I have created various failed systems for generating facial animations from tracked video, and now i consider looking for a 3d camera that is not IR dependent but rather stereo.
The reason I want to avoid IR is that the Kinect and such are extremely noisy and needs special software to read the “blob” that is your face and convert it to either primitive (cheap) or mediocre/good (expensive) facial poses.
There is only one, I repeat: ONE, software that can do this and that is Mycap Studio, but that program sucks as there is only automatic placement of trackers.
Various other software including Blender addons, rely on matching cameras to find 3d depth, but they do not work well when the cameras are parallel and would rather like cameras at 90 degree angle to each other. That is too bad as a face is rather round, and it is best to track from front view to see all markers in both cameras. Also matching the cameras in Blender to those from the shot is more of a hit and miss.
Is there any software, Blender addon or camera that lets you pair markers in both views, track it and move on with your production?