Geometry nodes index of nearest - ignore self ID

Hello,
Would anybody happen to know, how I can exclude points which share the same ID from the search in the “index of nearest” node.
To explain better, I scattered points and instanced circles on them. I’m storing the initial point index onto the circle, to have for each point of the circle the same ID. What I now want to achieve is that when using the index of nearest, it doesn’t look at the points with it’s own ID but only at others. On the image for example, the points in black should only look at points in white.


I tried to connect a “not equal” with the captured ID plugged into top and bottom and connect this to the Group ID of the index of nearest node, but doesn’t seem to work.

Thanks a lot in advanced,
Stefan

That can not be done easily. You’ll require Alpha 4.0’s Repeat Zone and a very slow and convoluted network.

Instead, try see if you can rephrase the problem…

e.g.: Can you do your nearest indexing in the point domain before instancing the circles?

Good luck!

You can get a good enough SIMD solution just working in the point domain:

If you want something perfect, you’ll need to use Simulation nodes or the Repeat Zone and do something similar to this example from higgsas:

Good luck!

3 Likes

thank you very much for you reply. I should have mentioned that I am using Blender 4’s repeat. At the end I want to have the points on the circles to expand from the original point position (from which they are instanced) until they touch another circle in order to simulate some sort of cell growth. It is working as such, however if I have high density points with very small starting circles, they keep seeing themselves in the minimum distance range and don’t start growing after all.


Based on your reply I assume it’s just not possible out of the box to ignore the self group ID, bummer.

Cheers,
Stefan

3 Likes

I see…

Just a suggestion: Before creating the circles, do an initial pass on the points and find any pairs that are closer than some threshold and merge them.

Will that help?

ah damn, right !!! I was thinking to complicated and tried to relax them also using index of nearest. I’d really love to know how the group ID works, can’t figure it out even when using an id based on for example vertex group…