A simplified example of what I hope to do is to have a video on loop of four people standing fairly still but not motionless (perhaps they are swaying slowly, or an arm might be swinging). When the viewer interacts with one of the people (e.g. by hovering a mouse cursor over the person) that figure will come to life and talk directly to the viewer with animated body gestures etc. When the viewer stops interacting (moves the cursor away from the figure), the figure will go back to how they were before.
I think that a very rough version of this is easily possible in jitter using a number layers of different video clips: cutting to the talking footage whenever the viewer interacts with the relevant figure. The problem is that when the video cuts back and forth there will be an awkward jolt due to small differences in the positioning of the figure. For example the hand will appear mid-swing, or the body mid-sway.
If this is even possible to do effectively in max, then I would be really grateful for any pointers on a way to approach this (any jit.objects I should be looking at etc) from kind folk who might have dealt with similar video editing projects.
I’d suggest to stay in GL world, using 4 [jit.gl.videoplane]s and [jit.gl.picker] for obtaining mouse information.
About cutting from one video to another, you may try reducing differences allowing video triggering only on particular frames. Also, a little crossfade would help.