_format(out, 4, req_comp, w, h);
+ if (ri->bits_per_channel == 16)
+ out = (stbi_uc *) stbi__convert_format16((stbi__uint16 *) out, 4, req_comp, w, h);
+ else
+ out = stbi__convert_format(out, 4, req_comp, w, h);
if (out == NULL) return out; // stbi__convert_format frees input on failure
}
@@ -5351,7 +6465,6 @@ static stbi_uc *stbi__pic_load_core(stbi__context *s,int width,int height,int *c
if (count >= 128) { // Repeated
stbi_uc value[4];
- int i;
if (count==128)
count = stbi__get16be(s);
@@ -5384,25 +6497,33 @@ static stbi_uc *stbi__pic_load_core(stbi__context *s,int width,int height,int *c
return result;
}
-static stbi_uc *stbi__pic_load(stbi__context *s,int *px,int *py,int *comp,int req_comp)
+static void *stbi__pic_load(stbi__context *s,int *px,int *py,int *comp,int req_comp, stbi__result_info *ri)
{
stbi_uc *result;
- int i, x,y;
+ int i, x,y, internal_comp;
+ STBI_NOTUSED(ri);
+
+ if (!comp) comp = &internal_comp;
for (i=0; i<92; ++i)
stbi__get8(s);
x = stbi__get16be(s);
y = stbi__get16be(s);
+
+ if (y > STBI_MAX_DIMENSIONS) return stbi__errpuc("too large","Very large image (corrupt?)");
+ if (x > STBI_MAX_DIMENSIONS) return stbi__errpuc("too large","Very large image (corrupt?)");
+
if (stbi__at_eof(s)) return stbi__errpuc("bad file","file too short (pic header)");
- if ((1 << 28) / x < y) return stbi__errpuc("too large", "Image too large to decode");
+ if (!stbi__mad3sizes_valid(x, y, 4, 0)) return stbi__errpuc("too large", "PIC image too large to decode");
stbi__get32be(s); //skip `ratio'
stbi__get16be(s); //skip `fields'
stbi__get16be(s); //skip `pad'
// intermediate buffer is RGBA
- result = (stbi_uc *) stbi__malloc(x*y*4);
+ result = (stbi_uc *) stbi__malloc_mad3(x, y, 4, 0);
+ if (!result) return stbi__errpuc("outofmem", "Out of memory");
memset(result, 0xff, x*y*4);
if (!stbi__pic_load_core(s,x,y,comp, result)) {
@@ -5440,10 +6561,12 @@ typedef struct
{
int w,h;
stbi_uc *out; // output buffer (always 4 components)
+ stbi_uc *background; // The current "background" as far as a gif is concerned
+ stbi_uc *history;
int flags, bgindex, ratio, transparent, eflags;
stbi_uc pal[256][4];
stbi_uc lpal[256][4];
- stbi__gif_lzw codes[4096];
+ stbi__gif_lzw codes[8192];
stbi_uc *color_table;
int parse, step;
int lflags;
@@ -5451,6 +6574,7 @@ typedef struct
int max_x, max_y;
int cur_x, cur_y;
int line_size;
+ int delay;
} stbi__gif;
static int stbi__gif_test_raw(stbi__context *s)
@@ -5499,6 +6623,9 @@ static int stbi__gif_header(stbi__context *s, stbi__gif *g, int *comp, int is_in
g->ratio = stbi__get8(s);
g->transparent = -1;
+ if (g->w > STBI_MAX_DIMENSIONS) return stbi__err("too large","Very large image (corrupt?)");
+ if (g->h > STBI_MAX_DIMENSIONS) return stbi__err("too large","Very large image (corrupt?)");
+
if (comp != 0) *comp = 4; // can't actually tell whether it's 3 or 4 until we parse the comments
if (is_info) return 1;
@@ -5511,19 +6638,23 @@ static int stbi__gif_header(stbi__context *s, stbi__gif *g, int *comp, int is_in
static int stbi__gif_info_raw(stbi__context *s, int *x, int *y, int *comp)
{
- stbi__gif g;
- if (!stbi__gif_header(s, &g, comp, 1)) {
+ stbi__gif* g = (stbi__gif*) stbi__malloc(sizeof(stbi__gif));
+ if (!g) return stbi__err("outofmem", "Out of memory");
+ if (!stbi__gif_header(s, g, comp, 1)) {
+ STBI_FREE(g);
stbi__rewind( s );
return 0;
}
- if (x) *x = g.w;
- if (y) *y = g.h;
+ if (x) *x = g->w;
+ if (y) *y = g->h;
+ STBI_FREE(g);
return 1;
}
static void stbi__out_gif_code(stbi__gif *g, stbi__uint16 code)
{
stbi_uc *p, *c;
+ int idx;
// recurse to decode the prefixes, since the linked-list is backwards,
// and working backwards through an interleaved image would be nasty
@@ -5532,10 +6663,12 @@ static void stbi__out_gif_code(stbi__gif *g, stbi__uint16 code)
if (g->cur_y >= g->max_y) return;
- p = &g->out[g->cur_x + g->cur_y];
- c = &g->color_table[g->codes[code].suffix * 4];
+ idx = g->cur_x + g->cur_y;
+ p = &g->out[idx];
+ g->history[idx / 4] = 1;
- if (c[3] >= 128) {
+ c = &g->color_table[g->codes[code].suffix * 4];
+ if (c[3] > 128) { // don't render transparent pixels;
p[0] = c[2];
p[1] = c[1];
p[2] = c[0];
@@ -5558,7 +6691,7 @@ static void stbi__out_gif_code(stbi__gif *g, stbi__uint16 code)
static stbi_uc *stbi__process_gif_raster(stbi__context *s, stbi__gif *g)
{
stbi_uc lzw_cs;
- stbi__int32 len, code;
+ stbi__int32 len, init_code;
stbi__uint32 first;
stbi__int32 codesize, codemask, avail, oldcode, bits, valid_bits, clear;
stbi__gif_lzw *p;
@@ -5571,10 +6704,10 @@ static stbi_uc *stbi__process_gif_raster(stbi__context *s, stbi__gif *g)
codemask = (1 << codesize) - 1;
bits = 0;
valid_bits = 0;
- for (code = 0; code < clear; code++) {
- g->codes[code].prefix = -1;
- g->codes[code].first = (stbi_uc) code;
- g->codes[code].suffix = (stbi_uc) code;
+ for (init_code = 0; init_code < clear; init_code++) {
+ g->codes[init_code].prefix = -1;
+ g->codes[init_code].first = (stbi_uc) init_code;
+ g->codes[init_code].suffix = (stbi_uc) init_code;
}
// support no starting clear code
@@ -5609,11 +6742,16 @@ static stbi_uc *stbi__process_gif_raster(stbi__context *s, stbi__gif *g)
stbi__skip(s,len);
return g->out;
} else if (code <= avail) {
- if (first) return stbi__errpuc("no clear code", "Corrupt GIF");
+ if (first) {
+ return stbi__errpuc("no clear code", "Corrupt GIF");
+ }
if (oldcode >= 0) {
p = &g->codes[avail++];
- if (avail > 4096) return stbi__errpuc("too many codes", "Corrupt GIF");
+ if (avail > 8192) {
+ return stbi__errpuc("too many codes", "Corrupt GIF");
+ }
+
p->prefix = (stbi__int16) oldcode;
p->first = g->codes[oldcode].first;
p->suffix = (code == avail) ? p->first : g->codes[code].first;
@@ -5635,43 +6773,75 @@ static stbi_uc *stbi__process_gif_raster(stbi__context *s, stbi__gif *g)
}
}
-static void stbi__fill_gif_background(stbi__gif *g)
-{
- int i;
- stbi_uc *c = g->pal[g->bgindex];
- // @OPTIMIZE: write a dword at a time
- for (i = 0; i < g->w * g->h * 4; i += 4) {
- stbi_uc *p = &g->out[i];
- p[0] = c[2];
- p[1] = c[1];
- p[2] = c[0];
- p[3] = c[3];
- }
-}
-
// this function is designed to support animated gifs, although stb_image doesn't support it
-static stbi_uc *stbi__gif_load_next(stbi__context *s, stbi__gif *g, int *comp, int req_comp)
+// two back is the image from two frames ago, used for a very specific disposal format
+static stbi_uc *stbi__gif_load_next(stbi__context *s, stbi__gif *g, int *comp, int req_comp, stbi_uc *two_back)
{
- int i;
- stbi_uc *old_out = 0;
+ int dispose;
+ int first_frame;
+ int pi;
+ int pcount;
+ STBI_NOTUSED(req_comp);
+ // on first frame, any non-written pixels get the background colour (non-transparent)
+ first_frame = 0;
if (g->out == 0) {
- if (!stbi__gif_header(s, g, comp,0)) return 0; // stbi__g_failure_reason set by stbi__gif_header
- g->out = (stbi_uc *) stbi__malloc(4 * g->w * g->h);
- if (g->out == 0) return stbi__errpuc("outofmem", "Out of memory");
- stbi__fill_gif_background(g);
+ if (!stbi__gif_header(s, g, comp,0)) return 0; // stbi__g_failure_reason set by stbi__gif_header
+ if (!stbi__mad3sizes_valid(4, g->w, g->h, 0))
+ return stbi__errpuc("too large", "GIF image is too large");
+ pcount = g->w * g->h;
+ g->out = (stbi_uc *) stbi__malloc(4 * pcount);
+ g->background = (stbi_uc *) stbi__malloc(4 * pcount);
+ g->history = (stbi_uc *) stbi__malloc(pcount);
+ if (!g->out || !g->background || !g->history)
+ return stbi__errpuc("outofmem", "Out of memory");
+
+ // image is treated as "transparent" at the start - ie, nothing overwrites the current background;
+ // background colour is only used for pixels that are not rendered first frame, after that "background"
+ // color refers to the color that was there the previous frame.
+ memset(g->out, 0x00, 4 * pcount);
+ memset(g->background, 0x00, 4 * pcount); // state of the background (starts transparent)
+ memset(g->history, 0x00, pcount); // pixels that were affected previous frame
+ first_frame = 1;
} else {
- // animated-gif-only path
- if (((g->eflags & 0x1C) >> 2) == 3) {
- old_out = g->out;
- g->out = (stbi_uc *) stbi__malloc(4 * g->w * g->h);
- if (g->out == 0) return stbi__errpuc("outofmem", "Out of memory");
- memcpy(g->out, old_out, g->w*g->h*4);
+ // second frame - how do we dispose of the previous one?
+ dispose = (g->eflags & 0x1C) >> 2;
+ pcount = g->w * g->h;
+
+ if ((dispose == 3) && (two_back == 0)) {
+ dispose = 2; // if I don't have an image to revert back to, default to the old background
+ }
+
+ if (dispose == 3) { // use previous graphic
+ for (pi = 0; pi < pcount; ++pi) {
+ if (g->history[pi]) {
+ memcpy( &g->out[pi * 4], &two_back[pi * 4], 4 );
+ }
+ }
+ } else if (dispose == 2) {
+ // restore what was changed last frame to background before that frame;
+ for (pi = 0; pi < pcount; ++pi) {
+ if (g->history[pi]) {
+ memcpy( &g->out[pi * 4], &g->background[pi * 4], 4 );
+ }
+ }
+ } else {
+ // This is a non-disposal case eithe way, so just
+ // leave the pixels as is, and they will become the new background
+ // 1: do not dispose
+ // 0: not specified.
}
+
+ // background is what out is after the undoing of the previou frame;
+ memcpy( g->background, g->out, 4 * g->w * g->h );
}
+ // clear my history;
+ memset( g->history, 0x00, g->w * g->h ); // pixels that were affected previous frame
+
for (;;) {
- switch (stbi__get8(s)) {
+ int tag = stbi__get8(s);
+ switch (tag) {
case 0x2C: /* Image Descriptor */
{
stbi__int32 x, y, w, h;
@@ -5692,6 +6862,13 @@ static stbi_uc *stbi__gif_load_next(stbi__context *s, stbi__gif *g, int *comp, i
g->cur_x = g->start_x;
g->cur_y = g->start_y;
+ // if the width of the specified rectangle is 0, that means
+ // we may not see *any* pixels or the image is malformed;
+ // to make sure this is caught, move the current y down to
+ // max_y (which is what out_gif_code checks).
+ if (w == 0)
+ g->cur_y = g->max_y;
+
g->lflags = stbi__get8(s);
if (g->lflags & 0x40) {
@@ -5706,38 +6883,60 @@ static stbi_uc *stbi__gif_load_next(stbi__context *s, stbi__gif *g, int *comp, i
stbi__gif_parse_colortable(s,g->lpal, 2 << (g->lflags & 7), g->eflags & 0x01 ? g->transparent : -1);
g->color_table = (stbi_uc *) g->lpal;
} else if (g->flags & 0x80) {
- for (i=0; i < 256; ++i) // @OPTIMIZE: stbi__jpeg_reset only the previous transparent
- g->pal[i][3] = 255;
- if (g->transparent >= 0 && (g->eflags & 0x01))
- g->pal[g->transparent][3] = 0;
g->color_table = (stbi_uc *) g->pal;
} else
return stbi__errpuc("missing color table", "Corrupt GIF");
o = stbi__process_gif_raster(s, g);
- if (o == NULL) return NULL;
+ if (!o) return NULL;
+
+ // if this was the first frame,
+ pcount = g->w * g->h;
+ if (first_frame && (g->bgindex > 0)) {
+ // if first frame, any pixel not drawn to gets the background color
+ for (pi = 0; pi < pcount; ++pi) {
+ if (g->history[pi] == 0) {
+ g->pal[g->bgindex][3] = 255; // just in case it was made transparent, undo that; It will be reset next frame if need be;
+ memcpy( &g->out[pi * 4], &g->pal[g->bgindex], 4 );
+ }
+ }
+ }
- if (req_comp && req_comp != 4)
- o = stbi__convert_format(o, 4, req_comp, g->w, g->h);
return o;
}
case 0x21: // Comment Extension.
{
int len;
- if (stbi__get8(s) == 0xF9) { // Graphic Control Extension.
+ int ext = stbi__get8(s);
+ if (ext == 0xF9) { // Graphic Control Extension.
len = stbi__get8(s);
if (len == 4) {
g->eflags = stbi__get8(s);
- stbi__get16le(s); // delay
- g->transparent = stbi__get8(s);
+ g->delay = 10 * stbi__get16le(s); // delay - 1/100th of a second, saving as 1/1000ths.
+
+ // unset old transparent
+ if (g->transparent >= 0) {
+ g->pal[g->transparent][3] = 255;
+ }
+ if (g->eflags & 0x01) {
+ g->transparent = stbi__get8(s);
+ if (g->transparent >= 0) {
+ g->pal[g->transparent][3] = 0;
+ }
+ } else {
+ // don't need transparent
+ stbi__skip(s, 1);
+ g->transparent = -1;
+ }
} else {
stbi__skip(s, len);
break;
}
}
- while ((len = stbi__get8(s)) != 0)
+ while ((len = stbi__get8(s)) != 0) {
stbi__skip(s, len);
+ }
break;
}
@@ -5750,19 +6949,128 @@ static stbi_uc *stbi__gif_load_next(stbi__context *s, stbi__gif *g, int *comp, i
}
}
-static stbi_uc *stbi__gif_load(stbi__context *s, int *x, int *y, int *comp, int req_comp)
+static void *stbi__load_gif_main_outofmem(stbi__gif *g, stbi_uc *out, int **delays)
+{
+ STBI_FREE(g->out);
+ STBI_FREE(g->history);
+ STBI_FREE(g->background);
+
+ if (out) STBI_FREE(out);
+ if (delays && *delays) STBI_FREE(*delays);
+ return stbi__errpuc("outofmem", "Out of memory");
+}
+
+static void *stbi__load_gif_main(stbi__context *s, int **delays, int *x, int *y, int *z, int *comp, int req_comp)
+{
+ if (stbi__gif_test(s)) {
+ int layers = 0;
+ stbi_uc *u = 0;
+ stbi_uc *out = 0;
+ stbi_uc *two_back = 0;
+ stbi__gif g;
+ int stride;
+ int out_size = 0;
+ int delays_size = 0;
+
+ STBI_NOTUSED(out_size);
+ STBI_NOTUSED(delays_size);
+
+ memset(&g, 0, sizeof(g));
+ if (delays) {
+ *delays = 0;
+ }
+
+ do {
+ u = stbi__gif_load_next(s, &g, comp, req_comp, two_back);
+ if (u == (stbi_uc *) s) u = 0; // end of animated gif marker
+
+ if (u) {
+ *x = g.w;
+ *y = g.h;
+ ++layers;
+ stride = g.w * g.h * 4;
+
+ if (out) {
+ void *tmp = (stbi_uc*) STBI_REALLOC_SIZED( out, out_size, layers * stride );
+ if (!tmp)
+ return stbi__load_gif_main_outofmem(&g, out, delays);
+ else {
+ out = (stbi_uc*) tmp;
+ out_size = layers * stride;
+ }
+
+ if (delays) {
+ int *new_delays = (int*) STBI_REALLOC_SIZED( *delays, delays_size, sizeof(int) * layers );
+ if (!new_delays)
+ return stbi__load_gif_main_outofmem(&g, out, delays);
+ *delays = new_delays;
+ delays_size = layers * sizeof(int);
+ }
+ } else {
+ out = (stbi_uc*)stbi__malloc( layers * stride );
+ if (!out)
+ return stbi__load_gif_main_outofmem(&g, out, delays);
+ out_size = layers * stride;
+ if (delays) {
+ *delays = (int*) stbi__malloc( layers * sizeof(int) );
+ if (!*delays)
+ return stbi__load_gif_main_outofmem(&g, out, delays);
+ delays_size = layers * sizeof(int);
+ }
+ }
+ memcpy( out + ((layers - 1) * stride), u, stride );
+ if (layers >= 2) {
+ two_back = out - 2 * stride;
+ }
+
+ if (delays) {
+ (*delays)[layers - 1U] = g.delay;
+ }
+ }
+ } while (u != 0);
+
+ // free temp buffer;
+ STBI_FREE(g.out);
+ STBI_FREE(g.history);
+ STBI_FREE(g.background);
+
+ // do the final conversion after loading everything;
+ if (req_comp && req_comp != 4)
+ out = stbi__convert_format(out, 4, req_comp, layers * g.w, g.h);
+
+ *z = layers;
+ return out;
+ } else {
+ return stbi__errpuc("not GIF", "Image was not as a gif type.");
+ }
+}
+
+static void *stbi__gif_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri)
{
stbi_uc *u = 0;
stbi__gif g;
memset(&g, 0, sizeof(g));
+ STBI_NOTUSED(ri);
- u = stbi__gif_load_next(s, &g, comp, req_comp);
+ u = stbi__gif_load_next(s, &g, comp, req_comp, 0);
if (u == (stbi_uc *) s) u = 0; // end of animated gif marker
if (u) {
*x = g.w;
*y = g.h;
+
+ // moved conversion to after successful load so that the same
+ // can be done for multiple frames.
