I just finally got around to playing around with these, and it’s one of those systems where there’s a lot of potential, but it can be difficult to image what sort of useful things can be implemented via geometry nodes (so far.) I was hoping to get some other people’s ideas about where geometry nodes are exciting for them.
For me, it seems like the current implementation is heavily focused on point instances, which isn’t something I’m into, but there are a few new things that I think are potentially useful:
Non-destructive joining (not booleaning) of geometry. I think I might have a few operations that this will simplify.
The ability to create some frequently used weight paint operations as reusable node groups. Setting up the ability to drop weights, invert weights, steal weights are all useful, but require multiple modifiers that are painful to set up. A single node group should simplify these operations considerably.
But other than those two, I’m having trouble understanding what I might do with geo nodes. It seems very WIP (color ramps without VCol attributes?)
For now geometry nodes are awesome for scattering, but it get’s better in 2.93.
I think it’s like with scripting , the more you do the more ideas you get. It’s quite nice to use them with other modifiers.
A very simple use case that I found the other day with 2.93 : I was modeling something with skin modifier and I wanted to assign another material to some parts of the mesh and that was not possible.
I made a very simple geo-node tree that allow me to set a material based on a vertex group.
You can also do more advanced arrays and add some randomness.
I think that was intentional and how geometry nodes were first implemented, however the list of nodes is growing rapidly and the potential applications will likely grow exponentially over the next few releases.