Curve estimation upon point deletion is sometimes inaccurate and extreme

[ryan]

Hi,

Most of the time, when I delete a point, the resulting curve estimation is a godsend. However, sometimes, handles go crazy. See GIF below; sometimes handles go even more off-frame and I have to go chase them down.

TIA,
Ryan

[ryan]

Thanks for the background, and for giving it some thought! Will let you know if any ideas come to my mind as well.

[frederik]

yeah, what it does now is very simple and straight forward: a curve tries to reconstruct itself by extracting a t-factor (where on the curve) from the length of the handles and the position on the curve. This works in most cases, but in some it does not work out... curve fitting is no option as it will add more points.

I remember checking recursively if a curve is back in place, but the errors where a lot bigger with common point removing then fixing edge cases.

I would plead this is a case where it serves most situation but not all. (I know this is the easy answer, Ive opened an issue for when there is more time to check this out)