So I love my subject line here, it says everything and nothing.

So recently I did this super simple setup [vimeo.com] to recreate an animation someone else had done in C4D, basically just for the fun of it. The setup was a mesh plus some noise on with a sphere at each intersection. Super simple.

But then I thought, what if I would want the same setup, but subdivided the mesh (in this case a resample) before the noise to get a more soft net-type look - but then I run into the issue of the copy node using all points for placing the spheres and there's no use for grouping the points before the resample as that kills the point order (thus the group).

My screenshot illustrate the issue pretty clear. As you can see, my crap ‘solution’ here was to split it up the mesh and use to copies of the noise - and though it works for this particular setup, it breaks the proceduralism. So unsurprisingly, crap setup is crap.

So, anyone have any ideas how to do this while keeping the proceduralism?

And no worries if someone has a VEX solution, just go for it, I'm a TP/Pflow/Thinkbox guy transitioning into Houdini so I'm used to working like that, just not in Houdini.

Personally, I don't see why splitting up the network breaks proceduralism. If you replace the torus with something else, then it should still work?

In anycase, I tried solving this by keeping the original points as an float attribute. Then when it gets resampled, the float attribute values on the resampled points gets interpolated, giving you a value like X.Y where Y is 0 if it was the original point. You can now create a group out of that and use it in the Copy SOP. I've attached my file (beware it's .hipnc).

edward
Personally, I don't see why splitting up the network breaks proceduralism. If you replace the torus with something else, then it should still work?

Well, this would be more a discussion about what is proceduralism than anything else, I just meant it as trying to keep it in one pipe (stream?) rather than splitting it up.

edward
In anycase, I tried solving this by keeping the original points as an float attribute. Then when it gets resampled, the float attribute values on the resampled points gets interpolated, giving you a value like X.Y where Y is 0 if it was the original point. You can now create a group out of that and use it in the Copy SOP. I've attached my file (beware it's .hipnc).

Sadly it works as long as you keep it as simple as you have here, but the resample breaks the polygon edges thus you gotta fuse those it after the resample and before the noise, but the fuse breaks your solution. I think you would have to use a per point attribute to get around it, like the memory function in Thinking Particles, though not sure how you would do that in Houdini. Yet.

But I solved it anyway. Seems just leaving it alone and revisiting it did the trick because how I solved it was ridiculously simple, hehe - I just used a copy of the copy setup and after the resample, I used the spheres (though scaled down) to create a group from the points inside and then I used that group in the final copy. So now I have one main object, one copied object and one force, which was my goal.

Edit: And I think you could have done this with an attributetransfer from the pre-resampled- & fused geo as well, gonna play around with that too.

I was here in 2023 From Edward solved solution, it's simple and work just need little bit of adjust
Don't know what was happen with the Resample SOP node in 2014 but in current time I am using Houdini 18.5 (H20 about release), Resample Sop node has two check box need to manage:

• Disable: Even Last Segment Same Length
  
• Enable: Maintain Last Vertex
  

That help to remain the poly line without any cut, no need fuse then no problem need to solve forward. Leave here for any one want to know in the future search.

I was doing Vellum string and need to pin the corner points from bunch of Connect Adjacent Pieces Sop through Resample. It works Look like a spider web to me
corner points group pin [media.giphy.com]