Thanks @calvinchai this looks great.

@jingjingwu1225 I have modified your multi-slice jp2000 converter. See linc_convert/modalities/df/multi_slice.py. The idea is to remove hardcoded references to the data organization. Instead you'll need to run it per subject, and to pass all of the input files (although a glob in the shell should work). Could you try it quickly on one of Suzanne's subjects?

Although I guess we could merge already and test/fix df.multi_slice later. As you wish.

Hi @Yael, the reason I used 2**level in multiscales part is because it causes error if there is half voxel difference between 2 adjacent resolution levels when loading in webknossos.

@balbasty I just talked to Jingjing and I think we might need to fix the problem she mentioned and test it. But I think it might be better to first merge this and do another PR to keep things tracked and organized (we need to update test data file in this process). Also we can test if the GitHub action is working properly through this kind of push and PR. What do you think?

I agree, let's merge this!

Where are we at with this? Should we merge?

Sorry @balbasty , I was out sick yesterday and I will merge this now.