README.md

---
license: cc-by-nc-4.0
library_name: pytorch
pipeline_tag: image-to-3d
base_model: microsoft/TRELLIS.2-4B
base_model_relation: adapter
tags:
  - applied-intuition
  - 3d-reconstruction
  - scene-generation
  - single-image-to-3d
  - mesh-generation
  - trellis
  - lora
  - arxiv:2610.08790
paper: 2610.08790
---

<p align="center">
  <img src="https://huggingface.co/AppliedIntuitionResearch/BuildRome/resolve/main/assets/icon.svg" width="120"/>
</p>

# Building Rome from a Single Image

<p align="center">
  <a href="https://arxiv.org/abs/2610.08790"><img src="https://img.shields.io/badge/arXiv-2610.08790-b31b1b.svg" alt="arXiv"></a>
  <a href="https://build-rome.github.io/"><img src="https://img.shields.io/badge/Project-Website-blue" alt="Project Page"></a>
  <a href="https://github.com/Applied-Intuition-Open-Source/BuildRome"><img src="https://img.shields.io/badge/GitHub-Code-black?logo=github" alt="Code"></a>
  <a href="https://creativecommons.org/licenses/by-nc/4.0/"><img src="https://img.shields.io/badge/License-CC%20BY--NC%204.0-lightgrey" alt="License"></a>
</p>

Building Rome generates a complete metric 3D scene mesh — visible surfaces and plausible occluded
geometry — from a single photograph, indoors and outdoors. It fine-tunes the TRELLIS.2 object
generator to work on scenes: a monocular depth estimate (MoGe-3) lifts image features onto the
observed surfaces, the scene is covered with chunks that grow with distance from the camera, and
the chunks are generated one at a time with latent outpainting and stitched into one mesh.

> ⚠️ **Research release.** This model is released for research purposes under the terms below. It is not a production system and does not include any Applied Intuition proprietary data, product code, or production checkpoints.

**Built with DINOv3.**

## Model Details

| | |
|---|---|
| **Developed by** | Applied Intuition — AI Research |
| **Model type** | LoRA adapters and depth-conditioning branches for the TRELLIS.2 sparse-structure (SS) and shape (SLat) flow transformers |
| **Paper** | [Building Rome from a Single Image](https://arxiv.org/abs/2610.08790) |
| **Code** | https://github.com/Applied-Intuition-Open-Source/BuildRome |
| **Project page** | https://build-rome.github.io/ |
| **Base model** | [microsoft/TRELLIS.2-4B](https://huggingface.co/microsoft/TRELLIS.2-4B) (MIT), downloaded separately |
| **Training data** | Indoor scenes from [SAGE-10k](https://huggingface.co/datasets/nvidia/SAGE-10k) and [Infinigen](https://github.com/princeton-vl/infinigen), and about 4,000 synthetic outdoor scenes (see the paper) |
| **License (weights)** | [CC BY-NC 4.0](https://creativecommons.org/licenses/by-nc/4.0/) |
| **Contact** | GitHub Issues on the code repo |

## Checkpoints

The files hold only the fine-tuned parameters; the inference code adds them to the public
TRELLIS.2 base weights.

| Checkpoint | Description | Size | Link |
|---|---|---|---|
| `ss_delta.safetensors` | Sparse-structure (occupancy) model: LoRA r32 adapters and depth-lift / clearance branches, 115M params. | 438 MiB | [ss_delta.safetensors](ss_delta.safetensors) |
| `slat_delta.safetensors` | Shape model: LoRA r32 adapters and depth-lift branch, 66M params. | 251 MiB | [slat_delta.safetensors](slat_delta.safetensors) |

SHA-256 checksums are in [SHA256SUMS](SHA256SUMS).

