Can seasoned Blender users can visualize and compose Geometry Node graphs in their minds? It seems to me that it would be pretty hard, simply because of their clunky design.
In my opinion, the only thing missing from Blender's amazing Geo Node system is icons, i.e. something compact and symbolic that one can store in their mental space as they would a word when composing a sentence.
I do also feel like a top down flow is more conductive to "process-oriented" or procedural design. I've done my fair share of procedural visual scripting with nodal systems such as UE4's Blueprints, PopcornFX or Blender's Geometry Nodes and none of them feel as "natural" as Houdini's.
It almost feels as though vertical lines occupy less space in my mind than horizontal lines do. Perhaps it's because our brains are hardwire to deal with lots of top down structured environments (plants, tree-trunks, other Sapiens).
> I do also feel like a top down flow is more conductive to "process-oriented" or procedural design. I've done my fair share of procedural visual scripting with nodal systems such as UE4's Blueprints, PopcornFX or Blender's Geometry Nodes and none of them feel as "natural" as Houdini's.
I agree with this. I've never used Houdini, but I have used Max/MSP and after using that, always found the horizontal visual programming systems very clunky and, as you say, they just feel like they take up a lot lot more space. Of course, Max's nodes are quite compact in comparison to something like Blueprints, but horizontal ones always are because each additional input/output makes the node one connection taller, while with a vertical layout you tend to fit many connections in the available width (of course, connection labels aren't clearly visible in that case, but I find that's something you get accustomed to).
Personally, I greatly prefer a vertical layout for visual programming systems.
Vertical node graphs like Houdini, Nuke or Katana tend to work for "pass the world" , whereas horizontal ones allow for more data specific passing instead.
It's just a difference of approach in what's going from one to the other. Houdini for example switches to horizontal data passing for VOPs.
I find horizontal data graphs easier to reason about at a glance, whereas vertical ones are easier to quickly prototype things.
I started with Blender, but have been working a bit in Houdini over the past quarter. I feel like both models work for both programs, and I find it pretty natural to work in a vertically oriented system in Houdini (outside of VOPs) and the horizontally oriented system in Blender.
I do also feel like a top down flow is more conductive to "process-oriented" or procedural design. I've done my fair share of procedural visual scripting with nodal systems such as UE4's Blueprints, PopcornFX or Blender's Geometry Nodes and none of them feel as "natural" as Houdini's.
It almost feels as though vertical lines occupy less space in my mind than horizontal lines do. Perhaps it's because our brains are hardwire to deal with lots of top down structured environments (plants, tree-trunks, other Sapiens).