search

NVIDIAGameWorks / kaolin

A PyTorch Library for Accelerating 3D Deep Learning Research
Apache License 2.0
4.45k stars 552 forks source link

Is there a differentiable function for inverse perspective geometry? #191

Closed SeokjuLee closed 2 months ago

SeokjuLee commented 4 years ago

Thanks for releasing such a powerful library for 3D vision. I am interested in estimating depth by differentiable rendering. If a dense correspondence map (F_12) and relative camera pose (T_12) between two frames (I_1, I_2) are given, can I inversely compute a relative reference depth in a differentiable way? Could you recommend some scripts to achieve this?

krrish94 commented 4 years ago

Currently the differentiable renderers support rendering of depth. Although there isn't a goto tutorial at the moment, it is a good suggestion and we will expose our depth image API in a future release.

shumash commented 2 months ago

Closing old issue.