Closed chenhao-user closed 1 year ago
Hi, do you figure out this problem?
I seem to have operated like this before, where 0.5 and 0.1 are values I tried out myself, but of course, you can also try other values. `face_texture = torch.full_like(face_texture, 0.5)
face_norm = self.compute_norm(face_shape)
face_norm_roted = face_norm @ rotation
face_gamma = torch.full_like(coef_dict['gamma'], 0.1)
face_color = self.compute_color(face_texture, face_norm_roted, face_gamma)`
Hi! I'd like to know how to generate this kind of facial image below? Based on my own understanding, I set the illumination to 1 in the compute_color function without multiplying it by the texture parameter. I made the following modification.
However, the results of this change were not satisfactory. One of the results is as follows: