Fix histogram sampling for tiny datasets

[fm3]

The position sampling for histogram and find data roughly works like a menger sponge, with a stopping criterion when the distance between the sampling points becomes too close. For tiny datasets, the first iteration already runs into the stopping criterion.

The fix is to always add the 32vx³ bucket at 0,0,0 to the position list. This way, the data from this corner (which tends to exist for tiny datasets) is included.

URL of deployed dev instance (used for testing):

• https://tinyhistogram.webknossos.xyz

Steps to test:

• open a very small dataset (tested with 50 × 50 × 30 voxel³)
• sidebar should show valid histogram

Issues:

• fixes Histogram request fails for tiny datasets #5444

---

• Updated (unreleased) changelog
• Needs datastore update after deployment
• Ready for review

[youri-k]

Nice idea 🎉

[philippotto]

Awesome to see this fixed!

The fix is to always add the 32vx³ bucket at 0,0,0 to the position list.

Is this relative to the datasource's bounding box?

[fm3]

Is this relative to the datasource's bounding box?

It will be in the next version 🙈 😆 good idea