Aizatulin

You can also apply orientation along curve node on circle for stable frames.

Hi, you can extract one point per each class and generate random attributes on the points. After you've applied sort on the points you can map the result back onto the classes. random_class_sort.hipnc

Hi, it looks like a numerical problem. A quick workaround: apply transform after merge with a small uniform scale value like 0.00001

Hi, maybe as a workaround you can evaluate the (NURBS)surface itself by a grid, where the intrinsic uv is the same as the texture uv. Now you can fit the surface to get close to the original shape. seamless-nurbs-uv.v2_mod.hipnc

I'm expecting, that there will be an offset since there are fewer guide curves, but it would be nice if you can provide a simple picture (or sequence of picutres) of your final goal.

Hi, you can also try wire deform for this purpose. curve_wrap_wiredeform.hipnc

Here is a simple setup using intersection analysis. intersection_curves_displace.hipnc

Hi, is this coming close to what you are trying to achieve? InfiniteSpiralCarve_mod.hipnc

Hi, here is an approach rebuilding the uv based on the distance to a given center point (for example (0.5, 0.5, 0)). noise_question_mod.hipnc

Hi, you can use addpoint() function on each prim to add the vector attribute as point. You can remove the prims after this and you can also apply fuse to remove duplicates. You can also apply the add function on subsets (groups) if needed. prim_vector_attribute_add.hipnc

Hi, there are probably many ways to do this, but one way can using the expand group node applied on the left open edge ring. Combined with some other attributes it should give you desired result. Question_sort_mod.hipnc

You can also use scatter with relax iteration to get quite equidistant sampling. It is a bit slow but you can also use density attribute. poisson_on_bottle_setup_scattering.hipnc

Hi, you can also loop over the edges. If an edge is convex, the dot product of the direction between the center of the first and second (incident) primitive with the normal of the first primitive should be less than zero. edges_convexity.hipnc

Hi, you can try to set creases on edges for (catmull-clark) subdivision to get sharper lines, here is an example. creases_lines.hipnc

Hi, here are some ideas: create a referene for each prim onto its parent (if not already done) for each prim store all children(ref) into one array (using the reference) map this array to the u position of the parent curve (by using distance functions) [there are probably multiple ways -> but xyzdist() should be fine here] push all u values into one array including the existing points and rebuild the curve (with this array sorted by u) Here is simple setup including some aspects prim_insert_vertices.hipnc