Is it able to do, if I:
1, Have an image-based video that has human’s motion in it, which I can get from anywhere like youtube, etc. In general, just a regular mp4 video, not something you have to take with kinect or stuff.
2, Turn the background of the video into black (or green etc). I leave only the motion object that can be visible in the video.
3, Put some virtual marker on the motion object, which I can do by using compositing program like After Effect or something.
4, Track those marker from the video (this is the part that I am wondering). Refine it, etc.