Considering masks in colmap2nerf script

[yorkpud]

I have images and corresponding alpha masks. I want to consider masks in the pose estimation, so that black mask pixels are not taken in the feature extraction phase. This is already available in Colmap.
Is it possible to do it in the colmap2nerf script? What should be changed? Any advices and hints on how to do this?