Bumping this up for visibility, as I’m having a similar issue.
In geometry nodes, I can resample a curve, but I can’t resample a curve to conform to the original shape where the sample isn’t evenly distributed. I see people using map ranges as a factor bias, but that only leans to distributing points toward one end more. I need something that resamples to the best overall shape of my curve. Added a picture for clarification!
Also the curve is just the simplest example for explanation purposes. The actual curves would be more dimensional than this 2D version. Would be nice if this could be added to Blender.
This is a bit tricky to do, basically you shouldn’t look into resample, but rather it’s about deleting points based on some rules.
But still it’s a bit complex, as in your example, you can keep points with their surrounding edges forming an angle of 90° to keep the corners , but to reduce the pointcount of the arc you’ll need another rule, more closer to the resample …
Probably you’ll have to separate these two operations , maybe by separating straight lines first and process them separately from curved ones…
I currently don’t have time to look into that, but at least this might give you pointers to move forward !
I just updated my Split Bezier node-group (Split Bezier(3.6)). You can use that in conjunction with the node-deletion mentioned by @sozap to accomplish what you’re after:
Basically here I’m splitting the curve into 3 curves on points where either of the handles are of type vector…
… then doing the resample, and then cleaning up by deleting any points that lie on a straight edge. Finally converting to mesh to merge vertices and converting back to a single curve.
