Hi,
For quite some time now we have been battling some issues related to the occlusion of secondary rays when using matte object flags. The example below demonstrates how when matte object is activated, those objects do not receive self-shadowing for secondary rays. In this example, the sun should be occluded, and not affect the sphere, however once matte object is activated the ground plane does not receive correct shadowing (evident in the light cache example attached). This results in GI bounce back onto the ball that is incorrect. We have gotten around situations like this in the past by creating separate sets for matte objects vs shadow casters, but in more complex scenarios this is not possible. The other option is to render without hold-outs and a separate matte, but this leads to incorrectly pre-multiplied pixels, especially in cases of extreme motion blur. The only other option is to render deep, and deep merge, but we would prefer to avoid expensive compositing operations if possible. Any chance this can be resolved? Or is it just a limitation of the matte object workflow.
Cheers,
Drew
For quite some time now we have been battling some issues related to the occlusion of secondary rays when using matte object flags. The example below demonstrates how when matte object is activated, those objects do not receive self-shadowing for secondary rays. In this example, the sun should be occluded, and not affect the sphere, however once matte object is activated the ground plane does not receive correct shadowing (evident in the light cache example attached). This results in GI bounce back onto the ball that is incorrect. We have gotten around situations like this in the past by creating separate sets for matte objects vs shadow casters, but in more complex scenarios this is not possible. The other option is to render without hold-outs and a separate matte, but this leads to incorrectly pre-multiplied pixels, especially in cases of extreme motion blur. The only other option is to render deep, and deep merge, but we would prefer to avoid expensive compositing operations if possible. Any chance this can be resolved? Or is it just a limitation of the matte object workflow.
Cheers,
Drew
Comment