idea: floating mode with eye-tracking

[abenz1267]

Hi,

would it be possible/feasible to have a floating-mode with eye-tracking?

Right now enabling floating mode captures the whole screen and adds 10000000000000000 labels, which is slightly overkill.

I think limiting the display area based on eye-tracking would be a rather cool idea.

What do you think?

[moverest]

You could try generating selection areas and pass them to the floating mode. The README tells you how to use the mode like that and you can find an example with the deleted wl-kbptr-sway-hints helper. I would be surprised if no one had already done a eye tracker in Python with OpenCV (or something akin) that could be reused for that.

Another way to do it would just be to specify the initial area with the -r (--restrict) parameter.