I am trying to procedurally insert edge loops. So basically what i want to do is to take an edge loop from a model, make a curve out of it, resample it and use the points of that curve as a insert location for new edge loops. And i am not sure how to do the last part.



So how can i take these points and use their location as a loops insert location for the original model?

Attachments:
loopsinsert.hip (134.2 KB)
Screenshot_1.png (344.0 KB)

Attachments:
loopsinsert_modX.hipnc (164.9 KB)

Aizatulin
Hi,

here are some ideas. Edge divide should do the job usually, but it some cases it messes up, but polygons can be rebuild manually using the points from edge divide. Once you have the polygons you can also capture the points by the curve and optionally rebuild it more smoothly.

Thank you for the effort.
This works pretty good, but there is one major problem with it( unless i am missing something ).
It splits every edge the same amount times. While with curve, resample point insert can vary, based on the length of each segment. That's the reason i wanted to use curve points as a insertion reference, because i need that feature.



When i use insert edge loop from the polygon toolbar, it adds a polysplit node with split location coordinates. Is it somehow possible to take the curves points coordinates, and use them in the polysplit node ?

Oh ... and i don't care about the smoothing of the mesh, just loop insertion control. Preserving the initial shape of the model is what matters to me, so the original edge loops and edges should stay.

Attachments:
Screenshot_1.png (442.8 KB)

Attachments:
loopsinsert_modXA.hipnc (221.9 KB)

Aizatulin
If you want equal point count on each curve, you can't resample the curves independently (to maintain quads). So you might probably resample one curve (the first curve). You can use the length between neighbours for example to determine the point count. But if you want to keep the shape anyway. Here is another file with some modifications merged from other files.

Oh no, not the dreaded skin node!
Seriously though, the reason i abandoned the "curves to skin" approach, discussed in the odforce forum is because things sometimes worked, sometime didn't. The skin node is very unreliable for something like this. That's why i decided that just splitting the geo into loops was a better idea.

For the last example you've posted it worked pretty well for the models i've tried. Tried maybe more than 10 basic cylinder/square type models. The only problem was that i had to manually switch the v wrap in the skin node to on or off because in some cases i had gaps that had to be closed. Fixed that problem by leaving it at off and then used some group gymnastics to procedurally select the gap edges and fill them with the "crosssectionsurface" node. Things were working pretty well.

Then i've tried something pretty basic like the one in the picture:



And the familiar result that ruined my hopes.

Basically i am trying to make a tool that should be fundamental for my module based modeling approach.
1.Create some simple shape.
2.Run it trough the algorithm to get some level of control over the edge loops.
3.Then use those as a base for path deform/sweep ... or other procedures.

I am gonna play around with your latest example a bit more, but i doubt it's gonna work with the skin node. That's why i wanted to use the polysplit node and just transfer the resampled curve point locations to the node.

Attachments:
Screenshot_1.png (442.9 KB)

Attachments:
loopsinsert_modXA2_v2.hipnc (337.3 KB)

Attachments:
loopsinsert_modXA2_v2_loopfix.hipnc (345.4 KB)