---
license: other
license_name: vggt-diff-research-license
license_link: https://github.com/chenkangjie1123/VGGT-Diff/blob/main/LICENSE
library_name: vggtdiff
pipeline_tag: image-to-video
base_model: Wan-AI/Wan2.1-I2V-14B-480P
tags:
  - novel-view-synthesis
  - multi-view
  - camera-conditioned-generation
  - vggt-omega
  - vggt-diff
  - "arxiv:2609.33253"
arxiv: "2609.33253"
---

# VGGT-Diff NVS 192x336 (100K)

This is the half-resolution independent-view scaling checkpoint trained on the
DL3DV clean-10K split (9,510 available scenes) for 100,000 effective optimizer
steps. It takes six posed source images and jointly predicts one or more target
views.

## Configuration

- Resolution: 192 x 336
- Task: independent-view novel-view synthesis
- Input: six posed source views and one or more target cameras
- VAE: source and target views are encoded independently
- Geometry: frozen VGGT-Omega features with camera conditioning
- Training data: DL3DV clean-10K (9,510 available scenes)
- Training: full-model fine-tuning
- Release checkpoint: 100K effective training steps

This release extends the original 18K fixed-compute setting to 100K steps.
Although individual checkpoints can fluctuate, the long-term evaluation trend
remains favorable and the 100K checkpoint is best across the reported
half-resolution image metrics. PSNR is approaching a plateau, while SSIM and
LPIPS continue to improve, so this release should not be interpreted as a
fully saturated endpoint.

## Usage

Use the `nvs` architecture mode:

```bash
python scripts/infer.py \
  --training-mode nvs \
  --checkpoint /path/to/vggt-diff.safetensors \
  --omega-checkpoint /path/to/vggt_omega_1b_512.pt \
  --height 192 --width 336
```

VGGT-Omega weights and Wan2.1 base-model files are separate dependencies and
are not redistributed in this repository.

## Links

- [Hugging Face Paper](https://huggingface.co/papers/2609.33253)
- [arXiv](https://arxiv.org/abs/2609.33253)
- [Code](https://github.com/chenkangjie1123/VGGT-Diff)
- [Project page](https://chenkangjie1123.github.io/VGGT-Diff/)

## License

Released under the non-commercial VGGT-Diff Research License. Third-party
models and datasets remain subject to their own terms.