Developer trains a CNN to rasterize triangles from final renders alone
TheRPGGamerMan · reddit · 2026-11-12
A developer trained a small Triangle-DNN (500K params, 10 layers) in Unity for real-time rendering: vertex positions are projected into screen space, and the network must fill in the triangles on its own, learning rasterization purely from final rendered images. After 52K training iterations, it fills humanoid meshes quite accurately. A counterintuitive experiment on replacing the traditional raster pipeline with a network.
More from Research
- Qwen 3.8 27B on a laptop can read model weights, replicating the Weight-Space Programming paper — dl_weekly · 2026-11-12
- Trask: Transformer weight columns are concept detectors, but many concepts remain missing — iamtrask · 2026-11-12
- Dense rewards lose to standard GRPO at matched compute, arXiv paper finds — burny_tech · 2026-11-12
- Wayve unveils Learned Safety Envelope: end-to-end trained live risk prediction for AI driving — alexgkendall · 2026-11-12
- HipKittens turns one: breaking AMD's 'CUDA moat' for a multi-silicon future — simran_s_arora · 2026-11-12
- Why world model teams pay for gameplay with inputs attached: actions are the scarce half — KnowledgeOk7634 · 2026-11-12