Add README with project page link
Browse files
README.md
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: other
|
| 3 |
+
tags:
|
| 4 |
+
- video-generation
|
| 5 |
+
- streaming
|
| 6 |
+
- interactive
|
| 7 |
+
- world-model
|
| 8 |
+
- minimax-h3
|
| 9 |
+
---
|
| 10 |
+
|
| 11 |
+
# StreamRender-H3: Interactive Coding Worlds with Real-time Streaming Diffusion Render
|
| 12 |
+
|
| 13 |
+
**Project page:** https://nvlabs.github.io/Sana/Sol-Engine/StreamRender-H3/
|
| 14 |
+
|
| 15 |
+
StreamRender-H3 splits an interactive world into a **Code Game Engine** and a **Streaming Diffusion Render**. The engine keeps the world state in code (maps, physics, rules) and streams a semantic map for every frame. A two-step streaming video model built on MiniMax-H3 Ref2VA renders those maps into photorealistic video as the player drives, with the look set by a reference image and a prompt.
|
| 16 |
+
|
| 17 |
+
- Two denoising steps per chunk in the distilled student.
|
| 18 |
+
- Diffusion render of 217.96 ms per streaming step on 8 GB200 GPUs, a 3.67× speedup over the PyTorch baseline.
|
| 19 |
+
- Fast Causal H3 VAE decodes 2 latent frames at 720p in 24 ms on one GB200, built by a self-improving agent loop.
|
| 20 |
+
|
| 21 |
+
Weights and code will be added here. See the project page for details.
|
| 22 |
+
|
| 23 |
+
## Citation
|
| 24 |
+
|
| 25 |
+
```bibtex
|
| 26 |
+
@misc{streamrenderh3_2026,
|
| 27 |
+
title = {StreamRender-H3: Interactive Coding Worlds with Real-time Streaming Diffusion Render},
|
| 28 |
+
author = {Yanzuo Lu and Tian Ye and Yitong Li and Shuchen Xue and Haozhe Liu and Song Han and Enze Xie},
|
| 29 |
+
year = {2026},
|
| 30 |
+
howpublished = {\url{https://nvlabs.github.io/Sana/Sol-Engine/StreamRender-H3/}},
|
| 31 |
+
note = {Project page}
|
| 32 |
+
}
|
| 33 |
+
```
|