[2506.22899] Neural Cellular Automata: From Cells to Pixels
Abstract:Neural Cellular Automata (NCAs) are bio-inspired dynamical systems in which identical cells iteratively apply a learned local update rule to self-organize into complex patterns, exhibiting regeneration, robustness, and spontaneous dynamics. Despite their success in texture synthesis and morphogenesis, NCAs remain largely confined to low-resolution outputs. This limitation stems from (1) training time and memory requirements that grow quadratically with grid size, (2) the strictly local propagation of information that impedes long-range cell communication, and (3) the heavy compute demands of real-time inference at high resolution. In this work, we overcome this limitation by pairing an NCA that evolves on a coarse grid with a lightweight implicit decoder that maps cell states and local coordinates to appearance attributes, enabling the same model to render outputs at arbitrary resolution. Moreover, because both the decoder and NCA updates are local, inference remains highly parallelizable. To supervise high-resolution outputs efficiently, we introduce task-specific losses for morphogenesis (growth from a seed) and texture synthesis with minimal additional memory and computation overhead. Our experiments across 2D/3D grids and mesh domains demonstrate that our hybrid models produce high-resolution outputs in real-time, and preserve the characteristic self-organizing behavior of NCAs.
Abstract:Neural Cellular Automata (NCAs) are bio-inspired dynamical systems in which identical cells iteratively apply a learned local update rule to self-organize into complex patterns, exhibiting regeneration, robustness, and spontaneous dynamics. Despite their success in texture synthesis and morphogenesis, NCAs remain largely confined to low-resolution outputs. This limitation stems from (1) training time and memory requirements that grow quadratically with grid size, (2) the strictly local propagation of information that impedes long-range cell communication, and (3) the heavy compute dem
saved by
related reading
- [2607.02954] Microcosmos: Reimagining Artificial Life for the GPU Eraarxiv.org
- Towards self-organized controlavariengien.github.io
- Learning Sensorimotor Capabilities in Cellular Automatadevelopmentalsystems.org
- pdfopenreview.net
- Digital Ecosystems: Interactive Multi-Agent Neural Cellular Automatapub.sakana.ai
- Self-Assembly Gets Automated in Reverse of ‘Game of Life’ | Quanta Magazinequantamagazine.org
- [2603.10055] Training Language Models via Neural Cellular Automataarxiv.org
- [2006.11239] Denoising Diffusion Probabilistic Modelsarxiv.org
- Probing the rules of cell coordination in live tissues by interpretable machine learning based on graph neural networkspmc.ncbi.nlm.nih.gov
- [2608.13335] Neural Quadratic Forms: A Unified Minimal Model for Sudden Learning and Scaling Lawsarxiv.org
- D-NeRF: Neural Radiance Fields for Dynamic Scenesopenaccess.thecvf.com
- DreamGaussian: Generative Gaussian Splatting for Efficient 3D Content Creation | PDFarxiv.org