Closed wawanbreton closed 1 month ago
Looks good! I don't fully remember why I initially did the alpha that way ... it might've been part of another shader first.
Maybe because the alpha can be modified during the OpenGL pipeline processing, so it is a bit more sensitive.
The only thing I do notice is that
if current_model_id >= 128:
can probably be replaced withif current_model_id >= 255:
now.
Good catch, that is true :) I'll make a commit
Instead of using a dynamic number of bits for the face index, store the object ID in the alpha channel, which was previously unused. This way we have the full 24 bits of the RGB channels to store the face index.
OpenGL blending is disabled for this pass, so the alpha channel is just stored as an extra channel but no calculation is performed based on it.
CURA-11984