I’ve just started leaning Nuke and am interested in ways people create depth maps for live action footage. I see you can do things as simple as a roto node with it’s feather and SilhouetteFX’s application can do some cool stuff. Are there any other ways people commonly do this?
Thanks
Roto and ramps is the way to do most of it. There are some other techniques that can work in some circumstances, but they usually require plug-ins. I’ve been out of the stereo conversion world for a couple of years now, but back when I was doing that we were using a set of plug-ins from a company called YUVSoft that could estimate depth from focus or from motion. At the time they were using block-based motion estimation for the latter, but my understanding is that they were working on an optical flow algorithm that would be more accurate, though slower.
At about that same time, The Pixel Farm released PFDepth, which makes depth maps based on a 3d track. You might be able to do something similar with a dense enough point cloud from any other 3d tracker, such as Nuke’s native CameraTracker node.
You can also do some interesting things with displacement of cards in the 3d system.
Thanks for the great advice Bryan. I been playing around with CameraTracker and the point clouds trying to create depth maps. I’m new so it’s been with limited success haha. YUVsoft’s product looks awesome! Bit out of reach though. Might try the PFDepth trial.