I’m looking for ways to make this more efficient.
I’m working on a mapping project of a sculpture. The content for the sculpture is viewed in vvvv from the ideal spectator position then projective textured back onto the model for projector output. As the spectator position defines the view of the content I can’t choose a view / framing that would be efficient (which in my case would be a top down view not side on).
To ensure enough detail in the resulting content texture once it’s reprojected I need to pump up the backbuffer resolution a lot. This slows down the shading in that first rendering stage, and also any subsequent pixel shaders of post processing stages, even though much of the output of these are now black.
Do you guys know any tricks to help with this? I’m using dx11. I can’t think of anything to help the first rendering stage, but I think there is some way to help the pixel shader post processing stages, maybe using stencil buffers (which I know nothing about). Or maybe it’s as simple as discarding pixels if there is no content there (ie if pixel colour matches background?