@alienmelon there's a project that does this on the OpenGL level https://web.archive.org/web/20100709233104/http://ogle.eyebeamresearch.org/
@justtesting @alienmelon I know! I haven’t actually used any of them for years, but I forget if they really are automated versions of your own idea; I.e. if the program isn’t rendering backfaces because of culling, the grabber can’t pull them. So you do still end up having to reconstruct a mesh and textures from multiple views.
I know at least one of worked by intercepting draw calls but I bet you could also do stuff like reconstructing a scene from the depth buffer etc
A Mastodon server friendly towards anti-fascists, members of the LGBTQ+ community, hackers, and the like.