I created a set of tools to generate plant placement to used in internally in houdini or externally via houdini engine. Plant placement is fully derived by climate space attributes (preciptation, sun occlusion, temperature and so on) and plant species optimality attributes. This in hopes solves the caveat of purely noise/topological based scattering techniques for many plants species. Go see slides here.

Slides

Results

Future Plans

As of currently, the seeding procdeure in the algorithm is causes the simulation to explode. I suspect the calculations doesn’t properly conserve resources. Futher investigation is needed.

To side step simulation time, a friend of mine suggested training a machine learning model. The idea is that when a climate and plant configuration is established, we could train a model to cheapily calculate plant placement for very large climate spaces.

References

Reference Images
Forza Horizon 5 (Game) - Playground Games
Far Cry 6 (Game) - Ubisoft
The Lion King Remake (Animated Feature Film) - Disney & MPC VFX

Textbook
Digital Design of Nature: Computer Generated Plants and Organics - Oliver Deussen, Bernd Lintermann

Current Methods
Synthetic Silviculture: Multi-scale Modeling of Plant Ecosystem https://storage.googleapis.com/pirk.io/projects/synthetic_silviculture/index.html

GPU-Based Procedural Placement in Horizon Zero Dawn https://www.slideshare.net/guerrillagames/gpubased-procedural-placement-in-horizon-zero-dawn

Foliage Atlas Textures
Quixel Megascans - https://quixel.com/megascans

Technology Used
SideFx Houdini - VEX, Hython, OpenCL
Unreal Engine 5 - Reference Renderer, Pathtracer