Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 95.69%. Comparing base (5f7a99c) to head (cc789cb).
Report is 1 commits behind head on master.

@@            Coverage Diff             @@
##           master    #1879      +/-   ##
==========================================
+ Coverage   95.68%   95.69%   +0.01%     
==========================================
  Files         125      125              
  Lines        9961     9988      +27     
==========================================
+ Hits         9531     9558      +27     
  Misses        430      430              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

I'm trying to understand the behavior with this PR. When the camera is rolled, the temporary up is used, but when do we restore the old up value? Only when we reset the camera? I think we should slowing converge back the temporary up direction to the initial up direction by doing a slerp

Thats not what we want here I think.

I'm trying to understand the behavior with this PR. When the camera is rolled, the temporary up is used, but when do we restore the old up value? Only when we reset the camera? I think we should slowing converge back the temporary up direction to the initial up direction by doing a slerp

The bug was that rotating the camera around the object while rolled causes the camera to jerk back to its original orientation. This PR does not change up in the renderer, it gives the interactor a definition to use after rolling with roll_camera that reflects the view up of the camera. When the camera is reset, the camera is oriented in a way where the renderer's up value is correct again, so at that point the flag is set to false.

Thanks for your contribution @samoncrief !