README.md

---
license: other
license_name: license
license_link: LICENSE
language:
- en
tags:
- image-to-3d
---
# SceneDreamer Model Card

<!-- Provide a quick summary of what the model is/does. -->

SceneDreamer learns to generate unbounded 3D scenes from in-the-wild 2D image collections.

Our method can synthesize diverse landscapes across different styles, with 3D consistency, well-defined depth, and free camera trajectory.

<tr>
    <img src="./teaser.gif" width="100%"/>
</tr>


## Model Details

### Model Sources

<!-- Provide the basic links for the model. -->

- **Repository:** https://github.com/FrozenBurning/SceneDreamer
- **Paper:** https://arxiv.org/abs/2302.01330
- **Demo:** https://huggingface.co/spaces/FrozenBurning/SceneDreamer

## Citation

<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->

**BibTeX:**
```
@ARTICLE {chen2023sd,
author = {Z. Chen and G. Wang and Z. Liu},
journal = {IEEE Transactions on Pattern Analysis &amp; Machine Intelligence},
title = {SceneDreamer: Unbounded 3D Scene Generation From 2D Image Collections},
year = {2023},
volume = {45},
number = {12},
issn = {1939-3539},
pages = {15562-15576},
keywords = {three-dimensional displays;solid modeling;semantics;cameras;training;rendering (computer graphics);geometry},
doi = {10.1109/TPAMI.2023.3321857},
publisher = {IEEE Computer Society},
address = {Los Alamitos, CA, USA},
month = {dec}
}
```

## Model Card Contact

Zhaoxi Chen: zhaoxi001@ntu.edu.sg