It’s been nearly two years since Blender 2.92 officially released (February 2021) and users gained access to an all-new geometry nodes editor. What have people created since then, and what spaces are seldom explored?
Youtube Analysis
To understand the popular applications of geometry nodes, I’ve viewed the top 25 most popular Youtube videos under the search term: “Blender Geometry Nodes,” and tagged them with keywords. While 25 videos isn’t many and popular Blender videos are typically beginner tutorials, they should indicate trends and popular perception of geometry nodes.
What features are people using?
The most popular feature by far is instancing, more specifically the ‘distribute points on faces’ -> ‘instance on points’ combination to achieve a scattering effect. Although scattering is nothing new to Blender (achievable previously through particle systems and the ‘scatter objects’ addon), it’s a simple place to start testing geometry nodes and has been possible since their inception.
Another feature is geometry nodes’ ability to create grids. People have made 2D and 3D grids using the grid node, subdividing cubes, or have turned scans into voxels to make legos.
Other feature trends are scarce among the top 25 videos, although creators of motion graphics seem to enjoy the extrude node.
What are people depicting?
Their procedural nature lends geometry nodes to the creation of certain subjects. As Joey Carlino says, “If you can think of rules for how you would like things to be made, then you can probably make it with geometry nodes.” This means that people are making a lot of things that would be difficult (or usually, tedious) to model by hand but not difficult to imagine.
Landscapes fall easily into this category — they’re just a bunch of things scattered on a flat thing, or they can be made in more creative ways too. They have the benefit of being recognizable even when guided by a simple set of rules.
Buildings are also popular, although more complicated because their parts need specific orientations relative to each other, unlike landscapes which just need their scattered elements not to overlap. By building more and more procedural levels, one can even generate entire procedural fantasy cities, landscape included.
Some people, instead of instancing hand-modeled objects, prefer to let the generative-ness of their systems shine through and make abstract art too. If the rules in the geometry nodes system are interesting enough, they can become art all on their own.
Summary
The larger and likely less experienced Blender user base is mostly using geometry nodes for instancing and making landscapes and motion graphics. This isn’t to say that these things are drained of potential, but because they’re introductory exercises to geometry nodes they’ve been branched off of a lot and are relatively well-explored.
What spaces lack exploration?
Always a difficult question to ask, because knowing what spaces to explore is half the challenge. Here is a few personal observations:
- Curves: I’ve seen surprisingly little use curve nodes, despite how aesthetically appealing they are (especially in animation), but what I have seen is very impressive. Examples: hairstyling, curve-based VFX.
- Food: There are plenty of simple foods made with geometry nodes thus far, but I think geometry nodes have unlocked the potential to model more chaotic, complex meals.
- Utility: Although geometry nodes creations are impressive by themselves, geometry nodes can also take a step out of the spotlight to assist in the creation of less procedural assets. For example in creating interesting high-poly meshes to bake into textures, or efficient ways of modeling.
As a conclusion to this snapshot of creative history, with Blender 3.5 coming soon and a whole branch of simulation geo nodes with them, the future of animation with geo nodes looks exciting. Simulation-like animation would have made the previous list of underexplored spaces if we weren’t seeing promising examples already.
