Hi. I'd like to know, is there any stereoscopic rendering optimization in mantra, like in PRman?. On odforce, someone said, that it is possible to force mantra to reuse shading samples during rendering second camera. Is it true? Thanks!

Sorry to dig up old thread, but I have also been trying to find out how to do this. Is this possible?

Nope, you can't do this in Mantra currently. Sorry!

Thank you Jason. Maybe sometime down the road, when stereo is used more sesi will consider implementing this.

I am not sure how much you can share
since even dicing is camera dependent (and dicing from 2 cameras at the same time would produce heavier mesh)

it seems that what renderman is using is big fake that can maybe work for old render engines, but I cannot imagine this for adaptive sampling in PBR and all the camera dependent rays
maybe even if something can be shared like theoretically goeometry or diffuse shading, the memory consumption will be higher than for one view, which sometimes is not enough even for that.

I prefer if sesi continues on improving mantra speed in general, but not introducing fake solutions for special cases like stereo multiview rendering, especially not if that would require customised shaders like it was in renderman (not sure about present version)

but if clever solution exists it may be a nice feature