I have an animated camera going through a large model. How could i remove all geo that is never seen by the camera for the entire frame range?

i know this needs a solver for one and also maybe the intersect function. Shooting from the camera to each point if we find an intersection then its occluded but later it might not be. If its not then hit=-1. So if i multiplied every frame the hit value with the new one if the end result is negative it means the camera saw the geo at least once so i keep it.

Does this sound reasonable or is there some other easier way to do this?

This -> ? https://www.youtube.com/watch?v=yweaCsLepXE [www.youtube.com]

yeah that method assigns a uv value even to occluded geo unfortunately as long as it is in the frame somewhere

I did this many times in production where I generally scatter points on faces to get more uniform raytracing towards the camera in a solver sop and if it ever not hit anything for any of its points, then I keep those faces.

animatrix_
I did this many times in production where I generally scatter points on faces to get more uniform raytracing towards the camera in a solver sop and if it ever not hit anything for any of its points, then I keep those faces.

Something like this?