Infer the other cameras

[yuxiaoz]

After training, how to utilize the model to render the other cameras and save the ply? From the code, I can forward using a camera and get a predict "rgb", how to get the dense point clouds? Appreciate.

[yuxiaoz]

3D Gaussian Splatting is not like DL, all of the dense gaussian points with info must be trained utilizing all viewed frames.