I’m always amazed at how high the bar for 3D graphics in film rises from year to year, and mostly I’ve been able to keep up with the times and understand how something was accomplished, but there is one area which has always puzzled me, and that is reflections.
Now certainly, I can understand putting 3d objects into live reflections using matting and spot-on camera tracking, but the opposite is a stunning illusion, putting live-action reflections onto a CGI object.
I simply can’t understand how it can be done. The closest thing I’ve come to is having a panoramic camera positioned where the 3D object is meant to be in the scene, and using the video from that camera to fake the reflections, but for some reason that doesn’t seem right, but it’s the only answer I can come close to short of building the entire set in CG, which sounds like a bit much, even for the degree of realism mentioned.
Does anyone know exactly how this feat is accomplished? It’s the one aspect of CG/live action integration that has truly stumped me.