In function transformed_params2depthplussilhouette(params, w2c, transformed_gaussians) , the rendevar["colors_precomp"]uses get_depth_and_silhouette(pts_3D, w2c) which processes the first parameter as gaussians's 3D coordinates in the world coordinates.
However, in transformed_params2depthplussilhouette, it uses get_depth_and_silhouette(transformed_gaussians['means3D'], w2c) , indicating that the coordinates are transferred to world coordinates twice.
In function
transformed_params2depthplussilhouette(params, w2c, transformed_gaussians), therendevar["colors_precomp"]usesget_depth_and_silhouette(pts_3D, w2c)which processes the first parameter as gaussians's 3D coordinates in the world coordinates.However, in
transformed_params2depthplussilhouette, it usesget_depth_and_silhouette(transformed_gaussians['means3D'], w2c), indicating that the coordinates are transferred to world coordinates twice.I don't whether it's a mistake.
Thanks for your reply! ☺️