Peter Lin

Hope this implemented soon as well!

DO NOT MERGE this one yet. We are temporarily making some change in the RVM repo. Will take some time to be available again.

This pull request is ready to merge again.

@NicolasHug Can you approve the merge?

@NicolasHug Hi Nicolas. Any updates?

Do we have GPU on the test environment? Otherwise the inference will be too slow.

I don't think I recently have time for making the changes. If we can just merge this I would really appreciate.

@NicolasHug

This PR fixes fp16 on implicit renderer: https://github.com/facebookresearch/pytorch3d/pull/946/files Please approve the merge.

@li-wenquan You need to train on stage 4 using Distinctions and AIM dataset. VideoMatte240k dataset has pretty bad foreground samples so the network will learn to make the background all...