## Intended Use & Limitations

**Intended use:** research on single-image 3D scene reconstruction and generation.

**Out of scope / limitations:**

- Outputs are geometry only (no texture). Metric scale and camera come from the MoGe-3 depth
  estimate, so depth errors carry into the scene.
- Occluded regions are plausible completions, not reconstructions of the true scene.
- Meshes are not guaranteed to be watertight and can contain holes, as with TRELLIS.2.
- Requires a CUDA GPU; a 98 m street scene (4 depth bands, 15 chunks) peaks at about 16 GiB and
  takes about 6 min on one A100.

**Commercial use is not permitted** under CC BY-NC 4.0.

## How to Use

```python
from build_rome import ScenePipeline

pipe = ScenePipeline.from_pretrained("AppliedIntuitionResearch/BuildRome")
scene = pipe("photo.jpg")
scene.save("results/photo")  # scene.glb, layout_bands.json, layout_topview.png, spiral.mp4
```

Command line:

```bash
hf download AppliedIntuitionResearch/BuildRome --local-dir checkpoints
python inference/image_to_scene.py --image photo.jpg \
    --ss-ckpt checkpoints/ss_delta.safetensors --slat-ckpt checkpoints/slat_delta.safetensors --out results/
```

The pipeline also downloads [TRELLIS.2-4B](https://huggingface.co/microsoft/TRELLIS.2-4B),
[TRELLIS-image-large](https://huggingface.co/microsoft/TRELLIS-image-large),
[DINOv3 ViT-L/16](https://huggingface.co/facebook/dinov3-vitl16-pretrain-lvd1689m) (gated: accept
its license on Hugging Face first) and [MoGe-3](https://huggingface.co/Ruicheng/moge-3-vitg). Full
installation and usage instructions: see the [GitHub repository](https://github.com/Applied-Intuition-Open-Source/BuildRome).

## License

The weights in this repository are released under
[CC BY-NC 4.0](https://creativecommons.org/licenses/by-nc/4.0/), **non-commercial use only**
(full text in [LICENSE](LICENSE)). They are used together with separately downloaded models that
keep their own terms:

- [TRELLIS.2-4B](https://huggingface.co/microsoft/TRELLIS.2-4B) and
  [TRELLIS-image-large](https://huggingface.co/microsoft/TRELLIS-image-large): MIT.
- [DINOv3](https://huggingface.co/facebook/dinov3-vitl16-pretrain-lvd1689m): the
  [DINOv3 License](https://ai.meta.com/resources/models-and-libraries/dinov3-license).
  Built with DINOv3.
- [MoGe-3](https://huggingface.co/Ruicheng/moge-3-vitg): MIT.

Training data: [SAGE-10k](https://huggingface.co/datasets/nvidia/SAGE-10k) (Apache-2.0),
[Infinigen](https://github.com/princeton-vl/infinigen) scenes (BSD-3-Clause generator), and
synthetic outdoor scenes produced with the pipeline described in the paper. No training data is
included in this repository.

## Citation

```bibtex
@article{yenphraphai2026rome,
  title   = {Building Rome from a Single Image},
  author  = {Yenphraphai, Jiraphon and Li, Fang and Xu, Tianshuo and Meng, Depu and Herau, Quentin and Hu, Yihan and Yeh, Raymond A. and Zhan, Wei},
  journal = {arXiv preprint arXiv:2610.08790},
  year    = {2026}
}
```

## Acknowledgments

This model builds on [TRELLIS.2](https://github.com/microsoft/TRELLIS.2) and
[TRELLIS](https://github.com/microsoft/TRELLIS) (Microsoft, MIT),
[MoGe](https://github.com/microsoft/MoGe) (Microsoft, MIT) and
[DINOv3](https://github.com/facebookresearch/dinov3) (Meta, DINOv3 License). Training uses
scenes from [SAGE-10k](https://huggingface.co/datasets/nvidia/SAGE-10k) (NVIDIA) and
[Infinigen](https://github.com/princeton-vl/infinigen) (Princeton). We thank the authors for
making their work available.