+ if (req_comp && req_comp != 4)
+ u = stbi__convert_format(u, 4, req_comp, g.w, g.h);
+ } else if (g.out) {
+ // if there was an error and we allocated an image buffer, free it!
+ STBI_FREE(g.out);
}
+ // free buffers needed for multiple frame loading;
+ STBI_FREE(g.history);
+ STBI_FREE(g.background);
+
return u;
}
@@ -5776,20 +7084,24 @@ static int stbi__gif_info(stbi__context *s, int *x, int *y, int *comp)
// Radiance RGBE HDR loader
// originally by Nicolas Schulz
#ifndef STBI_NO_HDR
-static int stbi__hdr_test_core(stbi__context *s)
+static int stbi__hdr_test_core(stbi__context *s, const char *signature)
{
- const char *signature = "#?RADIANCE\n";
int i;
for (i=0; signature[i]; ++i)
if (stbi__get8(s) != signature[i])
- return 0;
+ return 0;
+ stbi__rewind(s);
return 1;
}
static int stbi__hdr_test(stbi__context* s)
{
- int r = stbi__hdr_test_core(s);
+ int r = stbi__hdr_test_core(s, "#?RADIANCE\n");
stbi__rewind(s);
+ if(!r) {
+ r = stbi__hdr_test_core(s, "#?RGBE\n");
+ stbi__rewind(s);
+ }
return r;
}
@@ -5843,7 +7155,7 @@ static void stbi__hdr_convert(float *output, stbi_uc *input, int req_comp)
}
}
-static float *stbi__hdr_load(stbi__context *s, int *x, int *y, int *comp, int req_comp)
+static float *stbi__hdr_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri)
{
char buffer[STBI__HDR_BUFLEN];
char *token;
@@ -5854,10 +7166,12 @@ static float *stbi__hdr_load(stbi__context *s, int *x, int *y, int *comp, int re
int len;
unsigned char count, value;
int i, j, k, c1,c2, z;
-
+ const char *headerToken;
+ STBI_NOTUSED(ri);
// Check identifier
- if (strcmp(stbi__hdr_gettoken(s,buffer), "#?RADIANCE") != 0)
+ headerToken = stbi__hdr_gettoken(s,buffer);
+ if (strcmp(headerToken, "#?RADIANCE") != 0 && strcmp(headerToken, "#?RGBE") != 0)
return stbi__errpf("not HDR", "Corrupt HDR image");
// Parse header
@@ -5880,14 +7194,22 @@ static float *stbi__hdr_load(stbi__context *s, int *x, int *y, int *comp, int re
token += 3;
width = (int) strtol(token, NULL, 10);
+ if (height > STBI_MAX_DIMENSIONS) return stbi__errpf("too large","Very large image (corrupt?)");
+ if (width > STBI_MAX_DIMENSIONS) return stbi__errpf("too large","Very large image (corrupt?)");
+
*x = width;
*y = height;
if (comp) *comp = 3;
if (req_comp == 0) req_comp = 3;
+ if (!stbi__mad4sizes_valid(width, height, req_comp, sizeof(float), 0))
+ return stbi__errpf("too large", "HDR image is too large");
+
// Read data
- hdr_data = (float *) stbi__malloc(height * width * req_comp * sizeof(float));
+ hdr_data = (float *) stbi__malloc_mad4(width, height, req_comp, sizeof(float), 0);
+ if (!hdr_data)
+ return stbi__errpf("outofmem", "Out of memory");
// Load image data
// image data is stored as some number of sca
@@ -5926,20 +7248,29 @@ static float *stbi__hdr_load(stbi__context *s, int *x, int *y, int *comp, int re
len <<= 8;
len |= stbi__get8(s);
if (len != width) { STBI_FREE(hdr_data); STBI_FREE(scanline); return stbi__errpf("invalid decoded scanline length", "corrupt HDR"); }
- if (scanline == NULL) scanline = (stbi_uc *) stbi__malloc(width * 4);
+ if (scanline == NULL) {
+ scanline = (stbi_uc *) stbi__malloc_mad2(width, 4, 0);
+ if (!scanline) {
+ STBI_FREE(hdr_data);
+ return stbi__errpf("outofmem", "Out of memory");
+ }
+ }
for (k = 0; k < 4; ++k) {
+ int nleft;
i = 0;
- while (i < width) {
+ while ((nleft = width - i) > 0) {
count = stbi__get8(s);
if (count > 128) {
// Run
value = stbi__get8(s);
count -= 128;
+ if ((count == 0) || (count > nleft)) { STBI_FREE(hdr_data); STBI_FREE(scanline); return stbi__errpf("corrupt", "bad RLE data in HDR"); }
for (z = 0; z < count; ++z)
scanline[i++ * 4 + k] = value;
} else {
// Dump
+ if ((count == 0) || (count > nleft)) { STBI_FREE(hdr_data); STBI_FREE(scanline); return stbi__errpf("corrupt", "bad RLE data in HDR"); }
for (z = 0; z < count; ++z)
scanline[i++ * 4 + k] = stbi__get8(s);
}
@@ -5948,7 +7279,8 @@ static float *stbi__hdr_load(stbi__context *s, int *x, int *y, int *comp, int re
for (i=0; i < width; ++i)
stbi__hdr_convert(hdr_data+(j*width + i)*req_comp, scanline + i*4, req_comp);
}
- STBI_FREE(scanline);
+ if (scanline)
+ STBI_FREE(scanline);
}
return hdr_data;
@@ -5959,8 +7291,13 @@ static int stbi__hdr_info(stbi__context *s, int *x, int *y, int *comp)
char buffer[STBI__HDR_BUFLEN];
char *token;
int valid = 0;
+ int dummy;
- if (strcmp(stbi__hdr_gettoken(s,buffer), "#?RADIANCE") != 0) {
+ if (!x) x = &dummy;
+ if (!y) y = &dummy;
+ if (!comp) comp = &dummy;
+
+ if (stbi__hdr_test(s) == 0) {
stbi__rewind( s );
return 0;
}
@@ -5997,29 +7334,23 @@ static int stbi__hdr_info(stbi__context *s, int *x, int *y, int *comp)
#ifndef STBI_NO_BMP
static int stbi__bmp_info(stbi__context *s, int *x, int *y, int *comp)
{
- int hsz;
- if (stbi__get8(s) != 'B' || stbi__get8(s) != 'M') {
- stbi__rewind( s );
- return 0;
- }
- stbi__skip(s,12);
- hsz = stbi__get32le(s);
- if (hsz != 12 && hsz != 40 && hsz != 56 && hsz != 108 && hsz != 124) {
- stbi__rewind( s );
- return 0;
- }
- if (hsz == 12) {
- *x = stbi__get16le(s);
- *y = stbi__get16le(s);
- } else {
- *x = stbi__get32le(s);
- *y = stbi__get32le(s);
+ void *p;
+ stbi__bmp_data info;
+
+ info.all_a = 255;
+ p = stbi__bmp_parse_header(s, &info);
+ if (p == NULL) {
+ stbi__rewind( s );
+ return 0;
}
- if (stbi__get16le(s) != 1) {
- stbi__rewind( s );
- return 0;
+ if (x) *x = s->img_x;
+ if (y) *y = s->img_y;
+ if (comp) {
+ if (info.bpp == 24 && info.ma == 0xff000000)
+ *comp = 3;
+ else
+ *comp = info.ma ? 4 : 3;
}
- *comp = stbi__get16le(s) / 8;
return 1;
}
#endif
@@ -6027,7 +7358,10 @@ static int stbi__bmp_info(stbi__context *s, int *x, int *y, int *comp)
#ifndef STBI_NO_PSD
static int stbi__psd_info(stbi__context *s, int *x, int *y, int *comp)
{
- int channelCount;
+ int channelCount, dummy, depth;
+ if (!x) x = &dummy;
+ if (!y) y = &dummy;
+ if (!comp) comp = &dummy;
if (stbi__get32be(s) != 0x38425053) {
stbi__rewind( s );
return 0;
@@ -6044,7 +7378,8 @@ static int stbi__psd_info(stbi__context *s, int *x, int *y, int *comp)
}
*y = stbi__get32be(s);
*x = stbi__get32be(s);
- if (stbi__get16be(s) != 8) {
+ depth = stbi__get16be(s);
+ if (depth != 8 && depth != 16) {
stbi__rewind( s );
return 0;
}
@@ -6055,22 +7390,61 @@ static int stbi__psd_info(stbi__context *s, int *x, int *y, int *comp)
*comp = 4;
return 1;
}
+
+static int stbi__psd_is16(stbi__context *s)
+{
+ int channelCount, depth;
+ if (stbi__get32be(s) != 0x38425053) {
+ stbi__rewind( s );
+ return 0;
+ }
+ if (stbi__get16be(s) != 1) {
+ stbi__rewind( s );
+ return 0;
+ }
+ stbi__skip(s, 6);
+ channelCount = stbi__get16be(s);
+ if (channelCount < 0 || channelCount > 16) {
+ stbi__rewind( s );
+ return 0;
+ }
+ STBI_NOTUSED(stbi__get32be(s));
+ STBI_NOTUSED(stbi__get32be(s));
+ depth = stbi__get16be(s);
+ if (depth != 16) {
+ stbi__rewind( s );
+ return 0;
+ }
+ return 1;
+}
#endif
#ifndef STBI_NO_PIC
static int stbi__pic_info(stbi__context *s, int *x, int *y, int *comp)
{
- int act_comp=0,num_packets=0,chained;
+ int act_comp=0,num_packets=0,chained,dummy;
stbi__pic_packet packets[10];
- stbi__skip(s, 92);
+ if (!x) x = &dummy;
+ if (!y) y = &dummy;
+ if (!comp) comp = &dummy;
+
+ if (!stbi__pic_is4(s,"\x53\x80\xF6\x34")) {
+ stbi__rewind(s);
+ return 0;
+ }
+
+ stbi__skip(s, 88);
*x = stbi__get16be(s);
*y = stbi__get16be(s);
- if (stbi__at_eof(s)) return 0;
+ if (stbi__at_eof(s)) {
+ stbi__rewind( s);
+ return 0;
+ }
if ( (*x) != 0 && (1 << 28) / (*x) < (*y)) {
- stbi__rewind( s );
- return 0;
+ stbi__rewind( s );
+ return 0;
}
stbi__skip(s, 8);
@@ -6114,7 +7488,6 @@ static int stbi__pic_info(stbi__context *s, int *x, int *y, int *comp)
// Known limitations:
// Does not support comments in the header section
// Does not support ASCII image data (formats P2 and P3)
-// Does not support 16-bit-per-channel
#ifndef STBI_NO_PNM
@@ -6130,21 +7503,38 @@ static int stbi__pnm_test(stbi__context *s)
return 1;
}
-static stbi_uc *stbi__pnm_load(stbi__context *s, int *x, int *y, int *comp, int req_comp)
+static void *stbi__pnm_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri)
{
stbi_uc *out;
- if (!stbi__pnm_info(s, (int *)&s->img_x, (int *)&s->img_y, (int *)&s->img_n))
+ STBI_NOTUSED(ri);
+
+ ri->bits_per_channel = stbi__pnm_info(s, (int *)&s->img_x, (int *)&s->img_y, (int *)&s->img_n);
+ if (ri->bits_per_channel == 0)
return 0;
+
+ if (s->img_y > STBI_MAX_DIMENSIONS) return stbi__errpuc("too large","Very large image (corrupt?)");
+ if (s->img_x > STBI_MAX_DIMENSIONS) return stbi__errpuc("too large","Very large image (corrupt?)");
+
*x = s->img_x;
*y = s->img_y;
- *comp = s->img_n;
+ if (comp) *comp = s->img_n;
- out = (stbi_uc *) stbi__malloc(s->img_n * s->img_x * s->img_y);
+ if (!stbi__mad4sizes_valid(s->img_n, s->img_x, s->img_y, ri->bits_per_channel / 8, 0))
+ return stbi__errpuc("too large", "PNM too large");
+
+ out = (stbi_uc *) stbi__malloc_mad4(s->img_n, s->img_x, s->img_y, ri->bits_per_channel / 8, 0);
if (!out) return stbi__errpuc("outofmem", "Out of memory");
- stbi__getn(s, out, s->img_n * s->img_x * s->img_y);
+ if (!stbi__getn(s, out, s->img_n * s->img_x * s->img_y * (ri->bits_per_channel / 8))) {
+ STBI_FREE(out);
+ return stbi__errpuc("bad PNM", "PNM file truncated");
+ }
if (req_comp && req_comp != s->img_n) {
- out = stbi__convert_format(out, s->img_n, req_comp, s->img_x, s->img_y);
+ if (ri->bits_per_channel == 16) {
+ out = (stbi_uc *) stbi__convert_format16((stbi__uint16 *) out, s->img_n, req_comp, s->img_x, s->img_y);
+ } else {
+ out = stbi__convert_format(out, s->img_n, req_comp, s->img_x, s->img_y);
+ }
if (out == NULL) return out; // stbi__convert_format frees input on failure
}
return out;
@@ -6157,8 +7547,16 @@ static int stbi__pnm_isspace(char c)
static void stbi__pnm_skip_whitespace(stbi__context *s, char *c)
{
- while (!stbi__at_eof(s) && stbi__pnm_isspace(*c))
- *c = (char) stbi__get8(s);
+ for (;;) {
+ while (!stbi__at_eof(s) && stbi__pnm_isspace(*c))
+ *c = (char) stbi__get8(s);
+
+ if (stbi__at_eof(s) || *c != '#')
+ break;
+
+ while (!stbi__at_eof(s) && *c != '\n' && *c != '\r' )
+ *c = (char) stbi__get8(s);
+ }
}
static int stbi__pnm_isdigit(char c)
@@ -6173,6 +7571,8 @@ static int stbi__pnm_getinteger(stbi__context *s, char *c)
while (!stbi__at_eof(s) && stbi__pnm_isdigit(*c)) {
value = value*10 + (*c - '0');
*c = (char) stbi__get8(s);
+ if((value > 214748364) || (value == 214748364 && *c > '7'))
+ return stbi__err("integer parse overflow", "Parsing an integer in the PPM header overflowed a 32-bit int");
}
return value;
@@ -6180,16 +7580,20 @@ static int stbi__pnm_getinteger(stbi__context *s, char *c)
static int stbi__pnm_info(stbi__context *s, int *x, int *y, int *comp)
{
- int maxv;
+ int maxv, dummy;
char c, p, t;
- stbi__rewind( s );
+ if (!x) x = &dummy;
+ if (!y) y = &dummy;
+ if (!comp) comp = &dummy;
+
+ stbi__rewind(s);
// Get identifier
p = (char) stbi__get8(s);
t = (char) stbi__get8(s);
if (p != 'P' || (t != '5' && t != '6')) {
- stbi__rewind( s );
+ stbi__rewind(s);
return 0;
}
@@ -6199,17 +7603,29 @@ static int stbi__pnm_info(stbi__context *s, int *x, int *y, int *comp)
stbi__pnm_skip_whitespace(s, &c);
*x = stbi__pnm_getinteger(s, &c); // read width
+ if(*x == 0)
+ return stbi__err("invalid width", "PPM image header had zero or overflowing width");
stbi__pnm_skip_whitespace(s, &c);
*y = stbi__pnm_getinteger(s, &c); // read height
+ if (*y == 0)
+ return stbi__err("invalid width", "PPM image header had zero or overflowing width");
stbi__pnm_skip_whitespace(s, &c);
maxv = stbi__pnm_getinteger(s, &c); // read max value
-
- if (maxv > 255)
- return stbi__err("max value > 255", "PPM image not 8-bit");
+ if (maxv > 65535)
+ return stbi__err("max value > 65535", "PPM image supports only 8-bit and 16-bit images");
+ else if (maxv > 255)
+ return 16;
else
- return 1;
+ return 8;
+}
+
+static int stbi__pnm_is16(stbi__context *s)
+{
+ if (stbi__pnm_info(s, NULL, NULL, NULL) == 16)
+ return 1;
+ return 0;
}
#endif
@@ -6255,6 +7671,22 @@ static int stbi__info_main(stbi__context *s, int *x, int *y, int *comp)
return stbi__err("unknown image type", "Image not of any known type, or corrupt");
}
+static int stbi__is_16_main(stbi__context *s)
+{
+ #ifndef STBI_NO_PNG
+ if (stbi__png_is16(s)) return 1;
+ #endif
+
+ #ifndef STBI_NO_PSD
+ if (stbi__psd_is16(s)) return 1;
+ #endif
+
+ #ifndef STBI_NO_PNM
+ if (stbi__pnm_is16(s)) return 1;
+ #endif
+ return 0;
+}
+
#ifndef STBI_NO_STDIO
STBIDEF int stbi_info(char const *filename, int *x, int *y, int *comp)
{
@@ -6276,6 +7708,27 @@ STBIDEF int stbi_info_from_file(FILE *f, int *x, int *y, int *comp)
fseek(f,pos,SEEK_SET);
return r;
}
+
+STBIDEF int stbi_is_16_bit(char const *filename)
+{
+ FILE *f = stbi__fopen(filename, "rb");
+ int result;
+ if (!f) return stbi__err("can't fopen", "Unable to open file");
+ result = stbi_is_16_bit_from_file(f);
+ fclose(f);
+ return result;
+}
+
+STBIDEF int stbi_is_16_bit_from_file(FILE *f)
+{
+ int r;
+ stbi__context s;
+ long pos = ftell(f);
+ stbi__start_file(&s, f);
+ r = stbi__is_16_main(&s);
+ fseek(f,pos,SEEK_SET);
+ return r;
+}
#endif // !STBI_NO_STDIO
STBIDEF int stbi_info_from_memory(stbi_uc const *buffer, int len, int *x, int *y, int *comp)
@@ -6292,10 +7745,64 @@ STBIDEF int stbi_info_from_callbacks(stbi_io_callbacks const *c, void *user, int
return stbi__info_main(&s,x,y,comp);
}
+STBIDEF int stbi_is_16_bit_from_memory(stbi_uc const *buffer, int len)
+{
+ stbi__context s;
+ stbi__start_mem(&s,buffer,len);
+ return stbi__is_16_main(&s);
+}
+
+STBIDEF int stbi_is_16_bit_from_callbacks(stbi_io_callbacks const *c, void *user)
+{
+ stbi__context s;
+ stbi__start_callbacks(&s, (stbi_io_callbacks *) c, user);
