I’m not saying render the scene. I’m saying take your measurements from where the camera is (or would be).

It would require no extra infra or compute.