Regarding apple trackpads - For raw data, you get enough to do ellipsoid tracking, paired with 'state' information to help with palm rejection. This is my exper

Regarding apple trackpads - For raw data, you get enough to do ellipsoid tracking, paired with 'state' information to help with palm rejection. This is my experience from working with the multitouch private framework on macos - it's generally been enough to good enough work with and arguably prefer it rather than attempting to do any raw pixel-level tracking (which would tax the CPU) See: https://github.com/jnordberg/FingerMgmt https://github.com/jnordberg/FingerMgmt https://web.archive.org/web/20120323025118/http://www.steike.com/code/multitouch/ https://web.archive.org/web/20120323025118/http://www.steike... I haven't had time but should try to open-source whatever bits I've had over the years when I was bootstrapping thimblemac.com previous hn discussion on this: https://news.ycombinator.com/item?id=17547817 https://news.ycombinator.com/item?id=17547817 (where I also commented here: https://news.ycombinator.com/item?id=17551984 https://news.ycombinator.com/item?id=17551984 ) I haven't kept up with the windows ecosystem to look into the 'precision touchpad' data, assume it's something similar just delivered ~8 years later.