+ return stbi__is_16_main(&s);
+}
+
#endif // STB_IMAGE_IMPLEMENTATION
/*
revision history:
+ 2.20 (2019-02-07) support utf8 filenames in Windows; fix warnings and platform ifdefs
+ 2.19 (2018-02-11) fix warning
+ 2.18 (2018-01-30) fix warnings
+ 2.17 (2018-01-29) change sbti__shiftsigned to avoid clang -O2 bug
+ 1-bit BMP
+ *_is_16_bit api
+ avoid warnings
+ 2.16 (2017-07-23) all functions have 16-bit variants;
+ STBI_NO_STDIO works again;
+ compilation fixes;
+ fix rounding in unpremultiply;
+ optimize vertical flip;
+ disable raw_len validation;
+ documentation fixes
+ 2.15 (2017-03-18) fix png-1,2,4 bug; now all Imagenet JPGs decode;
+ warning fixes; disable run-time SSE detection on gcc;
+ uniform handling of optional "return" values;
+ thread-safe initialization of zlib tables
+ 2.14 (2017-03-03) remove deprecated STBI_JPEG_OLD; fixes for Imagenet JPGs
+ 2.13 (2016-11-29) add 16-bit API, only supported for PNG right now
+ 2.12 (2016-04-02) fix typo in 2.11 PSD fix that caused crashes
+ 2.11 (2016-04-02) allocate large structures on the stack
+ remove white matting for transparent PSD
+ fix reported channel count for PNG & BMP
+ re-enable SSE2 in non-gcc 64-bit
+ support RGB-formatted JPEG
+ read 16-bit PNGs (only as 8-bit)
+ 2.10 (2016-01-22) avoid warning introduced in 2.09 by STBI_REALLOC_SIZED
+ 2.09 (2016-01-16) allow comments in PNM files
+ 16-bit-per-pixel TGA (not bit-per-component)
+ info() for TGA could break due to .hdr handling
+ info() for BMP to shares code instead of sloppy parse
+ can use STBI_REALLOC_SIZED if allocator doesn't support realloc
+ code cleanup
+ 2.08 (2015-09-13) fix to 2.07 cleanup, reading RGB PSD as RGBA
+ 2.07 (2015-09-13) fix compiler warnings
+ partial animated GIF support
+ limited 16-bpc PSD support
+ #ifdef unused functions
+ bug with < 92 byte PIC,PNM,HDR,TGA
2.06 (2015-04-19) fix bug where PSD returns wrong '*comp' value
2.05 (2015-04-19) fix bug in progressive JPEG handling, fix warning
2.04 (2015-04-15) try to re-enable SIMD on MinGW 64-bit
@@ -6436,3 +7943,46 @@ STBIDEF int stbi_info_from_callbacks(stbi_io_callbacks const *c, void *user, int
0.50 (2006-11-19)
first released version
*/
+
+
+/*
+------------------------------------------------------------------------------
+This software is available under 2 licenses -- choose whichever you prefer.
+------------------------------------------------------------------------------
+ALTERNATIVE A - MIT License
+Copyright (c) 2017 Sean Barrett
+Permission is hereby granted, free of charge, to any person obtaining a copy of
+this software and associated documentation files (the "Software"), to deal in
+the Software without restriction, including without limitation the rights to
+use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies
+of the Software, and to permit persons to whom the Software is furnished to do
+so, subject to the following conditions:
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.
+------------------------------------------------------------------------------
+ALTERNATIVE B - Public Domain (www.unlicense.org)
+This is free and unencumbered software released into the public domain.
+Anyone is free to copy, modify, publish, use, compile, sell, or distribute this
+software, either in source code form or as a compiled binary, for any purpose,
+commercial or non-commercial, and by any means.
+In jurisdictions that recognize copyright laws, the author or authors of this
+software dedicate any and all copyright interest in the software to the public
+domain. We make this dedication for the benefit of the public at large and to
+the detriment of our heirs and successors. We intend this dedication to be an
+overt act of relinquishment in perpetuity of all present and future rights to
+this software under copyright law.
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+------------------------------------------------------------------------------
+*/
diff --git a/external/include/stb_image_write.h b/external/include/stb_image_write.h
index 59052f9c..e4b32ed1 100644
--- a/external/include/stb_image_write.h
+++ b/external/include/stb_image_write.h
@@ -1,5 +1,5 @@
-/* stb_image_write - v0.98 - public domain - http://nothings.org/stb/stb_image_write.h
- writes out PNG/BMP/TGA images to C stdio - Sean Barrett 2010
+/* stb_image_write - v1.16 - public domain - http://nothings.org/stb
+ writes out PNG/BMP/TGA/JPEG/HDR images to C stdio - Sean Barrett 2010-2015
no warranty implied; use at your own risk
Before #including,
@@ -12,29 +12,67 @@
ABOUT:
- This header file is a library for writing images to C stdio. It could be
- adapted to write to memory or a general streaming interface; let me know.
+ This header file is a library for writing images to C stdio or a callback.
The PNG output is not optimal; it is 20-50% larger than the file
- written by a decent optimizing implementation. This library is designed
- for source code compactness and simplicitly, not optimal image file size
- or run-time performance.
+ written by a decent optimizing implementation; though providing a custom
+ zlib compress function (see STBIW_ZLIB_COMPRESS) can mitigate that.
+ This library is designed for source code compactness and simplicity,
+ not optimal image file size or run-time performance.
BUILDING:
You can #define STBIW_ASSERT(x) before the #include to avoid using assert.h.
You can #define STBIW_MALLOC(), STBIW_REALLOC(), and STBIW_FREE() to replace
malloc,realloc,free.
- You can define STBIW_MEMMOVE() to replace memmove()
+ You can #define STBIW_MEMMOVE() to replace memmove()
+ You can #define STBIW_ZLIB_COMPRESS to use a custom zlib-style compress function
+ for PNG compression (instead of the builtin one), it must have the following signature:
+ unsigned char * my_compress(unsigned char *data, int data_len, int *out_len, int quality);
+ The returned data will be freed with STBIW_FREE() (free() by default),
+ so it must be heap allocated with STBIW_MALLOC() (malloc() by default),
+
+UNICODE:
+
+ If compiling for Windows and you wish to use Unicode filenames, compile
+ with
+ #define STBIW_WINDOWS_UTF8
+ and pass utf8-encoded filenames. Call stbiw_convert_wchar_to_utf8 to convert
+ Windows wchar_t filenames to utf8.
USAGE:
- There are four functions, one for each image file format:
+ There are five functions, one for each image file format:
int stbi_write_png(char const *filename, int w, int h, int comp, const void *data, int stride_in_bytes);
int stbi_write_bmp(char const *filename, int w, int h, int comp, const void *data);
int stbi_write_tga(char const *filename, int w, int h, int comp, const void *data);
- int stbi_write_hdr(char const *filename, int w, int h, int comp, const void *data);
+ int stbi_write_jpg(char const *filename, int w, int h, int comp, const void *data, int quality);
+ int stbi_write_hdr(char const *filename, int w, int h, int comp, const float *data);
+
+ void stbi_flip_vertically_on_write(int flag); // flag is non-zero to flip data vertically
+
+ There are also five equivalent functions that use an arbitrary write function. You are
+ expected to open/close your file-equivalent before and after calling these:
+
+ int stbi_write_png_to_func(stbi_write_func *func, void *context, int w, int h, int comp, const void *data, int stride_in_bytes);
+ int stbi_write_bmp_to_func(stbi_write_func *func, void *context, int w, int h, int comp, const void *data);
+ int stbi_write_tga_to_func(stbi_write_func *func, void *context, int w, int h, int comp, const void *data);
+ int stbi_write_hdr_to_func(stbi_write_func *func, void *context, int w, int h, int comp, const float *data);
+ int stbi_write_jpg_to_func(stbi_write_func *func, void *context, int x, int y, int comp, const void *data, int quality);
+
+ where the callback is:
+ void stbi_write_func(void *context, void *data, int size);
+
+ You can configure it with these global variables:
+ int stbi_write_tga_with_rle; // defaults to true; set to 0 to disable RLE
+ int stbi_write_png_compression_level; // defaults to 8; set to higher for more compression
+ int stbi_write_force_png_filter; // defaults to -1; set to 0..5 to force a filter mode
+
+
+ You can define STBI_WRITE_NO_STDIO to disable the file variant of these
+ functions, so the library will not use stdio.h at all. However, this will
+ also disable HDR writing, because it requires stdio for formatted output.
Each function returns 0 on failure and non-0 on success.
@@ -58,69 +96,145 @@
writer, both because it is in BGR order and because it may have padding
at the end of the line.)
+ PNG allows you to set the deflate compression level by setting the global
+ variable 'stbi_write_png_compression_level' (it defaults to 8).
+
HDR expects linear float data. Since the format is always 32-bit rgb(e)
data, alpha (if provided) is discarded, and for monochrome data it is
replicated across all three channels.
+ TGA supports RLE or non-RLE compressed data. To use non-RLE-compressed
+ data, set the global variable 'stbi_write_tga_with_rle' to 0.
+
+ JPEG does ignore alpha channels in input data; quality is between 1 and 100.
+ Higher quality looks better but results in a bigger image.
+ JPEG baseline (no JPEG progressive).
+
CREDITS:
- PNG/BMP/TGA
- Sean Barrett
- HDR
- Baldur Karlsson
- TGA monochrome:
- Jean-Sebastien Guay
- misc enhancements:
- Tim Kelsey
+
+ Sean Barrett - PNG/BMP/TGA
+ Baldur Karlsson - HDR
+ Jean-Sebastien Guay - TGA monochrome
+ Tim Kelsey - misc enhancements
+ Alan Hickman - TGA RLE
+ Emmanuel Julien - initial file IO callback implementation
+ Jon Olick - original jo_jpeg.cpp code
+ Daniel Gibson - integrate JPEG, allow external zlib
+ Aarni Koskela - allow choosing PNG filter
+
bugfixes:
github:Chribba
-
+ Guillaume Chereau
+ github:jry2
+ github:romigrou
+ Sergio Gonzalez
+ Jonas Karlsson
+ Filip Wasil
+ Thatcher Ulrich
+ github:poppolopoppo
+ Patrick Boettcher
+ github:xeekworx
+ Cap Petschulat
+ Simon Rodriguez
+ Ivan Tikhonov
+ github:ignotion
+ Adam Schackart
+ Andrew Kensler
+
LICENSE
-This software is in the public domain. Where that dedication is not
-recognized, you are granted a perpetual, irrevocable license to copy,
-distribute, and modify this file as you see fit.
+ See end of file for license information.
+
*/
#ifndef INCLUDE_STB_IMAGE_WRITE_H
#define INCLUDE_STB_IMAGE_WRITE_H
+#include
+
+// if STB_IMAGE_WRITE_STATIC causes problems, try defining STBIWDEF to 'inline' or 'static inline'
+#ifndef STBIWDEF
+#ifdef STB_IMAGE_WRITE_STATIC
+#define STBIWDEF static
+#else
#ifdef __cplusplus
-extern "C" {
+#define STBIWDEF extern "C"
+#else
+#define STBIWDEF extern
+#endif
+#endif
#endif
-extern int stbi_write_png(char const *filename, int w, int h, int comp, const void *data, int stride_in_bytes);
-extern int stbi_write_bmp(char const *filename, int w, int h, int comp, const void *data);
-extern int stbi_write_tga(char const *filename, int w, int h, int comp, const void *data);
-extern int stbi_write_hdr(char const *filename, int w, int h, int comp, const float *data);
+#ifndef STB_IMAGE_WRITE_STATIC // C++ forbids static forward declarations
+STBIWDEF int stbi_write_tga_with_rle;
+STBIWDEF int stbi_write_png_compression_level;
+STBIWDEF int stbi_write_force_png_filter;
+#endif
-#ifdef __cplusplus
-}
+#ifndef STBI_WRITE_NO_STDIO
+STBIWDEF int stbi_write_png(char const *filename, int w, int h, int comp, const void *data, int stride_in_bytes);
+STBIWDEF int stbi_write_bmp(char const *filename, int w, int h, int comp, const void *data);
+STBIWDEF int stbi_write_tga(char const *filename, int w, int h, int comp, const void *data);
+STBIWDEF int stbi_write_hdr(char const *filename, int w, int h, int comp, const float *data);
+STBIWDEF int stbi_write_jpg(char const *filename, int x, int y, int comp, const void *data, int quality);
+
+#ifdef STBIW_WINDOWS_UTF8
+STBIWDEF int stbiw_convert_wchar_to_utf8(char *buffer, size_t bufferlen, const wchar_t* input);
+#endif
#endif
+typedef void stbi_write_func(void *context, void *data, int size);
+
+STBIWDEF int stbi_write_png_to_func(stbi_write_func *func, void *context, int w, int h, int comp, const void *data, int stride_in_bytes);
+STBIWDEF int stbi_write_bmp_to_func(stbi_write_func *func, void *context, int w, int h, int comp, const void *data);
+STBIWDEF int stbi_write_tga_to_func(stbi_write_func *func, void *context, int w, int h, int comp, const void *data);
+STBIWDEF int stbi_write_hdr_to_func(stbi_write_func *func, void *context, int w, int h, int comp, const float *data);
+STBIWDEF int stbi_write_jpg_to_func(stbi_write_func *func, void *context, int x, int y, int comp, const void *data, int quality);
+
+STBIWDEF void stbi_flip_vertically_on_write(int flip_boolean);
+
#endif//INCLUDE_STB_IMAGE_WRITE_H
#ifdef STB_IMAGE_WRITE_IMPLEMENTATION
+#ifdef _WIN32
+ #ifndef _CRT_SECURE_NO_WARNINGS
+ #define _CRT_SECURE_NO_WARNINGS
+ #endif
+ #ifndef _CRT_NONSTDC_NO_DEPRECATE
+ #define _CRT_NONSTDC_NO_DEPRECATE
+ #endif
+#endif
+
+#ifndef STBI_WRITE_NO_STDIO
+#include
+#endif // STBI_WRITE_NO_STDIO
+
#include
#include
-#include
#include
#include
-#if defined(STBIW_MALLOC) && defined(STBIW_FREE) && defined(STBIW_REALLOC)
+#if defined(STBIW_MALLOC) && defined(STBIW_FREE) && (defined(STBIW_REALLOC) || defined(STBIW_REALLOC_SIZED))
// ok
-#elif !defined(STBIW_MALLOC) && !defined(STBIW_FREE) && !defined(STBIW_REALLOC)
+#elif !defined(STBIW_MALLOC) && !defined(STBIW_FREE) && !defined(STBIW_REALLOC) && !defined(STBIW_REALLOC_SIZED)
// ok
#else
-#error "Must define all or none of STBIW_MALLOC, STBIW_FREE, and STBIW_REALLOC."
