some questions about the ray position encoding

[liuzhidemaomao]

mvp/lib/models/position_encoding.py

Line 274 in 80eecd0

pixel_world = torch.bmm(pixel_camera-T.transpose(2, 1), R)

Is there are something wrong with this code, the last R may be R.T or inverse(R)? And I have another question, the pixel world coordinates calculated without Z_c in your code. Can it reflect the real position of pixels in the world coordinate system?