Hi @JuhaHuiskonen ,

Thanks a lot for the PR! I think it's a very valuable feature to be able to process all connected components within the segmentation.
However, could you explain a bit more on your thoughts here?
Particularly, do you think it's best to store all components of one tomogram / segmentation into a single .h5 container.

I'm a bit hesitant with this as it would break a bit the compatibility with subsequent steps (annotation in surforama, analysis in membrain-stats), so I would prefer to keep the 1 membrane <--> 1 .h5 container relation.
In this sense, I think it would make sense to simply store multiple .h5 containers for multiple connected components within a segmentation and just call them "seg_name_comp1.h5", "seg_name_comp2.h5" ...

What do you think?

Hi,

The meshes are written in the same container only if "--merge-outputs" option is given. The default is to write a separate file as you describe.

Cheers,
Juha

Ooh yes, I see it now. Thx for the clarification. Will try to review in detail as soon as possible!

Hey @JuhaHuiskonen

thanks for the PR! I had a look, and I’d prefer not to merge it for now. It’s quite a substantial change affecting many files, and I don’t currently have the right dataset or workflow to properly test all functionalities — especially the multiple membranes within a single .h5 container.

Would it make sense to simplify things for now by keeping one .h5 container per membrane and leaving multi-membrane support for a later update?

Of course. It seems to work for me as intended so we can proceed with our project. I am also happy to provide you some test data with multiple membranes in each tomogram.