In rendering, there are depth, distance mean and distance median. I am curious why using distance_median for tsdf reconstruction. In another word, if I want to add the depth supervision, which of those three is more appropriate for calculating the loss?
In rendering, there are depth, distance mean and distance median. I am curious why using distance_median for tsdf reconstruction. In another word, if I want to add the depth supervision, which of those three is more appropriate for calculating the loss?