SpatialCrafter: Single Image World Modeling with Generative 3D Proxies

TL;DR: SpatialCrafter builds generative 3D proxies from a single image to enable consistent, controllable video generation.

News

• [2026-08] Paper released on arXiv.

Release Plan

• Inference code && Pre-trained model weights
• Dataset (115K scenes)
• Training code

Usage

Installation

# TODO

Inference

# TODO

License

This project is released under the Apache 2.0 License.

The dataset is released for non-commercial research purposes only.

BibTeX

@article{spatialcrafter2026,
  title     = {SpatialCrafter: Single Image World Modeling with Generative 3D Proxies},
  author    = {Author names omitted for double-blind review},
  journal   = {arXiv preprint},
  year      = {2026}
}

Acknowledgements

We thank the teams behind TRELLIS and WanVideo for open-sourcing their models, and the authors of SpatialGen for the website template.