fixed loading of 0,0,0 block for 8bit data

[ekatrukha]

Hello @tpietzsch,

bad news, I've found a bug: rendering of 8bit multi resolution sometimes shows
black boxes. It occurs randomly, not every time and most of
the time this block is at the zero axis boundary or close to the origin of coordinates.

So to reproduce an error, I've made an example Debug8bitMultiRes.java,
where 8bit is loaded in green and the same 16bit data is loaded in red.

This is how it looks like (without the fix):

20250527_texture_load_BVV.mp4

The reason for that being (after reading the code) that the 0,0,0 block is loaded
in a special way, because of texSubImage3D bug.

Now, the good news: the easy correction of its loading in TextureCache seems to fix it (as submitted).

Let me know what you think.

Cheers,
Eugene

[tpietzsch]

Good catch! I'll merge and make a new release tomorrow.

[NicoKiaru]

@tpietzsch please let us know when you'll have time to merge it. Also - it's great if it's updated in pom-scijava:

https://github.com/scijava/pom-scijava/blob/260925d60eb2d277c1c0e54b03241e468b04aaf6/pom.xml#L1039

@ekatrukha when all of this is done, maybe we could synchronize and I can remove the 8-bit -> 16-bit conversion in bigdataviewer-image-loaders ? Or do you think it's needed for something else ?

Cheers

[ekatrukha]

@NicoKiaru Yes, you can remove it. I've already updated my codebase and I do not use it anymore.

[tpietzsch]

I finally merged the PR and released bigvolumeviewer-0.4.1

[tpietzsch]

I made a (draft) PR to pom-scijava: scijava/pom-scijava#294

However, as mentioned in the PR description, I think this might mess with scenery/SciView, so someone will need to look into that...