Trying to import raw data from LZ1 CZI does not try to fix incorrect stack size

[lacan]

When we import czi files from the LZ1 and choose resave as hdf5, the tiles are correctly placed and the stack sizes are fixed.

I we do the very same procedure and ask to keep the raw data for the dataset.xml file, all images are at position 0,0,0 and have the same number of slices...

Any help where this problem lies would be appreciated. I could create a pull request myself once I know where to look.

Best
Oli

[hoerldavid]

Hi @lacan,

I just checked with some example .czi files we have lying around here and could replicate the same z-size stack bug. That was working before but might have broken during one of the last updates. I will look into this with the files I have.

I could not reproduce the poosition (0,0,0) bug - could you maybe create a very small dataset (only a few planes) where it occurs and send it to us?

Regarding hacking on the code yourself, the AutoLoader is mainly implemented in https://github.com/PreibischLab/multiview-reconstruction/blob/master/src/main/java/net/preibisch/mvrecon/fiji/datasetmanager/FileListDatasetDefinition.java and https://github.com/PreibischLab/multiview-reconstruction/blob/master/src/main/java/net/preibisch/mvrecon/fiji/datasetmanager/FileListDatasetDefinitionUtil.java.
However, this is one of the more convoluted parts of the codebase, so it might not be the easiest part to work on.

Regardless, if you fix a bug, PRs are always welcome :)

Best,
David

[hoerldavid]

Hi @lacan,

Just a short update:
I found the reason why the stack sizes were not corrected, it should be fixed with c5ce4c2 and will be included in the next release in Fiji.
(I'm responsible for this breaking, as I had introduced a missing-exclamation-mark "typo-bug" in some refactoring done in 1142041. Sorry for that).

Best,
David