Ecosphere Card Generator (using Stable Diffusion)

A Python program that uses Stable Diffusion to generate Ecosphere card art (and then writes the card stats, just to be complete). AKA Ecosphere | SD | PIL.

Usage

Run the program with one of the following commands, executed from within the activated virtual environment in the base directory containing the README.md.

• python card-maker/make_stats.py: Only write card stats. Outputs to output/finished_cards/stats/.
• python card-maker/make_art.py: Only write card art. Outputs to output/finished_cards/art/.
• python card-maker/make_both.py: Write card stats and art. Outputs to output/finished_cards/art_and_stats/.

Card Stats

Card stats are read from the CSV file assets_data/card_database.csv. This file is the exact same as Neo's official Card Database Google Sheets document, except I've added an Art Description column after Text. This column contains part of the prompt used in image generation (additional prompt is added in the program). This column is incomplete, only having data for 12 cards.

Image Generation

Raw artwork is saved to output/generated_artwork/.

The seed for image generation is hardcoded to 27. This means that, when the program is run twice, and an image prompt is not changed, it will generate the exact same image. To change this, you can edit line 20 of card_maker/make_art.py

See the above Card Stats section for information on image generation prompts.

Installation

Virtual Environment

You'll need to setup a venv for this, because there are so many specific dependencies.

1. Install virtualenv globally with pip install virtualenv
2. Create a venv in your cwd with virtualenv --python C:/Python/Python310/python.exe venv (replacing with your absolute Python executable path)
3. Enter the venv with .\venv\Scripts\activate. Now, all commands run from the terminal will affect the virtual environment and will be isolated from the larger system

All commands (including below installation commands) should be executed from within the activated virtual environment.

Python Requirements

Install dependencies with pip install -r requirements.txt.

opencv-python

1. Download opencv-python wheel file here
2. Rename file to opencv_python-4.1.2.30-cp38-abi3-win_amd64.whl.
3. Install with pip install opencv_python-4.1.2.30-cp38-abi3-win_amd64.whl

CUDA/Torch

1. Download CUDA installer from here
2. Install CUDA from installer
3. Uninstall the wrong packages that were installed a few steps ago (sorry :/) with pip uninstall torch torchvision torchaudio (you'll need to enter "y" to confirm it)
4. Install torch packages with pip install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu117)

Stable Diffusion Model

Place model.ckpt at location: stable-diffusion-main/models/ldm/stable-diffusion-v1/model.ckpt

(If you need a model, download Stable Diffusion 1.4 from here)

License

I license all possible parts of this repository under GPLv3.

This is built off of the Optimized Stable Diffusion fork, which was forked while SD was licensed under an All Rights Reserved license (before it was changed to the CreativeML Open RAIL M License). I'm not sure what the legal implications are for licensing this, but note that parts of this might be licensed under that.