OptiX Ray Tracer optimization.
I think that is possible if :
1 ) Use Z-buffering first (for all objects), and then use it for avoid first (0-depth) level of ray tracing.
Saving all triangles that intersects from surface pixels by reflection rays
and in next frame use its for first time test reflection rays intersections for thats pixels.
That’s can work because another rays from camera to the corresponding pixels turn on small values from previous, usually.
Because, camera change it’s rotation coordinats very slowly for each frame.
That’s may be good for static scenes with 1-depth ray tracing level.