Show HN: High-Res Neural Cellular Automata

Category: ai-ml

Tags: cellular-automata, neural-networks, texture-synthesis

Score: 6.5/10 (Innovation: 7, Technical: 8, Documentation: 6, Utility: 5)

This project implements high-resolution Neural Cellular Automata (NCA) for generating textures, morphologies, and radiance fields directly on mesh surfaces and volumes, accepted at SIGGRAPH 2026. It combines cellular automata with a learned local pattern producing network to upscale coarse cell states into fine details, offering interactive web demos and Colab notebooks. The approach is innovative for merging NCA with mesh-based rendering but remains niche for graphics and AI researchers.

Target audience: computer graphics researchers, machine learning engineers

Repository: https://cells2pixels.github.io/ · Jupyter Notebook

View on Hacker News