Fail to reconstruct reasonable volume for EMPIAR10028

[AlenCarral]

I strictly followed the steps provided in the README file. However, I encountered an issue where the reconstruction protein model appears to have a hollow structure, whether w/wo gt-pose:

left: wo gt-pose right with gt-pose

ps: I have already validated the input data by running cryodrgn and cryodrgn2 using exact the same set of data, they produced reasonable structures:

Has anyone encountered a similar situation before?Or can someone provide some ideas or suggestions?

[smallelephant9516]

I think it can be solved by multiplying a negative sign to the ctf function or the image in the fourier space.

[1320harry]

I also encountered the same situation. Did you resolve the problem?

[LucunJi]

Same situation, and after multiplying -1 to the reconstructed volume, I get a seemingly correct result.