Geometry Nodes in Blender 3.0 - Status Update

If anyone needs any clarification on the current design, I will be glad to elaborate further.

That said, I know some people want to talk about the Get/Set nodes, and this is not the point of this post. The focus here is on getting everyone aligned for the 3.0 release, so everyone can help with testing, new files, documentation, …

Whether to support those nodes is a topic that can be revisited for 3.1. For now, there is enough on the agenda. And the current design of shareability is still the driving force of the system.


Please, no polls here.


Could you perhaps explain why ? knowing what the community thinks about this removal do not interest you? :frowning_face_with_open_mouth:

1 Like

Hi, can you share what you are trying to do with geometry nodes/fields at the moment?

As I mentioned, this post is about clarifying the current design. And for people to show what they can already do with it (and will be able to do in 3.0). I don’t see how the poll can help with that.

People are free to have this conversation elsewhere, of course. But I don’t want to scare away people who may be interested on the current system, but who won’t stick around if this turns into a lengthy debate about possible 3.1 only targets.

1 Like

perhaps because some users might not be happy about this design change


18 posts were merged into an existing topic: Named attribute nodes in 3.0

I tried to create something like “shape keys” with GN nodes.

There are two modifiers. M1 and M2 in the modifier stack. Custom attribute “my_cache” used to store vertex positions of the mesh.


M1 stores position to attribute “my_cache”. M2 reads a position value from “my_cache” and blends it with the current value of position.

M1 …

and M2 …

The idea was to use M1 to store the vertex positions into my_cache. Then hide the modifier and edit the mesh.
After, I would try to use ‘my_cache’ in M2 to blend the shapes.

It did not work as expected, because the value of my_cache is zero that moment I hide M1. But the little experiment also shows me some other things.

  • Multiple group inputs could help to get a clean node graph. (A very good workflow idea someone shared in the devtalk forum)
  • One also might try to use multiple group outputs for the same reason. Splitting outputs, to get a cleaner node graph. But this seems not to work. Using more than one group output gives wrong results for output attributes. (Edit, currently I am doing a mess with output nodes. May multiple output nodes be dangerous?)
1 Like

I think shareability is as important as to give users control. Also, for anyone starting out with GN, I think it’s important to have some initial path to start over. It makes me a little concerned me that GN looks, at the moment, to be target only towards the technical artist or director. I mean, some build in node groups with some common user cases would make it easier to people to start over, maybe a call for content for the 3.0 realease @dfelinto


I would be happy to contribute to a 3.0 asset pack. I’m sure there are many other experienced users that are willing to help too.

@LeonardSiebeneicher, how is your test expected to work? No changes were made between the store and set.


This is a good one! However I think the “texture pointer” idea in Luxcore is great to have. It’s basically a method to define a procedural or image texture in its own node editor, and can be shared and used globlly. User can change it in the node editor and anywhere it’s been referenced will sync with it automatically.


I thought, storing happens if I unhide M1.
If the stored attribute prevailed after hiding M1, I thought it could work like a blend shape. When hiding M1, the attribute gets zero (not initialized).

I expected GN custom attributes to act like caches. Probably, current custom GN attributes are for a different purpose.


That’s a “variable” definition, and I also expect this in some situations, not all situations.

@dfelinto, pardon me please. This page was not meant to become a discussion about the removal of get/set nodes, and I was one of those who helped push it in that direction.

However, I believe this situation is one that should be addressed soon. Please take the calls of the community into consideration.

@Laulau, if everything in blender development was settled on by a poll, blender 2.93 would still look like blender 1.0.

1 Like

Hi, as requested above, I moved discussion of named attribute nodes to a dedicated post. That is not because I don’t think it’s important, but because I think it’s best to respect the intention of this post.


There are still nodes missing the documentation in the user manual. I added a note to direct artist to the release notes for now.


Hi, I am willing share this demo file for the new curve support 3.0. Please affirm that this can be accepted so that I can go ahead with the cleanup process. (it also uses the old point instance)


This looks really nice, it will be great to include in the demos. (bonus if you can make them look nice when rendering).

Besides cleanup, can you please also add a README with instructions on how to explore the file, and basic credits and the corresponding CC license?

That said the new demo files should all have only the new nodes. So in your case you need to replace the Point Instance with the Distribute Points on Faces nodes.

Thanks, I will have a look at other demo files for examples. It also needs the subdivision surface surface node which does not seem to be available right now in master.

Of course

Hey Dimitri,

I understand your position, but it’s a procedural system to make custom modifiers, it will be pre-requisite that there is some amount of knowledge involved. There are already a ton of tutorials and courses to get people started, and they are at a level that anyone can follow.

But it’s totally ok for many users not to use the Geometry nodes themselves, they can use assets that other people made.

It’s like making apps for your smartphone. Yes the apps need to be easy, but what about the environment where you make the apps? If that is dumbed down too much, then you will only get dumb apps.

I’m not saying to go overboard and make it super-convoluted so only people with PHD in math and computer science use it… It needs to come with a certain ease of use so people can really be creative with it, and it can be inclusive.
I think currently it’s close to having a good balance. Anyone can play with the nodes and get something out of it, while still allowing more elaborate setups to happen.


Maybe its useful?

It is a fractal, created from a cube. Chaining the same GN modifier is used as recursion steps for the fractal.

The GN Nodes are simple

The Blender file …