tencent/Hy-Nomas
AI model · https://huggingface.co/tencent/Hy-Nomas
First seen by Oriel on .
AI summary
Hy-Nomas is a video world generation model that supports multi-turn text interaction and camera control by generating continuous worlds segment by segment.
- A segment-autoregressive diffusion transformer combines Qwen/Qwen2.5-VL-7B-Instruct text features, camera conditioning, cached history, and a PixelUMM video decoder.
- Users can introduce objects, alter styles or scenes, describe actions, and control viewpoint or movement across successive video segments.
- The distilled inference configuration generates 720 × 1280 video at 24 FPS using eight DiT denoising steps and Euler ODE sampling.
- Hy-Nomas-Bench provides 280 test cases across four benchmarks for free-form interaction, camera navigation, combined controls, and spatial memory.
- Standard inference requires eight GPUs, while a low-memory configuration targets a single NVIDIA GeForce RTX 4090 using INT8 weights, CPU history archiving, and an FP8 decoder.
#video generation #diffusion transformers #camera control #multimodal models
Generated by gpt-5.6-sol. Check the original for details.
Metric trends
Last 30 days · Up to 120 recorded samples per source and metric, with reposts kept separate · First and latest retained
Points
Score reported by the source.
- Hacker News: 10, latest observation ; +7 over observed period since
Likes
Total likes reported by the source.
- Hugging Face Models: 123, latest observation ; +14 over observed period since
Downloads (30d)
Rolling 30-day downloads. A change in this value is not new downloads during the observed period.
- Hugging Face Models: 52,249, latest observation ; +14,144 over observed period since
Comments
Comment count reported by the source.
- Hacker News: 0, latest observation ; 0 over observed period since
Aliases
| Source | External ID | URL |
|---|---|---|
| Hugging Face Models | tencent/Hy-Nomas | https://huggingface.co/tencent/Hy-Nomas Read discussion |
| Hacker News | 50260747 | https://huggingface.co/tencent/Hy-Nomas Read discussion |