Palm rejection for stylus users: Documentation, possibly improvements

[xieve]

I've been trying to get a good handwriting experience with IPTSD on my Surface Pro 9 with a Slim Pen 2:

DisableOnStylus: Eliminates all random touches while writing, but can be irritating
DisableOnPalm: Almost there. Generally, writing feels okay and doable, but especially when leaving space before the next word (i.e. moving my palm across the screen), the canvas jumps a bit. Feels like this should be enabled by default, possibly?

These two parameters should definitely be mentioned in the wiki.

I've calibrated IPTSD like outlined in the wiki (those instructions are also slightly outdated, as the tool now writes three presets to current folder).

I would like to know what parameters are likely to influence palm rejection performance, and how I should go about tweaking them. Are some tweaks device-specific? If so, and if they are known, it'd be great to find them somewhere in the wiki.

Afaik, at least some Windows drivers do palm rejection heuristically based on stylus state, which I think IPTSD currently does not. I think that my Dell Latitude 5290 2-in-1 had a setting where you could change whether you're right- or left-handed, and it would reject touch points on the bottom left or right of the stylus more aggressively.

Simpler heuristics I made up on the spot:

• When a stylus is in the vicinity, activate DisableOnPalm: This makes a lot of sense to me. When you're drawing with the stylus, you probably want DisableOnPalm to be enabled, in other cases, you might prefer it being disabled for a better touch experience. ← This heuristic should be the default in my opinion.
• When a stylus is touching the screen, disable touch. ← I would use this, personally. I'm pretty sure I have never wanted to use touch with the other hand while the stylus touches the screen.
• When a stylus is present or touching the screen, adjust palm rejection to be more aggressive.
• When a stylus is touching the screen, accept only taps, but no gestures.

Generally, a "grace period" could be nice for these, i.e. after the stylus stops hovering, let the adjustments linger for a few seconds at most, in case it gets into range again. Scalar values could be eased (maybe even depending on the distance of the stylus to the screen? Not sure if that makes sense).