NVIDIA Research · Streaming Game Rendering

StreamRender-H3

Part of NVIDIA's Sol-Engine work, after SoL-Refiner (week 41). A three.js racing game handles maps, physics and rules and streams a 13-class semantic map to a diffusion renderer built on MiniMax-H3 Ref2VA, which turns it into photorealistic video, with a reference image and prompt setting the look so one engine stream can be restyled. Training moves through teacher forcing, Resampling Forcing and distribution-matching distillation down to two denoising steps per chunk. Render time is 218 ms per streaming step on 8 GB200s, down from 799.7 ms, and a fast causal VAE decodes two latent frames at 720p in 23.8 ms. Code is in NVlabs/Sana; no end-to-end frame rate is stated.

Open the source ↗

Similar Video & Animation releases

  • OverSeer

    University of Oxford and Meta (Runjia Li, Andrea Vedaldi, Shikun Liu et al.) · Multi-agent world model · Week 45, 2026

  • MatAnyone-E

    Shuyun Wang, Shangchen Zhou, Peiqing Yang, Yuekun Dai, Xin Yu and Chen Change Loy · Efficient 4K matting · Week 45, 2026

  • Genie 4

    Google DeepMind · Interactive world model · Week 45, 2026

  • FastH3 Omni speed results

    Hao AI Lab @ UCSD · Fast Video Model · Week 44, 2026

  • Hy-Nomas

    Tencent Hunyuan · World Model · Week 44, 2026

  • Streaming Code World Model

    Tingting Liao · Causal World Streaming · Week 43, 2026

More from NVIDIA Research

  • SoL-Refiner

    NVIDIA Research · 4K Video Refiner · Week 41, 2026

  • ArtiFixer

    NVIDIA Research · 3D Generation · Week 26, 2026

  • D-Rex

    NVIDIA Research · 3D Generation · Week 19, 2026

  • SLM Agents Research

    NVIDIA Research · Research Paper · Week 3, 2026

Published by Quanta AI Labs, an AI agency in India. Browse every week in AI News Weekly.