+#error "Must define all or none of STBIW_MALLOC, STBIW_FREE, and STBIW_REALLOC (or STBIW_REALLOC_SIZED)."
#endif
#ifndef STBIW_MALLOC
-#define STBIW_MALLOC(sz) malloc(sz)
-#define STBIW_REALLOC(p,sz) realloc(p,sz)
-#define STBIW_FREE(p) free(p)
+#define STBIW_MALLOC(sz) malloc(sz)
+#define STBIW_REALLOC(p,newsz) realloc(p,newsz)
+#define STBIW_FREE(p) free(p)
+#endif
+
+#ifndef STBIW_REALLOC_SIZED
+#define STBIW_REALLOC_SIZED(p,oldsz,newsz) STBIW_REALLOC(p,newsz)
#endif
+
+
#ifndef STBIW_MEMMOVE
#define STBIW_MEMMOVE(a,b,sz) memmove(a,b,sz)
#endif
@@ -131,22 +245,129 @@ extern int stbi_write_hdr(char const *filename, int w, int h, int comp, const fl
#define STBIW_ASSERT(x) assert(x)
#endif
+#define STBIW_UCHAR(x) (unsigned char) ((x) & 0xff)
+
+#ifdef STB_IMAGE_WRITE_STATIC
+static int stbi_write_png_compression_level = 8;
+static int stbi_write_tga_with_rle = 1;
+static int stbi_write_force_png_filter = -1;
+#else
+int stbi_write_png_compression_level = 8;
+int stbi_write_tga_with_rle = 1;
+int stbi_write_force_png_filter = -1;
+#endif
+
+static int stbi__flip_vertically_on_write = 0;
+
+STBIWDEF void stbi_flip_vertically_on_write(int flag)
+{
+ stbi__flip_vertically_on_write = flag;
+}
+
+typedef struct
+{
+ stbi_write_func *func;
+ void *context;
+ unsigned char buffer[64];
+ int buf_used;
+} stbi__write_context;
+
+// initialize a callback-based context
+static void stbi__start_write_callbacks(stbi__write_context *s, stbi_write_func *c, void *context)
+{
+ s->func = c;
+ s->context = context;
+}
+
+#ifndef STBI_WRITE_NO_STDIO
+
+static void stbi__stdio_write(void *context, void *data, int size)
+{
+ fwrite(data,1,size,(FILE*) context);
+}
+
+#if defined(_WIN32) && defined(STBIW_WINDOWS_UTF8)
+#ifdef __cplusplus
+#define STBIW_EXTERN extern "C"
+#else
+#define STBIW_EXTERN extern
+#endif
+STBIW_EXTERN __declspec(dllimport) int __stdcall MultiByteToWideChar(unsigned int cp, unsigned long flags, const char *str, int cbmb, wchar_t *widestr, int cchwide);
+STBIW_EXTERN __declspec(dllimport) int __stdcall WideCharToMultiByte(unsigned int cp, unsigned long flags, const wchar_t *widestr, int cchwide, char *str, int cbmb, const char *defchar, int *used_default);
+
+STBIWDEF int stbiw_convert_wchar_to_utf8(char *buffer, size_t bufferlen, const wchar_t* input)
+{
+ return WideCharToMultiByte(65001 /* UTF8 */, 0, input, -1, buffer, (int) bufferlen, NULL, NULL);
+}
+#endif
+
+static FILE *stbiw__fopen(char const *filename, char const *mode)
+{
+ FILE *f;
+#if defined(_WIN32) && defined(STBIW_WINDOWS_UTF8)
+ wchar_t wMode[64];
+ wchar_t wFilename[1024];
+ if (0 == MultiByteToWideChar(65001 /* UTF8 */, 0, filename, -1, wFilename, sizeof(wFilename)/sizeof(*wFilename)))
+ return 0;
+
+ if (0 == MultiByteToWideChar(65001 /* UTF8 */, 0, mode, -1, wMode, sizeof(wMode)/sizeof(*wMode)))
+ return 0;
+
+#if defined(_MSC_VER) && _MSC_VER >= 1400
+ if (0 != _wfopen_s(&f, wFilename, wMode))
+ f = 0;
+#else
+ f = _wfopen(wFilename, wMode);
+#endif
+
+#elif defined(_MSC_VER) && _MSC_VER >= 1400
+ if (0 != fopen_s(&f, filename, mode))
+ f=0;
+#else
+ f = fopen(filename, mode);
+#endif
+ return f;
+}
+
+static int stbi__start_write_file(stbi__write_context *s, const char *filename)
+{
+ FILE *f = stbiw__fopen(filename, "wb");
+ stbi__start_write_callbacks(s, stbi__stdio_write, (void *) f);
+ return f != NULL;
+}
+
+static void stbi__end_write_file(stbi__write_context *s)
+{
+ fclose((FILE *)s->context);
+}
+
+#endif // !STBI_WRITE_NO_STDIO
+
typedef unsigned int stbiw_uint32;
typedef int stb_image_write_test[sizeof(stbiw_uint32)==4 ? 1 : -1];
-static void writefv(FILE *f, const char *fmt, va_list v)
+static void stbiw__writefv(stbi__write_context *s, const char *fmt, va_list v)
{
while (*fmt) {
switch (*fmt++) {
case ' ': break;
- case '1': { unsigned char x = (unsigned char) va_arg(v, int); fputc(x,f); break; }
- case '2': { int x = va_arg(v,int); unsigned char b[2];
- b[0] = (unsigned char) x; b[1] = (unsigned char) (x>>8);
- fwrite(b,2,1,f); break; }
- case '4': { stbiw_uint32 x = va_arg(v,int); unsigned char b[4];
- b[0]=(unsigned char)x; b[1]=(unsigned char)(x>>8);
- b[2]=(unsigned char)(x>>16); b[3]=(unsigned char)(x>>24);
- fwrite(b,4,1,f); break; }
+ case '1': { unsigned char x = STBIW_UCHAR(va_arg(v, int));
+ s->func(s->context,&x,1);
+ break; }
+ case '2': { int x = va_arg(v,int);
+ unsigned char b[2];
+ b[0] = STBIW_UCHAR(x);
+ b[1] = STBIW_UCHAR(x>>8);
+ s->func(s->context,b,2);
+ break; }
+ case '4': { stbiw_uint32 x = va_arg(v,int);
+ unsigned char b[4];
+ b[0]=STBIW_UCHAR(x);
+ b[1]=STBIW_UCHAR(x>>8);
+ b[2]=STBIW_UCHAR(x>>16);
+ b[3]=STBIW_UCHAR(x>>24);
+ s->func(s->context,b,4);
+ break; }
default:
STBIW_ASSERT(0);
return;
@@ -154,105 +375,273 @@ static void writefv(FILE *f, const char *fmt, va_list v)
}
}
-static void write3(FILE *f, unsigned char a, unsigned char b, unsigned char c)
+static void stbiw__writef(stbi__write_context *s, const char *fmt, ...)
+{
+ va_list v;
+ va_start(v, fmt);
+ stbiw__writefv(s, fmt, v);
+ va_end(v);
+}
+
+static void stbiw__write_flush(stbi__write_context *s)
+{
+ if (s->buf_used) {
+ s->func(s->context, &s->buffer, s->buf_used);
+ s->buf_used = 0;
+ }
+}
+
+static void stbiw__putc(stbi__write_context *s, unsigned char c)
+{
+ s->func(s->context, &c, 1);
+}
+
+static void stbiw__write1(stbi__write_context *s, unsigned char a)
+{
+ if ((size_t)s->buf_used + 1 > sizeof(s->buffer))
+ stbiw__write_flush(s);
+ s->buffer[s->buf_used++] = a;
+}
+
+static void stbiw__write3(stbi__write_context *s, unsigned char a, unsigned char b, unsigned char c)
{
- unsigned char arr[3];
- arr[0] = a, arr[1] = b, arr[2] = c;
- fwrite(arr, 3, 1, f);
+ int n;
+ if ((size_t)s->buf_used + 3 > sizeof(s->buffer))
+ stbiw__write_flush(s);
+ n = s->buf_used;
+ s->buf_used = n+3;
+ s->buffer[n+0] = a;
+ s->buffer[n+1] = b;
+ s->buffer[n+2] = c;
}
-static void write_pixels(FILE *f, int rgb_dir, int vdir, int x, int y, int comp, void *data, int write_alpha, int scanline_pad, int expand_mono)
+static void stbiw__write_pixel(stbi__write_context *s, int rgb_dir, int comp, int write_alpha, int expand_mono, unsigned char *d)
{
unsigned char bg[3] = { 255, 0, 255}, px[3];
+ int k;
+
+ if (write_alpha < 0)
+ stbiw__write1(s, d[comp - 1]);
+
+ switch (comp) {
+ case 2: // 2 pixels = mono + alpha, alpha is written separately, so same as 1-channel case
+ case 1:
+ if (expand_mono)
+ stbiw__write3(s, d[0], d[0], d[0]); // monochrome bmp
+ else
+ stbiw__write1(s, d[0]); // monochrome TGA
+ break;
+ case 4:
+ if (!write_alpha) {
+ // composite against pink background
+ for (k = 0; k < 3; ++k)
+ px[k] = bg[k] + ((d[k] - bg[k]) * d[3]) / 255;
+ stbiw__write3(s, px[1 - rgb_dir], px[1], px[1 + rgb_dir]);
+ break;
+ }
+ /* FALLTHROUGH */
+ case 3:
+ stbiw__write3(s, d[1 - rgb_dir], d[1], d[1 + rgb_dir]);
+ break;
+ }
+ if (write_alpha > 0)
+ stbiw__write1(s, d[comp - 1]);
+}
+
+static void stbiw__write_pixels(stbi__write_context *s, int rgb_dir, int vdir, int x, int y, int comp, void *data, int write_alpha, int scanline_pad, int expand_mono)
+{
stbiw_uint32 zero = 0;
- int i,j,k, j_end;
+ int i,j, j_end;
if (y <= 0)
return;
- if (vdir < 0)
- j_end = -1, j = y-1;
- else
- j_end = y, j = 0;
+ if (stbi__flip_vertically_on_write)
+ vdir *= -1;
+
+ if (vdir < 0) {
+ j_end = -1; j = y-1;
+ } else {
+ j_end = y; j = 0;
+ }
for (; j != j_end; j += vdir) {
for (i=0; i < x; ++i) {
unsigned char *d = (unsigned char *) data + (j*x+i)*comp;
- if (write_alpha < 0)
- fwrite(&d[comp-1], 1, 1, f);
- switch (comp) {
- case 1: fwrite(d, 1, 1, f);
- break;
- case 2: if (expand_mono)
- write3(f, d[0],d[0],d[0]); // monochrome bmp
- else
- fwrite(d, 1, 1, f); // monochrome TGA
- break;
- case 4:
- if (!write_alpha) {
- // composite against pink background
- for (k=0; k < 3; ++k)
- px[k] = bg[k] + ((d[k] - bg[k]) * d[3])/255;
- write3(f, px[1-rgb_dir],px[1],px[1+rgb_dir]);
- break;
- }
- /* FALLTHROUGH */
- case 3:
- write3(f, d[1-rgb_dir],d[1],d[1+rgb_dir]);
- break;
- }
- if (write_alpha > 0)
- fwrite(&d[comp-1], 1, 1, f);
+ stbiw__write_pixel(s, rgb_dir, comp, write_alpha, expand_mono, d);
}
- fwrite(&zero,scanline_pad,1,f);
+ stbiw__write_flush(s);
+ s->func(s->context, &zero, scanline_pad);
}
}
-static int outfile(char const *filename, int rgb_dir, int vdir, int x, int y, int comp, int expand_mono, void *data, int alpha, int pad, const char *fmt, ...)
+static int stbiw__outfile(stbi__write_context *s, int rgb_dir, int vdir, int x, int y, int comp, int expand_mono, void *data, int alpha, int pad, const char *fmt, ...)
