Matt Faw
January 27th, 2012, 07:24 PM
Dear all,
I'm creating a short doc project that includes a good deal of 3D CGI inserted into stereo live-action shots, and even more stereo live-action (shot in front of greenscreen) into CGI environments. The latter is accomplished by bitmapping one eye at a time onto a plane in 3D Studio Max, placing that plane into the CG environment, and rendering out using the 3Dhippie stereo rig.
I'm very happy with the results so far (especially considering how brand new I am both to 3D Studio Max and to stereo). However, I can imagine a solution that would speed up my workflow considerably, and it's hard to believe that someone else has not yet come up with it. Hopefully someone on this forum has heard of such a solution that could help me out.
Specifically what I'm wondering, is if any of the 3D CGI programs (or plug-ins) will analyze the depth of a stereo image, and create a simulated 3D environment (or object), based on that analysis? I've seen a beta program that creates a depth map from a 2D image (based on heuristics about horizon, perspective, etc.), so it seems like it should be very easy for a computer to come up with an accurate depth map based on a stereo image.
Applications for this are manifold: I could take a live stereo shot, and fly a CGI object around it, including passing behind objects in the foreground, because the computer should know where those foreground objects' edges are, and what their depth is. Also, I could take one of my live-action greenscreen actors, and create a 3D person to interact in a CGI world, without having to fool around with the planes, etc (obviously, that person would not be true 3D, in the CGI sense, because it wouldn't have a backside, but it could still have some perspective shift for camera moves). Plus, it seems like auto-rotoscoping and match-moving should be a breeze, since the computer could identify edges so easily.
Suggestions, anyone?
I'm creating a short doc project that includes a good deal of 3D CGI inserted into stereo live-action shots, and even more stereo live-action (shot in front of greenscreen) into CGI environments. The latter is accomplished by bitmapping one eye at a time onto a plane in 3D Studio Max, placing that plane into the CG environment, and rendering out using the 3Dhippie stereo rig.
I'm very happy with the results so far (especially considering how brand new I am both to 3D Studio Max and to stereo). However, I can imagine a solution that would speed up my workflow considerably, and it's hard to believe that someone else has not yet come up with it. Hopefully someone on this forum has heard of such a solution that could help me out.
Specifically what I'm wondering, is if any of the 3D CGI programs (or plug-ins) will analyze the depth of a stereo image, and create a simulated 3D environment (or object), based on that analysis? I've seen a beta program that creates a depth map from a 2D image (based on heuristics about horizon, perspective, etc.), so it seems like it should be very easy for a computer to come up with an accurate depth map based on a stereo image.
Applications for this are manifold: I could take a live stereo shot, and fly a CGI object around it, including passing behind objects in the foreground, because the computer should know where those foreground objects' edges are, and what their depth is. Also, I could take one of my live-action greenscreen actors, and create a 3D person to interact in a CGI world, without having to fool around with the planes, etc (obviously, that person would not be true 3D, in the CGI sense, because it wouldn't have a backside, but it could still have some perspective shift for camera moves). Plus, it seems like auto-rotoscoping and match-moving should be a breeze, since the computer could identify edges so easily.
Suggestions, anyone?