Color thumbnails with view config

[frcroth]

URL of deployed dev instance (used for testing):

• https://___.webknossos.xyz

Steps to test:

• Set view config of some dataset to include color
• View thumbnail
  

Notes

• Only tested with one dataset and uint16 data Tested with uint8 and unit16
• I changed the intensity value calculation to use ints instead of bytes. Since bytes are signed in jvm (not a fan), clamping from 0 to 255 and then setting toByte lead to negative values, which messed up some calculations. Now it uses positive ints, which should work just as well.
• I used the Color class I found here and in Automatically assign color and channel name for omero ngff #7251, don't know if that class should be used.
• Optional TODO: Respect inverted setting Done

Issues:

• fixes Thumbnails should use layer color from view configuration #7237

---

(Please delete unneeded items, merge only when none are left open)

• Updated changelog
• Needs datastore update after deployment
• Remove dev-only changes (Use cache again)

[fm3]

Sweet, looking forward to this!

• Reusing the existing Color class is fine by me
• Changing the data types of the intermediate results is also fine, as long as it does not create data loss
• Passing the invert setting would be a good follow up issue :) (I guess if you consider it to be easily done, feel free to include it here as well)
• I think there are still some overflows happening, I’m getting something like this
  
  for a uint8 dataset (ROI2017_wkw) with color setting [200,5,5]
  The frontend rendering shows this (not the exact same location, sorry, but you get the idea)
  

[frcroth]

@fm3 I can't reproduce the faulty behavior. Maybe you can take another look at this?
I tested it with ROI2017 and got this (h=400, w=400)

[fm3]

I just tried again and I can actually still reproduce it.

This his what my dataset configuration looks like (from the dataset settings view)

And visiting http://localhost:9000/api/datasets/sample_organization/ROI2017_wkw/layers/color/thumbnail?w=100&h=100 (with ctrl+f5 for cache-clear refresh) yields this image

[fm3]

Looks like it does work when intensityRange is also set. 🤔 Even when it covers the full range (0,255)

[frcroth]

Ok, I think it was once again a weirdness of the byte type, which resulted in no intesity range being different to given full intesity range. Should be fixed now.

[fm3]

Works nicely now, thanks!

Added two more small comments about naming.

Also don’t forget to enable the caching again, then this should be good to go :)

[fm3]

🎉