> Am 13.05.2021 um 20:02 schrieb Henning Hraban Ramm <texml@fiee.net>:
> 
> Thank you, that is what I was asking for!
> 
> Unfortunately, the deviation of edge points is too big – try my attached test file.

Hi Aditya et al.,

coming back to this: I found "fill" was not working, since do_sketchfill didn’t use sketchrandomized. That was an easy fix.

But I couldn’t fix the edge points issue: the less segments I use, the more some of the edge points deviate (see attachment with 10 segments; I used the setup from your theorems article).
Lower left and upper right are good, what happens with the others?

I just can’t understand the loop that splits the original path segments:

        for t = 0 step 1/sketch_segments until 1-1/sketch_segments :
            ((point       (t*arclength(p))                     on p) randomshifted s) .. controls
            ((postcontrol (t*arclength(p))                     on p) randomshifted s) and
            ((precontrol  ((t+1/sketch_segments)*arclength(p)) on p) randomshifted s) ..
        endfor

Would it make sense to check if a point is an edge point (without curve controls)?


Next question: Can I apply this to \framed and TABLE lines? (Probably only via backgrounds?)


Hraban