{
- FILE *f;
- if (y < 0 || x < 0) return 0;
- f = fopen(filename, "wb");
- if (f) {
+ if (y < 0 || x < 0) {
+ return 0;
+ } else {
va_list v;
va_start(v, fmt);
- writefv(f, fmt, v);
+ stbiw__writefv(s, fmt, v);
va_end(v);
- write_pixels(f,rgb_dir,vdir,x,y,comp,data,alpha,pad,expand_mono);
- fclose(f);
+ stbiw__write_pixels(s,rgb_dir,vdir,x,y,comp,data,alpha,pad, expand_mono);
+ return 1;
}
- return f != NULL;
}
-int stbi_write_bmp(char const *filename, int x, int y, int comp, const void *data)
+static int stbi_write_bmp_core(stbi__write_context *s, int x, int y, int comp, const void *data)
+{
+ if (comp != 4) {
+ // write RGB bitmap
+ int pad = (-x*3) & 3;
+ return stbiw__outfile(s,-1,-1,x,y,comp,1,(void *) data,0,pad,
+ "11 4 22 4" "4 44 22 444444",
+ 'B', 'M', 14+40+(x*3+pad)*y, 0,0, 14+40, // file header
+ 40, x,y, 1,24, 0,0,0,0,0,0); // bitmap header
+ } else {
+ // RGBA bitmaps need a v4 header
+ // use BI_BITFIELDS mode with 32bpp and alpha mask
+ // (straight BI_RGB with alpha mask doesn't work in most readers)
+ return stbiw__outfile(s,-1,-1,x,y,comp,1,(void *)data,1,0,
+ "11 4 22 4" "4 44 22 444444 4444 4 444 444 444 444",
+ 'B', 'M', 14+108+x*y*4, 0, 0, 14+108, // file header
+ 108, x,y, 1,32, 3,0,0,0,0,0, 0xff0000,0xff00,0xff,0xff000000u, 0, 0,0,0, 0,0,0, 0,0,0, 0,0,0); // bitmap V4 header
+ }
+}
+
+STBIWDEF int stbi_write_bmp_to_func(stbi_write_func *func, void *context, int x, int y, int comp, const void *data)
+{
+ stbi__write_context s = { 0 };
+ stbi__start_write_callbacks(&s, func, context);
+ return stbi_write_bmp_core(&s, x, y, comp, data);
+}
+
+#ifndef STBI_WRITE_NO_STDIO
+STBIWDEF int stbi_write_bmp(char const *filename, int x, int y, int comp, const void *data)
{
- int pad = (-x*3) & 3;
- return outfile(filename,-1,-1,x,y,comp,1,(void *) data,0,pad,
- "11 4 22 4" "4 44 22 444444",
- 'B', 'M', 14+40+(x*3+pad)*y, 0,0, 14+40, // file header
- 40, x,y, 1,24, 0,0,0,0,0,0); // bitmap header
+ stbi__write_context s = { 0 };
+ if (stbi__start_write_file(&s,filename)) {
+ int r = stbi_write_bmp_core(&s, x, y, comp, data);
+ stbi__end_write_file(&s);
+ return r;
+ } else
+ return 0;
}
+#endif //!STBI_WRITE_NO_STDIO
-int stbi_write_tga(char const *filename, int x, int y, int comp, const void *data)
+static int stbi_write_tga_core(stbi__write_context *s, int x, int y, int comp, void *data)
{
int has_alpha = (comp == 2 || comp == 4);
int colorbytes = has_alpha ? comp-1 : comp;
int format = colorbytes < 2 ? 3 : 2; // 3 color channels (RGB/RGBA) = 2, 1 color channel (Y/YA) = 3
- return outfile(filename, -1,-1, x, y, comp, 0, (void *) data, has_alpha, 0,
- "111 221 2222 11", 0,0,format, 0,0,0, 0,0,x,y, (colorbytes+has_alpha)*8, has_alpha*8);
+
+ if (y < 0 || x < 0)
+ return 0;
+
+ if (!stbi_write_tga_with_rle) {
+ return stbiw__outfile(s, -1, -1, x, y, comp, 0, (void *) data, has_alpha, 0,
+ "111 221 2222 11", 0, 0, format, 0, 0, 0, 0, 0, x, y, (colorbytes + has_alpha) * 8, has_alpha * 8);
+ } else {
+ int i,j,k;
+ int jend, jdir;
+
+ stbiw__writef(s, "111 221 2222 11", 0,0,format+8, 0,0,0, 0,0,x,y, (colorbytes + has_alpha) * 8, has_alpha * 8);
+
+ if (stbi__flip_vertically_on_write) {
+ j = 0;
+ jend = y;
+ jdir = 1;
+ } else {
+ j = y-1;
+ jend = -1;
+ jdir = -1;
+ }
+ for (; j != jend; j += jdir) {
+ unsigned char *row = (unsigned char *) data + j * x * comp;
+ int len;
+
+ for (i = 0; i < x; i += len) {
+ unsigned char *begin = row + i * comp;
+ int diff = 1;
+ len = 1;
+
+ if (i < x - 1) {
+ ++len;
+ diff = memcmp(begin, row + (i + 1) * comp, comp);
+ if (diff) {
+ const unsigned char *prev = begin;
+ for (k = i + 2; k < x && len < 128; ++k) {
+ if (memcmp(prev, row + k * comp, comp)) {
+ prev += comp;
+ ++len;
+ } else {
+ --len;
+ break;
+ }
+ }
+ } else {
+ for (k = i + 2; k < x && len < 128; ++k) {
+ if (!memcmp(begin, row + k * comp, comp)) {
+ ++len;
+ } else {
+ break;
+ }
+ }
+ }
+ }
+
+ if (diff) {
+ unsigned char header = STBIW_UCHAR(len - 1);
+ stbiw__write1(s, header);
+ for (k = 0; k < len; ++k) {
+ stbiw__write_pixel(s, -1, comp, has_alpha, 0, begin + k * comp);
+ }
+ } else {
+ unsigned char header = STBIW_UCHAR(len - 129);
+ stbiw__write1(s, header);
+ stbiw__write_pixel(s, -1, comp, has_alpha, 0, begin);
+ }
+ }
+ }
+ stbiw__write_flush(s);
+ }
+ return 1;
}
+STBIWDEF int stbi_write_tga_to_func(stbi_write_func *func, void *context, int x, int y, int comp, const void *data)
+{
+ stbi__write_context s = { 0 };
+ stbi__start_write_callbacks(&s, func, context);
+ return stbi_write_tga_core(&s, x, y, comp, (void *) data);
+}
+
+#ifndef STBI_WRITE_NO_STDIO
+STBIWDEF int stbi_write_tga(char const *filename, int x, int y, int comp, const void *data)
+{
+ stbi__write_context s = { 0 };
+ if (stbi__start_write_file(&s,filename)) {
+ int r = stbi_write_tga_core(&s, x, y, comp, (void *) data);
+ stbi__end_write_file(&s);
+ return r;
+ } else
+ return 0;
+}
+#endif
+
// *************************************************************************************************
// Radiance RGBE HDR writer
// by Baldur Karlsson
+
#define stbiw__max(a, b) ((a) > (b) ? (a) : (b))
-void stbiw__linear_to_rgbe(unsigned char *rgbe, float *linear)
+#ifndef STBI_WRITE_NO_STDIO
+
+static void stbiw__linear_to_rgbe(unsigned char *rgbe, float *linear)
{
int exponent;
float maxcomp = stbiw__max(linear[0], stbiw__max(linear[1], linear[2]));
- if (maxcomp < 1e-32) {
+ if (maxcomp < 1e-32f) {
rgbe[0] = rgbe[1] = rgbe[2] = rgbe[3] = 0;
} else {
float normalize = (float) frexp(maxcomp, &exponent) * 256.0f/maxcomp;
@@ -264,23 +653,23 @@ void stbiw__linear_to_rgbe(unsigned char *rgbe, float *linear)
}
}
-void stbiw__write_run_data(FILE *f, int length, unsigned char databyte)
+static void stbiw__write_run_data(stbi__write_context *s, int length, unsigned char databyte)
{
- unsigned char lengthbyte = (unsigned char) (length+128);
+ unsigned char lengthbyte = STBIW_UCHAR(length+128);
STBIW_ASSERT(length+128 <= 255);
- fwrite(&lengthbyte, 1, 1, f);
- fwrite(&databyte, 1, 1, f);
+ s->func(s->context, &lengthbyte, 1);
+ s->func(s->context, &databyte, 1);
}
-void stbiw__write_dump_data(FILE *f, int length, unsigned char *data)
+static void stbiw__write_dump_data(stbi__write_context *s, int length, unsigned char *data)
{
- unsigned char lengthbyte = (unsigned char )(length & 0xff);
+ unsigned char lengthbyte = STBIW_UCHAR(length);
STBIW_ASSERT(length <= 128); // inconsistent with spec but consistent with official code
- fwrite(&lengthbyte, 1, 1, f);
- fwrite(data, length, 1, f);
+ s->func(s->context, &lengthbyte, 1);
+ s->func(s->context, data, length);
}
-void stbiw__write_hdr_scanline(FILE *f, int width, int comp, unsigned char *scratch, const float *scanline)
+static void stbiw__write_hdr_scanline(stbi__write_context *s, int width, int ncomp, unsigned char *scratch, float *scanline)
{
unsigned char scanlineheader[4] = { 2, 2, 0, 0 };
unsigned char rgbe[4];
@@ -293,31 +682,31 @@ void stbiw__write_hdr_scanline(FILE *f, int width, int comp, unsigned char *scra
/* skip RLE for images too small or large */
if (width < 8 || width >= 32768) {
for (x=0; x < width; x++) {
- switch (comp) {
+ switch (ncomp) {
case 4: /* fallthrough */
- case 3: linear[2] = scanline[x*comp + 2];
- linear[1] = scanline[x*comp + 1];
- linear[0] = scanline[x*comp + 0];
+ case 3: linear[2] = scanline[x*ncomp + 2];
+ linear[1] = scanline[x*ncomp + 1];
+ linear[0] = scanline[x*ncomp + 0];
break;
- case 2: /* fallthrough */
- case 1: linear[0] = linear[1] = linear[2] = scanline[x*comp + 0];
+ default:
+ linear[0] = linear[1] = linear[2] = scanline[x*ncomp + 0];
break;
}
stbiw__linear_to_rgbe(rgbe, linear);
- fwrite(rgbe, 4, 1, f);
+ s->func(s->context, rgbe, 4);
}
} else {
int c,r;
/* encode into scratch buffer */
for (x=0; x < width; x++) {
- switch(comp) {
+ switch(ncomp) {
case 4: /* fallthrough */
- case 3: linear[2] = scanline[x*comp + 2];
- linear[1] = scanline[x*comp + 1];
- linear[0] = scanline[x*comp + 0];
+ case 3: linear[2] = scanline[x*ncomp + 2];
+ linear[1] = scanline[x*ncomp + 1];
+ linear[0] = scanline[x*ncomp + 0];
break;
- case 2: /* fallthrough */
- case 1: linear[0] = linear[1] = linear[2] = scanline[x*comp + 0];
+ default:
+ linear[0] = linear[1] = linear[2] = scanline[x*ncomp + 0];
break;
}
stbiw__linear_to_rgbe(rgbe, linear);
@@ -327,7 +716,7 @@ void stbiw__write_hdr_scanline(FILE *f, int width, int comp, unsigned char *scra
scratch[x + width*3] = rgbe[3];
}
- fwrite(scanlineheader, 4, 1, f);
+ s->func(s->context, scanlineheader, 4);
/* RLE each component separately */
for (c=0; c < 4; c++) {
@@ -348,7 +737,7 @@ void stbiw__write_hdr_scanline(FILE *f, int width, int comp, unsigned char *scra
while (x < r) {
int len = r-x;
if (len > 128) len = 128;
- stbiw__write_dump_data(f, len, &comp[x]);
+ stbiw__write_dump_data(s, len, &comp[x]);
x += len;
}
// if there's a run, output it
@@ -360,7 +749,7 @@ void stbiw__write_hdr_scanline(FILE *f, int width, int comp, unsigned char *scra
while (x < r) {
int len = r-x;
if (len > 127) len = 127;
- stbiw__write_run_data(f, len, comp[x]);
+ stbiw__write_run_data(s, len, comp[x]);
x += len;
}
}
@@ -369,30 +758,60 @@ void stbiw__write_hdr_scanline(FILE *f, int width, int comp, unsigned char *scra
}
}
-int stbi_write_hdr(char const *filename, int x, int y, int comp, const float *data)
+static int stbi_write_hdr_core(stbi__write_context *s, int x, int y, int comp, float *data)
{
- int i;
- FILE *f;
- if (y <= 0 || x <= 0 || data == NULL) return 0;
- f = fopen(filename, "wb");
- if (f) {
- /* Each component is stored separately. Allocate scratch space for full output scanline. */
+ if (y <= 0 || x <= 0 || data == NULL)
+ return 0;
+ else {
+ // Each component is stored separately. Allocate scratch space for full output scanline.
unsigned char *scratch = (unsigned char *) STBIW_MALLOC(x*4);
- fprintf(f, "#?RADIANCE\n# Written by stb_image_write.h\nFORMAT=32-bit_rle_rgbe\n" );
- fprintf(f, "EXPOSURE= 1.0000000000000\n\n-Y %d +X %d\n" , y, x);
+ int i, len;
+ char buffer[128];
+ char header[] = "#?RADIANCE\n# Written by stb_image_write.h\nFORMAT=32-bit_rle_rgbe\n";
+ s->func(s->context, header, sizeof(header)-1);
+
+#ifdef __STDC_LIB_EXT1__
+ len = sprintf_s(buffer, sizeof(buffer), "EXPOSURE= 1.0000000000000\n\n-Y %d +X %d\n", y, x);
+#else
+ len = sprintf(buffer, "EXPOSURE= 1.0000000000000\n\n-Y %d +X %d\n", y, x);
+#endif
+ s->func(s->context, buffer, len);
+
for(i=0; i < y; i++)
- stbiw__write_hdr_scanline(f, x, comp, scratch, data + comp*i*x);
+ stbiw__write_hdr_scanline(s, x, comp, scratch, data + comp*x*(stbi__flip_vertically_on_write ? y-1-i : i));
STBIW_FREE(scratch);
- fclose(f);
+ return 1;
}
- return f != NULL;
}
-/////////////////////////////////////////////////////////
-// PNG
+STBIWDEF int stbi_write_hdr_to_func(stbi_write_func *func, void *context, int x, int y, int comp, const float *data)
+{
+ stbi__write_context s = { 0 };
+ stbi__start_write_callbacks(&s, func, context);
+ return stbi_write_hdr_core(&s, x, y, comp, (float *) data);
+}
+
+STBIWDEF int stbi_write_hdr(char const *filename, int x, int y, int comp, const float *data)
+{
+ stbi__write_context s = { 0 };
+ if (stbi__start_write_file(&s,filename)) {
+ int r = stbi_write_hdr_core(&s, x, y, comp, (float *) data);
+ stbi__end_write_file(&s);
+ return r;
+ } else
+ return 0;
+}
+#endif // STBI_WRITE_NO_STDIO
+
+
+//////////////////////////////////////////////////////////////////////////////
+//
+// PNG writer
+//
+#ifndef STBIW_ZLIB_COMPRESS
// stretchy buffer; stbiw__sbpush() == vector<>::push_back() -- stbiw__sbcount() == vector<>::size()
-#define stbiw__sbraw(a) ((int *) (a) - 2)
+#define stbiw__sbraw(a) ((int *) (void *) (a) - 2)
#define stbiw__sbm(a) stbiw__sbraw(a)[0]
#define stbiw__sbn(a) stbiw__sbraw(a)[1]
@@ -407,7 +826,7 @@ int stbi_write_hdr(char const *filename, int x, int y, int comp, const float *da
static void *stbiw__sbgrowf(void **arr, int increment, int itemsize)
{
int m = *arr ? 2*stbiw__sbm(*arr)+increment : increment+1;
- void *p = STBIW_REALLOC(*arr ? stbiw__sbraw(*arr) : 0, itemsize * m + sizeof(int)*2);
+ void *p = STBIW_REALLOC_SIZED(*arr ? stbiw__sbraw(*arr) : 0, *arr ? (stbiw__sbm(*arr)*itemsize + sizeof(int)*2) : 0, itemsize * m + sizeof(int)*2);
STBIW_ASSERT(p);
if (p) {
if (!*arr) ((int *) p)[1] = 0;
@@ -420,7 +839,7 @@ static void *stbiw__sbgrowf(void **arr, int increment, int itemsize)
static unsigned char *stbiw__zlib_flushf(unsigned char *data, unsigned int *bitbuffer, int *bitcount)
{
while (*bitcount >= 8) {
- stbiw__sbpush(data, (unsigned char) *bitbuffer);
+ stbiw__sbpush(data, STBIW_UCHAR(*bitbuffer));
*bitbuffer >>= 8;
*bitcount -= 8;
}
@@ -471,8 +890,14 @@ static unsigned int stbiw__zhash(unsigned char *data)
#define stbiw__ZHASH 16384
-unsigned char * stbi_zlib_compress(unsigned char *data, int data_len, int *out_len, int quality)
+#endif // STBIW_ZLIB_COMPRESS
+
+STBIWDEF unsigned char * stbi_zlib_compress(unsigned char *data, int data_len, int *out_len, int quality)
{
+#ifdef STBIW_ZLIB_COMPRESS
+ // user provided a zlib compress implementation, use that
+ return STBIW_ZLIB_COMPRESS(data, data_len, out_len, quality);
+#else // use builtin
static unsigned short lengthc[] = { 3,4,5,6,7,8,9,10,11,13,15,17,19,23,27,31,35,43,51,59,67,83,99,115,131,163,195,227,258, 259 };
static unsigned char lengtheb[]= { 0,0,0,0,0,0,0, 0, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 4, 4, 4, 4, 5, 5, 5, 5, 0 };
static unsigned short distc[] = { 1,2,3,4,5,7,9,13,17,25,33,49,65,97,129,193,257,385,513,769,1025,1537,2049,3073,4097,6145,8193,12289,16385,24577, 32768 };
@@ -480,7 +905,9 @@ unsigned char * stbi_zlib_compress(unsigned char *data, int data_len, int *out_l
unsigned int bitbuf=0;
int i,j, bitcount=0;
unsigned char *out = NULL;
- unsigned char **hash_table[stbiw__ZHASH]; // 64KB on the stack!
