Smart Depth Deep Overscan/Segmentation

[im.thatoneguy]

It would be great if there was a way to do a pre-pass on deep rendering with motion blur and DOF on much like the shader-helper that would detect occluded image regions/depths that will be revealed by foreground motion blur or DOF. I'm imagining the prepass segmenting the deep pre-pass into like 2-3 depth segments (user setting or auto) and then during the deep rendering firing an extra primary view ray from ((Front-Back * 0.2) + Back) Depth to avoid the foreground object but still sample the background.

Even without that overscan feature, it would just be nice to have a smarter Deep fragment segmentation algorithm which doesn't increase file sizes quite so much.

[vlado]

I'm not sure what would be the purpose of this? Deep OpenEXR files do not support multiple deep points with full alpha in the same pixel.

There are already several deep fragment merge modes available in the VRayOptionRE render element.

Best regards,
Vlado