HN Showcase
Back to browse
High-Res Neural Cellular Automata

High-Res Neural Cellular Automata

by esychology·Jun 17, 2026·102 points·18 comments
Visit ProjectView on HN

AI Analysis

●●●BangerWizardryZero to OneRabbit Hole

NCAs that heal when damaged, now at HD resolution with implicit neural decoders.

Strengths
  • •LPPN decoder breaks quadratic memory scaling, enabling arbitrary resolution output
  • •Three interactive demos prove real-time inference with damage regeneration
  • •Joint end-to-end training of NCA and decoder with task-specific losses
Weaknesses
  • •Research paper with limited production tooling or integration examples
  • •SIGGRAPH 2026 publication means limited community adoption so far
Category
AI/ML
Target Audience

Graphics researchers, ML engineers, procedural generation developers

Similar To

Self-Organizing Textures · Growing Neural Cellular Automata · NeRF

Post Description

Neural CAs model self-organizing pattern formation.

Now they can generate patterns at HD resolution in real-time, enabled by turning each CA cell into a Neural Field.

Try 3 demos: grow a pattern from a seed (and damage it, it heals), synthesize PBR textures that can regenerate, or create 3D textures like clouds.

Similar Projects