+ unsigned char ***hash_table = (unsigned char***) STBIW_MALLOC(stbiw__ZHASH * sizeof(unsigned char**));
+ if (hash_table == NULL)
+ return NULL;
if (quality < 5) quality = 5;
stbiw__sbpush(out, 0x78); // DEFLATE 32K window
@@ -501,7 +928,7 @@ unsigned char * stbi_zlib_compress(unsigned char *data, int data_len, int *out_l
for (j=0; j < n; ++j) {
if (hlist[j]-data > i-32768) { // if entry lies within window
int d = stbiw__zlib_countm(hlist[j], data+i, data_len-i);
- if (d >= best) best=d,bestloc=hlist[j];
+ if (d >= best) { best=d; bestloc=hlist[j]; }
}
}
// when hash table entry is too long, delete half the entries
@@ -552,43 +979,98 @@ unsigned char * stbi_zlib_compress(unsigned char *data, int data_len, int *out_l
for (i=0; i < stbiw__ZHASH; ++i)
(void) stbiw__sbfree(hash_table[i]);
+ STBIW_FREE(hash_table);
+
+ // store uncompressed instead if compression was worse
+ if (stbiw__sbn(out) > data_len + 2 + ((data_len+32766)/32767)*5) {
+ stbiw__sbn(out) = 2; // truncate to DEFLATE 32K window and FLEVEL = 1
+ for (j = 0; j < data_len;) {
+ int blocklen = data_len - j;
+ if (blocklen > 32767) blocklen = 32767;
+ stbiw__sbpush(out, data_len - j == blocklen); // BFINAL = ?, BTYPE = 0 -- no compression
+ stbiw__sbpush(out, STBIW_UCHAR(blocklen)); // LEN
+ stbiw__sbpush(out, STBIW_UCHAR(blocklen >> 8));
+ stbiw__sbpush(out, STBIW_UCHAR(~blocklen)); // NLEN
+ stbiw__sbpush(out, STBIW_UCHAR(~blocklen >> 8));
+ memcpy(out+stbiw__sbn(out), data+j, blocklen);
+ stbiw__sbn(out) += blocklen;
+ j += blocklen;
+ }
+ }
{
// compute adler32 on input
- unsigned int i=0, s1=1, s2=0, blocklen = data_len % 5552;
- int j=0;
+ unsigned int s1=1, s2=0;
+ int blocklen = (int) (data_len % 5552);
+ j=0;
while (j < data_len) {
- for (i=0; i < blocklen; ++i) s1 += data[j+i], s2 += s1;
- s1 %= 65521, s2 %= 65521;
+ for (i=0; i < blocklen; ++i) { s1 += data[j+i]; s2 += s1; }
+ s1 %= 65521; s2 %= 65521;
j += blocklen;
blocklen = 5552;
}
- stbiw__sbpush(out, (unsigned char) (s2 >> 8));
- stbiw__sbpush(out, (unsigned char) s2);
- stbiw__sbpush(out, (unsigned char) (s1 >> 8));
- stbiw__sbpush(out, (unsigned char) s1);
+ stbiw__sbpush(out, STBIW_UCHAR(s2 >> 8));
+ stbiw__sbpush(out, STBIW_UCHAR(s2));
+ stbiw__sbpush(out, STBIW_UCHAR(s1 >> 8));
+ stbiw__sbpush(out, STBIW_UCHAR(s1));
}
*out_len = stbiw__sbn(out);
// make returned pointer freeable
STBIW_MEMMOVE(stbiw__sbraw(out), out, *out_len);
return (unsigned char *) stbiw__sbraw(out);
+#endif // STBIW_ZLIB_COMPRESS
}
-unsigned int stbiw__crc32(unsigned char *buffer, int len)
+static unsigned int stbiw__crc32(unsigned char *buffer, int len)
{
- static unsigned int crc_table[256];
+#ifdef STBIW_CRC32
+ return STBIW_CRC32(buffer, len);
+#else
+ static unsigned int crc_table[256] =
+ {
+ 0x00000000, 0x77073096, 0xEE0E612C, 0x990951BA, 0x076DC419, 0x706AF48F, 0xE963A535, 0x9E6495A3,
+ 0x0eDB8832, 0x79DCB8A4, 0xE0D5E91E, 0x97D2D988, 0x09B64C2B, 0x7EB17CBD, 0xE7B82D07, 0x90BF1D91,
+ 0x1DB71064, 0x6AB020F2, 0xF3B97148, 0x84BE41DE, 0x1ADAD47D, 0x6DDDE4EB, 0xF4D4B551, 0x83D385C7,
+ 0x136C9856, 0x646BA8C0, 0xFD62F97A, 0x8A65C9EC, 0x14015C4F, 0x63066CD9, 0xFA0F3D63, 0x8D080DF5,
+ 0x3B6E20C8, 0x4C69105E, 0xD56041E4, 0xA2677172, 0x3C03E4D1, 0x4B04D447, 0xD20D85FD, 0xA50AB56B,
+ 0x35B5A8FA, 0x42B2986C, 0xDBBBC9D6, 0xACBCF940, 0x32D86CE3, 0x45DF5C75, 0xDCD60DCF, 0xABD13D59,
+ 0x26D930AC, 0x51DE003A, 0xC8D75180, 0xBFD06116, 0x21B4F4B5, 0x56B3C423, 0xCFBA9599, 0xB8BDA50F,
+ 0x2802B89E, 0x5F058808, 0xC60CD9B2, 0xB10BE924, 0x2F6F7C87, 0x58684C11, 0xC1611DAB, 0xB6662D3D,
+ 0x76DC4190, 0x01DB7106, 0x98D220BC, 0xEFD5102A, 0x71B18589, 0x06B6B51F, 0x9FBFE4A5, 0xE8B8D433,
+ 0x7807C9A2, 0x0F00F934, 0x9609A88E, 0xE10E9818, 0x7F6A0DBB, 0x086D3D2D, 0x91646C97, 0xE6635C01,
+ 0x6B6B51F4, 0x1C6C6162, 0x856530D8, 0xF262004E, 0x6C0695ED, 0x1B01A57B, 0x8208F4C1, 0xF50FC457,
+ 0x65B0D9C6, 0x12B7E950, 0x8BBEB8EA, 0xFCB9887C, 0x62DD1DDF, 0x15DA2D49, 0x8CD37CF3, 0xFBD44C65,
+ 0x4DB26158, 0x3AB551CE, 0xA3BC0074, 0xD4BB30E2, 0x4ADFA541, 0x3DD895D7, 0xA4D1C46D, 0xD3D6F4FB,
+ 0x4369E96A, 0x346ED9FC, 0xAD678846, 0xDA60B8D0, 0x44042D73, 0x33031DE5, 0xAA0A4C5F, 0xDD0D7CC9,
+ 0x5005713C, 0x270241AA, 0xBE0B1010, 0xC90C2086, 0x5768B525, 0x206F85B3, 0xB966D409, 0xCE61E49F,
+ 0x5EDEF90E, 0x29D9C998, 0xB0D09822, 0xC7D7A8B4, 0x59B33D17, 0x2EB40D81, 0xB7BD5C3B, 0xC0BA6CAD,
+ 0xEDB88320, 0x9ABFB3B6, 0x03B6E20C, 0x74B1D29A, 0xEAD54739, 0x9DD277AF, 0x04DB2615, 0x73DC1683,
+ 0xE3630B12, 0x94643B84, 0x0D6D6A3E, 0x7A6A5AA8, 0xE40ECF0B, 0x9309FF9D, 0x0A00AE27, 0x7D079EB1,
+ 0xF00F9344, 0x8708A3D2, 0x1E01F268, 0x6906C2FE, 0xF762575D, 0x806567CB, 0x196C3671, 0x6E6B06E7,
+ 0xFED41B76, 0x89D32BE0, 0x10DA7A5A, 0x67DD4ACC, 0xF9B9DF6F, 0x8EBEEFF9, 0x17B7BE43, 0x60B08ED5,
+ 0xD6D6A3E8, 0xA1D1937E, 0x38D8C2C4, 0x4FDFF252, 0xD1BB67F1, 0xA6BC5767, 0x3FB506DD, 0x48B2364B,
+ 0xD80D2BDA, 0xAF0A1B4C, 0x36034AF6, 0x41047A60, 0xDF60EFC3, 0xA867DF55, 0x316E8EEF, 0x4669BE79,
+ 0xCB61B38C, 0xBC66831A, 0x256FD2A0, 0x5268E236, 0xCC0C7795, 0xBB0B4703, 0x220216B9, 0x5505262F,
+ 0xC5BA3BBE, 0xB2BD0B28, 0x2BB45A92, 0x5CB36A04, 0xC2D7FFA7, 0xB5D0CF31, 0x2CD99E8B, 0x5BDEAE1D,
+ 0x9B64C2B0, 0xEC63F226, 0x756AA39C, 0x026D930A, 0x9C0906A9, 0xEB0E363F, 0x72076785, 0x05005713,
+ 0x95BF4A82, 0xE2B87A14, 0x7BB12BAE, 0x0CB61B38, 0x92D28E9B, 0xE5D5BE0D, 0x7CDCEFB7, 0x0BDBDF21,
+ 0x86D3D2D4, 0xF1D4E242, 0x68DDB3F8, 0x1FDA836E, 0x81BE16CD, 0xF6B9265B, 0x6FB077E1, 0x18B74777,
+ 0x88085AE6, 0xFF0F6A70, 0x66063BCA, 0x11010B5C, 0x8F659EFF, 0xF862AE69, 0x616BFFD3, 0x166CCF45,
+ 0xA00AE278, 0xD70DD2EE, 0x4E048354, 0x3903B3C2, 0xA7672661, 0xD06016F7, 0x4969474D, 0x3E6E77DB,
+ 0xAED16A4A, 0xD9D65ADC, 0x40DF0B66, 0x37D83BF0, 0xA9BCAE53, 0xDEBB9EC5, 0x47B2CF7F, 0x30B5FFE9,
+ 0xBDBDF21C, 0xCABAC28A, 0x53B39330, 0x24B4A3A6, 0xBAD03605, 0xCDD70693, 0x54DE5729, 0x23D967BF,
+ 0xB3667A2E, 0xC4614AB8, 0x5D681B02, 0x2A6F2B94, 0xB40BBE37, 0xC30C8EA1, 0x5A05DF1B, 0x2D02EF8D
+ };
+
unsigned int crc = ~0u;
- int i,j;
- if (crc_table[1] == 0)
- for(i=0; i < 256; i++)
- for (crc_table[i]=i, j=0; j < 8; ++j)
- crc_table[i] = (crc_table[i] >> 1) ^ (crc_table[i] & 1 ? 0xedb88320 : 0);
+ int i;
for (i=0; i < len; ++i)
crc = (crc >> 8) ^ crc_table[buffer[i] ^ (crc & 0xff)];
return ~crc;
+#endif
}
-#define stbiw__wpng4(o,a,b,c,d) ((o)[0]=(unsigned char)(a),(o)[1]=(unsigned char)(b),(o)[2]=(unsigned char)(c),(o)[3]=(unsigned char)(d),(o)+=4)
+#define stbiw__wpng4(o,a,b,c,d) ((o)[0]=STBIW_UCHAR(a),(o)[1]=STBIW_UCHAR(b),(o)[2]=STBIW_UCHAR(c),(o)[3]=STBIW_UCHAR(d),(o)+=4)
#define stbiw__wp32(data,v) stbiw__wpng4(data, (v)>>24,(v)>>16,(v)>>8,(v));
#define stbiw__wptag(data,s) stbiw__wpng4(data, s[0],s[1],s[2],s[3])
@@ -601,66 +1083,97 @@ static void stbiw__wpcrc(unsigned char **data, int len)
static unsigned char stbiw__paeth(int a, int b, int c)
{
int p = a + b - c, pa = abs(p-a), pb = abs(p-b), pc = abs(p-c);
- if (pa <= pb && pa <= pc) return (unsigned char) a;
- if (pb <= pc) return (unsigned char) b;
- return (unsigned char) c;
+ if (pa <= pb && pa <= pc) return STBIW_UCHAR(a);
+ if (pb <= pc) return STBIW_UCHAR(b);
+ return STBIW_UCHAR(c);
+}
+
+// @OPTIMIZE: provide an option that always forces left-predict or paeth predict
+static void stbiw__encode_png_line(unsigned char *pixels, int stride_bytes, int width, int height, int y, int n, int filter_type, signed char *line_buffer)
+{
+ static int mapping[] = { 0,1,2,3,4 };
+ static int firstmap[] = { 0,1,0,5,6 };
+ int *mymap = (y != 0) ? mapping : firstmap;
+ int i;
+ int type = mymap[filter_type];
+ unsigned char *z = pixels + stride_bytes * (stbi__flip_vertically_on_write ? height-1-y : y);
+ int signed_stride = stbi__flip_vertically_on_write ? -stride_bytes : stride_bytes;
+
+ if (type==0) {
+ memcpy(line_buffer, z, width*n);
+ return;
+ }
+
+ // first loop isn't optimized since it's just one pixel
+ for (i = 0; i < n; ++i) {
+ switch (type) {
+ case 1: line_buffer[i] = z[i]; break;
+ case 2: line_buffer[i] = z[i] - z[i-signed_stride]; break;
+ case 3: line_buffer[i] = z[i] - (z[i-signed_stride]>>1); break;
+ case 4: line_buffer[i] = (signed char) (z[i] - stbiw__paeth(0,z[i-signed_stride],0)); break;
+ case 5: line_buffer[i] = z[i]; break;
+ case 6: line_buffer[i] = z[i]; break;
+ }
+ }
+ switch (type) {
+ case 1: for (i=n; i < width*n; ++i) line_buffer[i] = z[i] - z[i-n]; break;
+ case 2: for (i=n; i < width*n; ++i) line_buffer[i] = z[i] - z[i-signed_stride]; break;
+ case 3: for (i=n; i < width*n; ++i) line_buffer[i] = z[i] - ((z[i-n] + z[i-signed_stride])>>1); break;
+ case 4: for (i=n; i < width*n; ++i) line_buffer[i] = z[i] - stbiw__paeth(z[i-n], z[i-signed_stride], z[i-signed_stride-n]); break;
+ case 5: for (i=n; i < width*n; ++i) line_buffer[i] = z[i] - (z[i-n]>>1); break;
+ case 6: for (i=n; i < width*n; ++i) line_buffer[i] = z[i] - stbiw__paeth(z[i-n], 0,0); break;
+ }
}
-unsigned char *stbi_write_png_to_mem(unsigned char *pixels, int stride_bytes, int x, int y, int n, int *out_len)
+STBIWDEF unsigned char *stbi_write_png_to_mem(const unsigned char *pixels, int stride_bytes, int x, int y, int n, int *out_len)
{
+ int force_filter = stbi_write_force_png_filter;
int ctype[5] = { -1, 0, 4, 2, 6 };
unsigned char sig[8] = { 137,80,78,71,13,10,26,10 };
unsigned char *out,*o, *filt, *zlib;
signed char *line_buffer;
- int i,j,k,p,zlen;
+ int j,zlen;
if (stride_bytes == 0)
stride_bytes = x * n;
+ if (force_filter >= 5) {
+ force_filter = -1;
+ }
+
filt = (unsigned char *) STBIW_MALLOC((x*n+1) * y); if (!filt) return 0;
line_buffer = (signed char *) STBIW_MALLOC(x * n); if (!line_buffer) { STBIW_FREE(filt); return 0; }
for (j=0; j < y; ++j) {
- static int mapping[] = { 0,1,2,3,4 };
- static int firstmap[] = { 0,1,0,5,6 };
- int *mymap = j ? mapping : firstmap;
- int best = 0, bestval = 0x7fffffff;
- for (p=0; p < 2; ++p) {
- for (k= p?best:0; k < 5; ++k) {
- int type = mymap[k],est=0;
- unsigned char *z = pixels + stride_bytes*j;
- for (i=0; i < n; ++i)
- switch (type) {
- case 0: line_buffer[i] = z[i]; break;
- case 1: line_buffer[i] = z[i]; break;
- case 2: line_buffer[i] = z[i] - z[i-stride_bytes]; break;
- case 3: line_buffer[i] = z[i] - (z[i-stride_bytes]>>1); break;
- case 4: line_buffer[i] = (signed char) (z[i] - stbiw__paeth(0,z[i-stride_bytes],0)); break;
- case 5: line_buffer[i] = z[i]; break;
- case 6: line_buffer[i] = z[i]; break;
- }
- for (i=n; i < x*n; ++i) {
- switch (type) {
- case 0: line_buffer[i] = z[i]; break;
- case 1: line_buffer[i] = z[i] - z[i-n]; break;
- case 2: line_buffer[i] = z[i] - z[i-stride_bytes]; break;
- case 3: line_buffer[i] = z[i] - ((z[i-n] + z[i-stride_bytes])>>1); break;
- case 4: line_buffer[i] = z[i] - stbiw__paeth(z[i-n], z[i-stride_bytes], z[i-stride_bytes-n]); break;
- case 5: line_buffer[i] = z[i] - (z[i-n]>>1); break;
- case 6: line_buffer[i] = z[i] - stbiw__paeth(z[i-n], 0,0); break;
- }
- }
- if (p) break;
- for (i=0; i < x*n; ++i)
+ int filter_type;
+ if (force_filter > -1) {
+ filter_type = force_filter;
+ stbiw__encode_png_line((unsigned char*)(pixels), stride_bytes, x, y, j, n, force_filter, line_buffer);
+ } else { // Estimate the best filter by running through all of them:
+ int best_filter = 0, best_filter_val = 0x7fffffff, est, i;
+ for (filter_type = 0; filter_type < 5; filter_type++) {
+ stbiw__encode_png_line((unsigned char*)(pixels), stride_bytes, x, y, j, n, filter_type, line_buffer);
+
+ // Estimate the entropy of the line using this filter; the less, the better.
