Jason Y. Zhang

I had a similar problem that I solved because I did not clone with `--recursive`. Even though the Eigen folder was there, the submodule was not cloned.

When I got the warning about Pytorch being compiled with g++ but c++ being the default compiler, I fixed it by doing this: ``` CC=gcc CXX=g++ python setup.py install ```

Hi, I suspect the issue might result from a lack of person detections in the first frame. The easiest way to debug this though is by uploading the video you...

Yes, we've tested on in-the-wild videos. Most of the demo videos are scraped from Instagram. Could you upload the video that you are having trouble with?

Hi, the training and evaluation code are coming soon! I'm working on making them ready for release. Best, Jason

Hi, In order to run on your own videos, you need to have a video tracklet cropped around a person. The easiest way to do this is to compute bounding...

Hi, This is a very challenging sequence. Here's what I might try to get better results: - spend time getting the camera poses as accurate as possible - remove the...

Hi Ali, To use NeRS out of the box, you will need to set initial cuboid dimensions and the camera extrinsics via [azimuth/elevation representation](https://en.wikipedia.org/wiki/Horizontal_coordinate_system) in degrees. The metadata.json is just...

Hi, You will need to convert your camera poses to the Pytorch3D camera convention. Given the 4x4 camera extrinsics tensor from COLMAP, you should be able to convert it to...

I had this issue as well. Turns out I was using the pytorch version without cuda support (since I only have CPU). I reinstalled pytorch with the proper CUDA version,...