If we're talking about the same xbox games, it's way less processor-intensive than the machine-learning done to detect human forms with that Kinect stuff (and doesn't need those lovely infrared-projecting deprh-field sensors that a kinect has either. Which come to think of it my phone does have, hah, funny story. And it's really funny to me that there's a bunch of other phones with this stuff now but just use it for like fancy blur on photos or trivial stuff like that, but I digress).
So anyway A) it does work and B) the image-processing requirements aren't that bad, and anyway at least as long as the table, camera, and projector stay in one place, can just be run once at the beginning but C) it also doesn't "magically work" like those old kinect games did; you have to print out some symbols and tape them to the table to get (cheap and fairly reliable) spatial references.
At the moment we also have to throw a towel over the table partway through the process for reasons.
If I were trying to do this more "magic and fancy," like, I know there are things I could do to try to figure out where the table is without printing out the markers, but it'd just be more work for me and yet ultimately produce something less reliable.
Edited by pokari, 18 August 2022 - 11:26 AM.