In depth2normal_tf.py, line 121, you do normal_vector = normalize_l2(normals0). Your final normal vector should be some weighted average of the four resulting cross products normalized by l2 norm, right? line 121 should be
normal_vector = normalize_l2(normal_vector).
Also, isn't it kind of redundant to normalize these vectors multiple times? You should only need to normalize at the end?
In depth2normal_tf.py, line 121, you do
normal_vector = normalize_l2(normals0). Your final normal vector should be some weighted average of the four resulting cross products normalized by l2 norm, right? line 121 should benormal_vector = normalize_l2(normal_vector).Also, isn't it kind of redundant to normalize these vectors multiple times? You should only need to normalize at the end?