+ est = 0;
+ for (i = 0; i < x*n; ++i) {
est += abs((signed char) line_buffer[i]);
- if (est < bestval) { bestval = est; best = k; }
+ }
+ if (est < best_filter_val) {
+ best_filter_val = est;
+ best_filter = filter_type;
+ }
+ }
+ if (filter_type != best_filter) { // If the last iteration already got us the best filter, don't redo it
+ stbiw__encode_png_line((unsigned char*)(pixels), stride_bytes, x, y, j, n, best_filter, line_buffer);
+ filter_type = best_filter;
}
}
- // when we get here, best contains the filter type, and line_buffer contains the data
- filt[j*(x*n+1)] = (unsigned char) best;
+ // when we get here, filter_type contains the filter type, and line_buffer contains the data
+ filt[j*(x*n+1)] = (unsigned char) filter_type;
STBIW_MEMMOVE(filt+j*(x*n+1)+1, line_buffer, x*n);
}
STBIW_FREE(line_buffer);
- zlib = stbi_zlib_compress(filt, y*( x*n+1), &zlen, 8); // increase 8 to get smaller but use more memory
+ zlib = stbi_zlib_compress(filt, y*( x*n+1), &zlen, stbi_write_png_compression_level);
STBIW_FREE(filt);
if (!zlib) return 0;
@@ -676,7 +1189,7 @@ unsigned char *stbi_write_png_to_mem(unsigned char *pixels, int stride_bytes, in
stbiw__wp32(o, x);
stbiw__wp32(o, y);
*o++ = 8;
- *o++ = (unsigned char) ctype[n];
+ *o++ = STBIW_UCHAR(ctype[n]);
*o++ = 0;
*o++ = 0;
*o++ = 0;
@@ -698,22 +1211,456 @@ unsigned char *stbi_write_png_to_mem(unsigned char *pixels, int stride_bytes, in
return out;
}
-int stbi_write_png(char const *filename, int x, int y, int comp, const void *data, int stride_bytes)
+#ifndef STBI_WRITE_NO_STDIO
+STBIWDEF int stbi_write_png(char const *filename, int x, int y, int comp, const void *data, int stride_bytes)
{
FILE *f;
int len;
- unsigned char *png = stbi_write_png_to_mem((unsigned char *) data, stride_bytes, x, y, comp, &len);
- if (!png) return 0;
- f = fopen(filename, "wb");
+ unsigned char *png = stbi_write_png_to_mem((const unsigned char *) data, stride_bytes, x, y, comp, &len);
+ if (png == NULL) return 0;
+
+ f = stbiw__fopen(filename, "wb");
if (!f) { STBIW_FREE(png); return 0; }
fwrite(png, 1, len, f);
fclose(f);
STBIW_FREE(png);
return 1;
}
+#endif
+
+STBIWDEF int stbi_write_png_to_func(stbi_write_func *func, void *context, int x, int y, int comp, const void *data, int stride_bytes)
+{
+ int len;
+ unsigned char *png = stbi_write_png_to_mem((const unsigned char *) data, stride_bytes, x, y, comp, &len);
+ if (png == NULL) return 0;
+ func(context, png, len);
+ STBIW_FREE(png);
+ return 1;
+}
+
+
+/* ***************************************************************************
+ *
+ * JPEG writer
+ *
+ * This is based on Jon Olick's jo_jpeg.cpp:
+ * public domain Simple, Minimalistic JPEG writer - http://www.jonolick.com/code.html
+ */
+
+static const unsigned char stbiw__jpg_ZigZag[] = { 0,1,5,6,14,15,27,28,2,4,7,13,16,26,29,42,3,8,12,17,25,30,41,43,9,11,18,
+ 24,31,40,44,53,10,19,23,32,39,45,52,54,20,22,33,38,46,51,55,60,21,34,37,47,50,56,59,61,35,36,48,49,57,58,62,63 };
+
+static void stbiw__jpg_writeBits(stbi__write_context *s, int *bitBufP, int *bitCntP, const unsigned short *bs) {
+ int bitBuf = *bitBufP, bitCnt = *bitCntP;
+ bitCnt += bs[1];
+ bitBuf |= bs[0] << (24 - bitCnt);
+ while(bitCnt >= 8) {
+ unsigned char c = (bitBuf >> 16) & 255;
+ stbiw__putc(s, c);
+ if(c == 255) {
+ stbiw__putc(s, 0);
+ }
+ bitBuf <<= 8;
+ bitCnt -= 8;
+ }
+ *bitBufP = bitBuf;
+ *bitCntP = bitCnt;
+}
+
+static void stbiw__jpg_DCT(float *d0p, float *d1p, float *d2p, float *d3p, float *d4p, float *d5p, float *d6p, float *d7p) {
+ float d0 = *d0p, d1 = *d1p, d2 = *d2p, d3 = *d3p, d4 = *d4p, d5 = *d5p, d6 = *d6p, d7 = *d7p;
+ float z1, z2, z3, z4, z5, z11, z13;
+
+ float tmp0 = d0 + d7;
+ float tmp7 = d0 - d7;
+ float tmp1 = d1 + d6;
+ float tmp6 = d1 - d6;
+ float tmp2 = d2 + d5;
+ float tmp5 = d2 - d5;
+ float tmp3 = d3 + d4;
+ float tmp4 = d3 - d4;
+
+ // Even part
+ float tmp10 = tmp0 + tmp3; // phase 2
+ float tmp13 = tmp0 - tmp3;
+ float tmp11 = tmp1 + tmp2;
+ float tmp12 = tmp1 - tmp2;
+
+ d0 = tmp10 + tmp11; // phase 3
+ d4 = tmp10 - tmp11;
+
+ z1 = (tmp12 + tmp13) * 0.707106781f; // c4
+ d2 = tmp13 + z1; // phase 5
+ d6 = tmp13 - z1;
+
+ // Odd part
+ tmp10 = tmp4 + tmp5; // phase 2
+ tmp11 = tmp5 + tmp6;
+ tmp12 = tmp6 + tmp7;
+
+ // The rotator is modified from fig 4-8 to avoid extra negations.
+ z5 = (tmp10 - tmp12) * 0.382683433f; // c6
+ z2 = tmp10 * 0.541196100f + z5; // c2-c6
+ z4 = tmp12 * 1.306562965f + z5; // c2+c6
+ z3 = tmp11 * 0.707106781f; // c4
+
+ z11 = tmp7 + z3; // phase 5
+ z13 = tmp7 - z3;
+
+ *d5p = z13 + z2; // phase 6
+ *d3p = z13 - z2;
+ *d1p = z11 + z4;
+ *d7p = z11 - z4;
+
+ *d0p = d0; *d2p = d2; *d4p = d4; *d6p = d6;
+}
+
+static void stbiw__jpg_calcBits(int val, unsigned short bits[2]) {
+ int tmp1 = val < 0 ? -val : val;
+ val = val < 0 ? val-1 : val;
+ bits[1] = 1;
+ while(tmp1 >>= 1) {
+ ++bits[1];
+ }
+ bits[0] = val & ((1<0)&&(DU[end0pos]==0); --end0pos) {
+ }
+ // end0pos = first element in reverse order !=0
+ if(end0pos == 0) {
+ stbiw__jpg_writeBits(s, bitBuf, bitCnt, EOB);
+ return DU[0];
+ }
+ for(i = 1; i <= end0pos; ++i) {
+ int startpos = i;
+ int nrzeroes;
+ unsigned short bits[2];
+ for (; DU[i]==0 && i<=end0pos; ++i) {
+ }
+ nrzeroes = i-startpos;
+ if ( nrzeroes >= 16 ) {
+ int lng = nrzeroes>>4;
+ int nrmarker;
+ for (nrmarker=1; nrmarker <= lng; ++nrmarker)
+ stbiw__jpg_writeBits(s, bitBuf, bitCnt, M16zeroes);
+ nrzeroes &= 15;
+ }
+ stbiw__jpg_calcBits(DU[i], bits);
+ stbiw__jpg_writeBits(s, bitBuf, bitCnt, HTAC[(nrzeroes<<4)+bits[1]]);
+ stbiw__jpg_writeBits(s, bitBuf, bitCnt, bits);
+ }
+ if(end0pos != 63) {
+ stbiw__jpg_writeBits(s, bitBuf, bitCnt, EOB);
+ }
+ return DU[0];
+}
+
+static int stbi_write_jpg_core(stbi__write_context *s, int width, int height, int comp, const void* data, int quality) {
+ // Constants that don't pollute global namespace
+ static const unsigned char std_dc_luminance_nrcodes[] = {0,0,1,5,1,1,1,1,1,1,0,0,0,0,0,0,0};
+ static const unsigned char std_dc_luminance_values[] = {0,1,2,3,4,5,6,7,8,9,10,11};
+ static const unsigned char std_ac_luminance_nrcodes[] = {0,0,2,1,3,3,2,4,3,5,5,4,4,0,0,1,0x7d};
+ static const unsigned char std_ac_luminance_values[] = {
+ 0x01,0x02,0x03,0x00,0x04,0x11,0x05,0x12,0x21,0x31,0x41,0x06,0x13,0x51,0x61,0x07,0x22,0x71,0x14,0x32,0x81,0x91,0xa1,0x08,
+ 0x23,0x42,0xb1,0xc1,0x15,0x52,0xd1,0xf0,0x24,0x33,0x62,0x72,0x82,0x09,0x0a,0x16,0x17,0x18,0x19,0x1a,0x25,0x26,0x27,0x28,
+ 0x29,0x2a,0x34,0x35,0x36,0x37,0x38,0x39,0x3a,0x43,0x44,0x45,0x46,0x47,0x48,0x49,0x4a,0x53,0x54,0x55,0x56,0x57,0x58,0x59,
+ 0x5a,0x63,0x64,0x65,0x66,0x67,0x68,0x69,0x6a,0x73,0x74,0x75,0x76,0x77,0x78,0x79,0x7a,0x83,0x84,0x85,0x86,0x87,0x88,0x89,
+ 0x8a,0x92,0x93,0x94,0x95,0x96,0x97,0x98,0x99,0x9a,0xa2,0xa3,0xa4,0xa5,0xa6,0xa7,0xa8,0xa9,0xaa,0xb2,0xb3,0xb4,0xb5,0xb6,
+ 0xb7,0xb8,0xb9,0xba,0xc2,0xc3,0xc4,0xc5,0xc6,0xc7,0xc8,0xc9,0xca,0xd2,0xd3,0xd4,0xd5,0xd6,0xd7,0xd8,0xd9,0xda,0xe1,0xe2,
+ 0xe3,0xe4,0xe5,0xe6,0xe7,0xe8,0xe9,0xea,0xf1,0xf2,0xf3,0xf4,0xf5,0xf6,0xf7,0xf8,0xf9,0xfa
+ };
+ static const unsigned char std_dc_chrominance_nrcodes[] = {0,0,3,1,1,1,1,1,1,1,1,1,0,0,0,0,0};
+ static const unsigned char std_dc_chrominance_values[] = {0,1,2,3,4,5,6,7,8,9,10,11};
+ static const unsigned char std_ac_chrominance_nrcodes[] = {0,0,2,1,2,4,4,3,4,7,5,4,4,0,1,2,0x77};
+ static const unsigned char std_ac_chrominance_values[] = {
+ 0x00,0x01,0x02,0x03,0x11,0x04,0x05,0x21,0x31,0x06,0x12,0x41,0x51,0x07,0x61,0x71,0x13,0x22,0x32,0x81,0x08,0x14,0x42,0x91,
+ 0xa1,0xb1,0xc1,0x09,0x23,0x33,0x52,0xf0,0x15,0x62,0x72,0xd1,0x0a,0x16,0x24,0x34,0xe1,0x25,0xf1,0x17,0x18,0x19,0x1a,0x26,
+ 0x27,0x28,0x29,0x2a,0x35,0x36,0x37,0x38,0x39,0x3a,0x43,0x44,0x45,0x46,0x47,0x48,0x49,0x4a,0x53,0x54,0x55,0x56,0x57,0x58,
+ 0x59,0x5a,0x63,0x64,0x65,0x66,0x67,0x68,0x69,0x6a,0x73,0x74,0x75,0x76,0x77,0x78,0x79,0x7a,0x82,0x83,0x84,0x85,0x86,0x87,
+ 0x88,0x89,0x8a,0x92,0x93,0x94,0x95,0x96,0x97,0x98,0x99,0x9a,0xa2,0xa3,0xa4,0xa5,0xa6,0xa7,0xa8,0xa9,0xaa,0xb2,0xb3,0xb4,
+ 0xb5,0xb6,0xb7,0xb8,0xb9,0xba,0xc2,0xc3,0xc4,0xc5,0xc6,0xc7,0xc8,0xc9,0xca,0xd2,0xd3,0xd4,0xd5,0xd6,0xd7,0xd8,0xd9,0xda,
+ 0xe2,0xe3,0xe4,0xe5,0xe6,0xe7,0xe8,0xe9,0xea,0xf2,0xf3,0xf4,0xf5,0xf6,0xf7,0xf8,0xf9,0xfa
+ };
+ // Huffman tables
+ static const unsigned short YDC_HT[256][2] = { {0,2},{2,3},{3,3},{4,3},{5,3},{6,3},{14,4},{30,5},{62,6},{126,7},{254,8},{510,9}};
+ static const unsigned short UVDC_HT[256][2] = { {0,2},{1,2},{2,2},{6,3},{14,4},{30,5},{62,6},{126,7},{254,8},{510,9},{1022,10},{2046,11}};
+ static const unsigned short YAC_HT[256][2] = {
+ {10,4},{0,2},{1,2},{4,3},{11,4},{26,5},{120,7},{248,8},{1014,10},{65410,16},{65411,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {12,4},{27,5},{121,7},{502,9},{2038,11},{65412,16},{65413,16},{65414,16},{65415,16},{65416,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {28,5},{249,8},{1015,10},{4084,12},{65417,16},{65418,16},{65419,16},{65420,16},{65421,16},{65422,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {58,6},{503,9},{4085,12},{65423,16},{65424,16},{65425,16},{65426,16},{65427,16},{65428,16},{65429,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {59,6},{1016,10},{65430,16},{65431,16},{65432,16},{65433,16},{65434,16},{65435,16},{65436,16},{65437,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {122,7},{2039,11},{65438,16},{65439,16},{65440,16},{65441,16},{65442,16},{65443,16},{65444,16},{65445,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {123,7},{4086,12},{65446,16},{65447,16},{65448,16},{65449,16},{65450,16},{65451,16},{65452,16},{65453,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {250,8},{4087,12},{65454,16},{65455,16},{65456,16},{65457,16},{65458,16},{65459,16},{65460,16},{65461,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {504,9},{32704,15},{65462,16},{65463,16},{65464,16},{65465,16},{65466,16},{65467,16},{65468,16},{65469,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {505,9},{65470,16},{65471,16},{65472,16},{65473,16},{65474,16},{65475,16},{65476,16},{65477,16},{65478,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {506,9},{65479,16},{65480,16},{65481,16},{65482,16},{65483,16},{65484,16},{65485,16},{65486,16},{65487,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {1017,10},{65488,16},{65489,16},{65490,16},{65491,16},{65492,16},{65493,16},{65494,16},{65495,16},{65496,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {1018,10},{65497,16},{65498,16},{65499,16},{65500,16},{65501,16},{65502,16},{65503,16},{65504,16},{65505,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {2040,11},{65506,16},{65507,16},{65508,16},{65509,16},{65510,16},{65511,16},{65512,16},{65513,16},{65514,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {65515,16},{65516,16},{65517,16},{65518,16},{65519,16},{65520,16},{65521,16},{65522,16},{65523,16},{65524,16},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {2041,11},{65525,16},{65526,16},{65527,16},{65528,16},{65529,16},{65530,16},{65531,16},{65532,16},{65533,16},{65534,16},{0,0},{0,0},{0,0},{0,0},{0,0}
+ };
+ static const unsigned short UVAC_HT[256][2] = {
+ {0,2},{1,2},{4,3},{10,4},{24,5},{25,5},{56,6},{120,7},{500,9},{1014,10},{4084,12},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {11,4},{57,6},{246,8},{501,9},{2038,11},{4085,12},{65416,16},{65417,16},{65418,16},{65419,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {26,5},{247,8},{1015,10},{4086,12},{32706,15},{65420,16},{65421,16},{65422,16},{65423,16},{65424,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {27,5},{248,8},{1016,10},{4087,12},{65425,16},{65426,16},{65427,16},{65428,16},{65429,16},{65430,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {58,6},{502,9},{65431,16},{65432,16},{65433,16},{65434,16},{65435,16},{65436,16},{65437,16},{65438,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {59,6},{1017,10},{65439,16},{65440,16},{65441,16},{65442,16},{65443,16},{65444,16},{65445,16},{65446,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {121,7},{2039,11},{65447,16},{65448,16},{65449,16},{65450,16},{65451,16},{65452,16},{65453,16},{65454,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {122,7},{2040,11},{65455,16},{65456,16},{65457,16},{65458,16},{65459,16},{65460,16},{65461,16},{65462,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {249,8},{65463,16},{65464,16},{65465,16},{65466,16},{65467,16},{65468,16},{65469,16},{65470,16},{65471,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {503,9},{65472,16},{65473,16},{65474,16},{65475,16},{65476,16},{65477,16},{65478,16},{65479,16},{65480,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {504,9},{65481,16},{65482,16},{65483,16},{65484,16},{65485,16},{65486,16},{65487,16},{65488,16},{65489,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {505,9},{65490,16},{65491,16},{65492,16},{65493,16},{65494,16},{65495,16},{65496,16},{65497,16},{65498,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {506,9},{65499,16},{65500,16},{65501,16},{65502,16},{65503,16},{65504,16},{65505,16},{65506,16},{65507,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {2041,11},{65508,16},{65509,16},{65510,16},{65511,16},{65512,16},{65513,16},{65514,16},{65515,16},{65516,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {16352,14},{65517,16},{65518,16},{65519,16},{65520,16},{65521,16},{65522,16},{65523,16},{65524,16},{65525,16},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {1018,10},{32707,15},{65526,16},{65527,16},{65528,16},{65529,16},{65530,16},{65531,16},{65532,16},{65533,16},{65534,16},{0,0},{0,0},{0,0},{0,0},{0,0}
+ };
+ static const int YQT[] = {16,11,10,16,24,40,51,61,12,12,14,19,26,58,60,55,14,13,16,24,40,57,69,56,14,17,22,29,51,87,80,62,18,22,
+ 37,56,68,109,103,77,24,35,55,64,81,104,113,92,49,64,78,87,103,121,120,101,72,92,95,98,112,100,103,99};
+ static const int UVQT[] = {17,18,24,47,99,99,99,99,18,21,26,66,99,99,99,99,24,26,56,99,99,99,99,99,47,66,99,99,99,99,99,99,
+ 99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99};
+ static const float aasf[] = { 1.0f * 2.828427125f, 1.387039845f * 2.828427125f, 1.306562965f * 2.828427125f, 1.175875602f * 2.828427125f,
+ 1.0f * 2.828427125f, 0.785694958f * 2.828427125f, 0.541196100f * 2.828427125f, 0.275899379f * 2.828427125f };
+
+ int row, col, i, k, subsample;
+ float fdtbl_Y[64], fdtbl_UV[64];
+ unsigned char YTable[64], UVTable[64];
+
+ if(!data || !width || !height || comp > 4 || comp < 1) {
+ return 0;
+ }
+
+ quality = quality ? quality : 90;
+ subsample = quality <= 90 ? 1 : 0;
+ quality = quality < 1 ? 1 : quality > 100 ? 100 : quality;
+ quality = quality < 50 ? 5000 / quality : 200 - quality * 2;
+
+ for(i = 0; i < 64; ++i) {
+ int uvti, yti = (YQT[i]*quality+50)/100;
+ YTable[stbiw__jpg_ZigZag[i]] = (unsigned char) (yti < 1 ? 1 : yti > 255 ? 255 : yti);
+ uvti = (UVQT[i]*quality+50)/100;
+ UVTable[stbiw__jpg_ZigZag[i]] = (unsigned char) (uvti < 1 ? 1 : uvti > 255 ? 255 : uvti);
+ }
+
+ for(row = 0, k = 0; row < 8; ++row) {
+ for(col = 0; col < 8; ++col, ++k) {
+ fdtbl_Y[k] = 1 / (YTable [stbiw__jpg_ZigZag[k]] * aasf[row] * aasf[col]);
+ fdtbl_UV[k] = 1 / (UVTable[stbiw__jpg_ZigZag[k]] * aasf[row] * aasf[col]);
+ }
+ }
+
+ // Write Headers
+ {
+ static const unsigned char head0[] = { 0xFF,0xD8,0xFF,0xE0,0,0x10,'J','F','I','F',0,1,1,0,0,1,0,1,0,0,0xFF,0xDB,0,0x84,0 };
+ static const unsigned char head2[] = { 0xFF,0xDA,0,0xC,3,1,0,2,0x11,3,0x11,0,0x3F,0 };
+ const unsigned char head1[] = { 0xFF,0xC0,0,0x11,8,(unsigned char)(height>>8),STBIW_UCHAR(height),(unsigned char)(width>>8),STBIW_UCHAR(width),
+ 3,1,(unsigned char)(subsample?0x22:0x11),0,2,0x11,1,3,0x11,1,0xFF,0xC4,0x01,0xA2,0 };
+ s->func(s->context, (void*)head0, sizeof(head0));
+ s->func(s->context, (void*)YTable, sizeof(YTable));
+ stbiw__putc(s, 1);
+ s->func(s->context, UVTable, sizeof(UVTable));
+ s->func(s->context, (void*)head1, sizeof(head1));
+ s->func(s->context, (void*)(std_dc_luminance_nrcodes+1), sizeof(std_dc_luminance_nrcodes)-1);
+ s->func(s->context, (void*)std_dc_luminance_values, sizeof(std_dc_luminance_values));
+ stbiw__putc(s, 0x10); // HTYACinfo
+ s->func(s->context, (void*)(std_ac_luminance_nrcodes+1), sizeof(std_ac_luminance_nrcodes)-1);
+ s->func(s->context, (void*)std_ac_luminance_values, sizeof(std_ac_luminance_values));
+ stbiw__putc(s, 1); // HTUDCinfo
+ s->func(s->context, (void*)(std_dc_chrominance_nrcodes+1), sizeof(std_dc_chrominance_nrcodes)-1);
+ s->func(s->context, (void*)std_dc_chrominance_values, sizeof(std_dc_chrominance_values));
+ stbiw__putc(s, 0x11); // HTUACinfo
+ s->func(s->context, (void*)(std_ac_chrominance_nrcodes+1), sizeof(std_ac_chrominance_nrcodes)-1);
+ s->func(s->context, (void*)std_ac_chrominance_values, sizeof(std_ac_chrominance_values));
+ s->func(s->context, (void*)head2, sizeof(head2));
+ }
+
+ // Encode 8x8 macroblocks
+ {
+ static const unsigned short fillBits[] = {0x7F, 7};
+ int DCY=0, DCU=0, DCV=0;
+ int bitBuf=0, bitCnt=0;
+ // comp == 2 is grey+alpha (alpha is ignored)
+ int ofsG = comp > 2 ? 1 : 0, ofsB = comp > 2 ? 2 : 0;
+ const unsigned char *dataR = (const unsigned char *)data;
+ const unsigned char *dataG = dataR + ofsG;
+ const unsigned char *dataB = dataR + ofsB;
+ int x, y, pos;
+ if(subsample) {
+ for(y = 0; y < height; y += 16) {
+ for(x = 0; x < width; x += 16) {
+ float Y[256], U[256], V[256];
+ for(row = y, pos = 0; row < y+16; ++row) {
+ // row >= height => use last input row
+ int clamped_row = (row < height) ? row : height - 1;
+ int base_p = (stbi__flip_vertically_on_write ? (height-1-clamped_row) : clamped_row)*width*comp;
+ for(col = x; col < x+16; ++col, ++pos) {
+ // if col >= width => use pixel from last input column
+ int p = base_p + ((col < width) ? col : (width-1))*comp;
+ float r = dataR[p], g = dataG[p], b = dataB[p];
+ Y[pos]= +0.29900f*r + 0.58700f*g + 0.11400f*b - 128;
+ U[pos]= -0.16874f*r - 0.33126f*g + 0.50000f*b;
+ V[pos]= +0.50000f*r - 0.41869f*g - 0.08131f*b;
+ }
+ }
+ DCY = stbiw__jpg_processDU(s, &bitBuf, &bitCnt, Y+0, 16, fdtbl_Y, DCY, YDC_HT, YAC_HT);
+ DCY = stbiw__jpg_processDU(s, &bitBuf, &bitCnt, Y+8, 16, fdtbl_Y, DCY, YDC_HT, YAC_HT);
+ DCY = stbiw__jpg_processDU(s, &bitBuf, &bitCnt, Y+128, 16, fdtbl_Y, DCY, YDC_HT, YAC_HT);
+ DCY = stbiw__jpg_processDU(s, &bitBuf, &bitCnt, Y+136, 16, fdtbl_Y, DCY, YDC_HT, YAC_HT);
+
+ // subsample U,V
+ {
+ float subU[64], subV[64];
+ int yy, xx;
+ for(yy = 0, pos = 0; yy < 8; ++yy) {
+ for(xx = 0; xx < 8; ++xx, ++pos) {
+ int j = yy*32+xx*2;
+ subU[pos] = (U[j+0] + U[j+1] + U[j+16] + U[j+17]) * 0.25f;
+ subV[pos] = (V[j+0] + V[j+1] + V[j+16] + V[j+17]) * 0.25f;
+ }
+ }
+ DCU = stbiw__jpg_processDU(s, &bitBuf, &bitCnt, subU, 8, fdtbl_UV, DCU, UVDC_HT, UVAC_HT);
+ DCV = stbiw__jpg_processDU(s, &bitBuf, &bitCnt, subV, 8, fdtbl_UV, DCV, UVDC_HT, UVAC_HT);
+ }
+ }
+ }
+ } else {
+ for(y = 0; y < height; y += 8) {
+ for(x = 0; x < width; x += 8) {
+ float Y[64], U[64], V[64];
+ for(row = y, pos = 0; row < y+8; ++row) {
+ // row >= height => use last input row
+ int clamped_row = (row < height) ? row : height - 1;
+ int base_p = (stbi__flip_vertically_on_write ? (height-1-clamped_row) : clamped_row)*width*comp;
+ for(col = x; col < x+8; ++col, ++pos) {
+ // if col >= width => use pixel from last input column
+ int p = base_p + ((col < width) ? col : (width-1))*comp;
+ float r = dataR[p], g = dataG[p], b = dataB[p];
+ Y[pos]= +0.29900f*r + 0.58700f*g + 0.11400f*b - 128;
+ U[pos]= -0.16874f*r - 0.33126f*g + 0.50000f*b;
+ V[pos]= +0.50000f*r - 0.41869f*g - 0.08131f*b;
+ }
+ }
+
+ DCY = stbiw__jpg_processDU(s, &bitBuf, &bitCnt, Y, 8, fdtbl_Y, DCY, YDC_HT, YAC_HT);
+ DCU = stbiw__jpg_processDU(s, &bitBuf, &bitCnt, U, 8, fdtbl_UV, DCU, UVDC_HT, UVAC_HT);
+ DCV = stbiw__jpg_processDU(s, &bitBuf, &bitCnt, V, 8, fdtbl_UV, DCV, UVDC_HT, UVAC_HT);
+ }
+ }
+ }
+
+ // Do the bit alignment of the EOI marker
+ stbiw__jpg_writeBits(s, &bitBuf, &bitCnt, fillBits);
+ }
+
+ // EOI
+ stbiw__putc(s, 0xFF);
+ stbiw__putc(s, 0xD9);
+
+ return 1;
+}
+
+STBIWDEF int stbi_write_jpg_to_func(stbi_write_func *func, void *context, int x, int y, int comp, const void *data, int quality)
+{
+ stbi__write_context s = { 0 };
+ stbi__start_write_callbacks(&s, func, context);
+ return stbi_write_jpg_core(&s, x, y, comp, (void *) data, quality);
+}
+
+
+#ifndef STBI_WRITE_NO_STDIO
+STBIWDEF int stbi_write_jpg(char const *filename, int x, int y, int comp, const void *data, int quality)
+{
+ stbi__write_context s = { 0 };
+ if (stbi__start_write_file(&s,filename)) {
+ int r = stbi_write_jpg_core(&s, x, y, comp, data, quality);
+ stbi__end_write_file(&s);
+ return r;
+ } else
+ return 0;
+}
+#endif
+
#endif // STB_IMAGE_WRITE_IMPLEMENTATION
/* Revision history
+ 1.16 (2021-07-11)
+ make Deflate code emit uncompressed blocks when it would otherwise expand
+ support writing BMPs with alpha channel
+ 1.15 (2020-07-13) unknown
+ 1.14 (2020-02-02) updated JPEG writer to downsample chroma channels
+ 1.13
+ 1.12
+ 1.11 (2019-08-11)
+
+ 1.10 (2019-02-07)
+ support utf8 filenames in Windows; fix warnings and platform ifdefs
+ 1.09 (2018-02-11)
+ fix typo in zlib quality API, improve STB_I_W_STATIC in C++
+ 1.08 (2018-01-29)
+ add stbi__flip_vertically_on_write, external zlib, zlib quality, choose PNG filter
+ 1.07 (2017-07-24)
+ doc fix
+ 1.06 (2017-07-23)
+ writing JPEG (using Jon Olick's code)
+ 1.05 ???
+ 1.04 (2017-03-03)
+ monochrome BMP expansion
+ 1.03 ???
+ 1.02 (2016-04-02)
+ avoid allocating large structures on the stack
+ 1.01 (2016-01-16)
+ STBIW_REALLOC_SIZED: support allocators with no realloc support
+ avoid race-condition in crc initialization
+ minor compile issues
+ 1.00 (2015-09-14)
+ installable file IO function
+ 0.99 (2015-09-13)
+ warning fixes; TGA rle support
0.98 (2015-04-08)
added STBIW_MALLOC, STBIW_ASSERT etc
0.97 (2015-01-18)
@@ -722,7 +1669,7 @@ int stbi_write_png(char const *filename, int x, int y, int comp, const void *dat
add HDR output
fix monochrome BMP
0.95 (2014-08-17)
- add monochrome TGA output
+ add monochrome TGA output
0.94 (2014-05-31)
rename private functions to avoid conflicts with stb_image.h
0.93 (2014-05-27)
@@ -733,3 +1680,45 @@ int stbi_write_png(char const *filename, int x, int y, int comp, const void *dat
first public release
0.90 first internal release
*/
+
+/*
+------------------------------------------------------------------------------
+This software is available under 2 licenses -- choose whichever you prefer.
+------------------------------------------------------------------------------
+ALTERNATIVE A - MIT License
+Copyright (c) 2017 Sean Barrett
+Permission is hereby granted, free of charge, to any person obtaining a copy of
+this software and associated documentation files (the "Software"), to deal in
+the Software without restriction, including without limitation the rights to
+use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies
+of the Software, and to permit persons to whom the Software is furnished to do
+so, subject to the following conditions:
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.
+------------------------------------------------------------------------------
+ALTERNATIVE B - Public Domain (www.unlicense.org)
+This is free and unencumbered software released into the public domain.
+Anyone is free to copy, modify, publish, use, compile, sell, or distribute this
+software, either in source code form or as a compiled binary, for any purpose,
+commercial or non-commercial, and by any means.
+In jurisdictions that recognize copyright laws, the author or authors of this
+software dedicate any and all copyright interest in the software to the public
+domain. We make this dedication for the benefit of the public at large and to
+the detriment of our heirs and successors. We intend this dedication to be an
+overt act of relinquishment in perpetuity of all present and future rights to
+this software under copyright law.
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+------------------------------------------------------------------------------
+*/
diff --git a/external/include/tiny_gltf.h b/external/include/tiny_gltf.h
new file mode 100644
index 00000000..6d8c3ffb
--- /dev/null
+++ b/external/include/tiny_gltf.h
@@ -0,0 +1,8733 @@
+//
+// Header-only tiny glTF 2.0 loader and serializer.
+//
+//
+// The MIT License (MIT)
+//
+// Copyright (c) 2015 - Present Syoyo Fujita, Aurélien Chatelain and many
+// contributors.
+//
+// Permission is hereby granted, free of charge, to any person obtaining a copy
+// of this software and associated documentation files (the "Software"), to deal
+// in the Software without restriction, including without limitation the rights
+// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+// copies of the Software, and to permit persons to whom the Software is
+// furnished to do so, subject to the following conditions:
+//
+// The above copyright notice and this permission notice shall be included in
+// all copies or substantial portions of the Software.
+//
+// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+// THE SOFTWARE.
+
+// Version: - v2.9.*
+// See https://github.com/syoyo/tinygltf/releases for release history.
+//
+// Tiny glTF loader is using following third party libraries:
+//
+// - jsonhpp: C++ JSON library.
+// - base64: base64 decode/encode library.
+// - stb_image: Image loading library.
+//
+#ifndef TINY_GLTF_H_
+#define TINY_GLTF_H_
+
+#include
+#include
+#include // std::fabs
+#include
+#include
+#include
+#include
+#include
+#include