I’m mostly new to Geometry Nodes as I’ve only dabbled around it recently. I was wondering what features it needs the most since I am noticing that there are still a ton of limitations to what could open up a ton of possibilities. I’m not aware of Houdini so I don’t have a frame of reference on what Blender needs but I know that Blender can’t realistically compete. You could do a lot of stuff with Geometry Nodes currently but what features do you think that it really needs that you’re looking forward to?
Loops, and particle/physics nodes.
Built in simple nodes or node groups for artists that allow you to do cool things without the need for you to be a programmer or mathematician…
Just to give just one example among the many possibilities: Morphing Transition between meshes Node Group. Start mesh at input 1, end mesh at input 2. Dropdown menu to choose effects for transition: bounce, pixelated, gridded, etc.
I was going to add, a few more tutorials.
I dont mind making a few simple ones myself, however, until things settle down, new nodes, or new features on nodes appear so quickly Ive resisted the urge.
The current roadmap (along with dev. discussions) talk about loops, physics solvers (with a preference for realtime solutions), and hair/particles (with hair and particles becoming separate systems).
If all goes to plan, we should all be using a version of these things later this year.
Yes, I trawl the geonodes forum for added diffs most weeks!
Hum, I’m not sure it’s the good way to think about it…
That a bit unfair for both software to compare them like this, in one hand you have houdini that is node based from the beginning and that is as old as blender (Houdini 1.0 : 1996) , and blender GN is available in blender since… a year ? Given that Houdini is still evolving, good chances are that it will still be far more advanced at least for years. It’s not a matter of adding some features to GN to be on part. There are big features but also a lot of small things that will help to make it more powerful or easy to use.
I think it’s way less frustrating to think in terms of what it can do now rather than what it could do if …
At this point, if you know what you’re doing you can already do really a lot of things, it’s just a bit cumbersome and not as easy as it could be if everything was equally polished. The initial goal of GN for now is to do asset scattering and procedural modeling to some extent… but it’s flexible enough to do a lot of other things if you’re a bit clever. And of course, if you want to crash a giant UFO on a big city it’s better to stick to houdini for now .
Of course good thinks are coming that will unlock new possibilities and the road map is IMO very well though. But if you feel stuck right now, you’ll probably have that same feeling for a long time. Especially because so much areas needs to be taken care of, that even if all the devs resources where involved , still it will take a bunch of time. In the meantime if you’re interested , play with the tools , try to push them to their boundaries, you’ll see that you can do much more than you thought initially. And many issues can be worked around even with the small node set that we have now.
I’m sure Geometry nodes is already very capable. The only reason why I pointed it out is because I’m not sure what to further expect from geonodes. I’m not asking about how it compares with Houdini but how Blender’s Geometry Nodes can further improve and perhaps even beyond what Houdini currently offers in the area of procedural geometry.
Also as someone who works more on character models and rigging, I’m looking forward to features that will help with my workflow.
One thing I look forward to is a bridge between geometry nodes and shader nodes. The ability to use a lot of the same nodes in both editors.
Copy a set of math nodes from one to the other.
A texture coordinates input node for GN.
Or a random number node for the shader editor, now there is one for GN I can not understand why we can not have one for shaders as well.
There is a large etc.
Edit I do understand that this is all very new and will get more integrated in the future.
I wish that there was an easy way to visualize texture or color outputs while working in Geometry nodes. I always tend to make a specific set-up only to test and visualize things easier.
Actually, perhaps a much better way to pre-view things visually in 3D and not just limited to the Spreadsheet?
Yes I find myself doing that, setting up a dummy texture just to visualise the result and then having to totally redo it in GN, in my tests I was not able to copy-past the nodes, even the math nodes which should be the same!
I would also like to see a better way of controlling everything from the modifier panel.
For a better control of params that we can expose.
I have been wanting the devs. to make a special socket type for widgets that allow their exposure (in UI’s like group nodes). Ten years later, nothing has happened.
Though a few things that seemed to be just a pipe dream falling on deaf ears have actually been implemented lately (such as colorized connection lines), Everything Nodes has changed everything in terms of the team listening to feedback in the area.
We definitely need much better representation of exposed parameters and options to add stuff like separators, headers or collapsible panels. Hell, not even a boolean is represented as a simple checkbox, but a value spinner.
And please, while you’re at it, add the goddamn image input socket to the shader editor.
Agreed. I think Color can be used as an input now though?
I feel like those kinds of input will only be added if Blender will also allow you to have that much of customization with Custom Properties.
Solvers all the way !
I hope they’ll soon fix the Join Geometry node to not remove stuff like custom normals or UVs.
export of the spreadsheet to Json or txt.
easy export of any value to txt from geonodes.
Id like a live mouse x,y coords and a way to define a tablet area in the node area.
Be great for vj’ing, live recording some random shakes etc.
Actions nodes from the the NLA editor, fall off nodes without having to code custom fall off nodes, anything related to animation, more mesh and curve primitives, …
In general Nodes are tool low level to be directly useful for a lot of people aren’t programmers or mathematicians. For programmers, the field system feels backward and unintuitive compared to just … writing code…