---
license: apache-2.0
task_categories:
- image-to-3d
---

# SparseCam4D: Spatio-Temporally Consistent 4D Reconstruction from Sparse Cameras

This repository contains the demo dataset for **SparseCam4D**, a framework for high-quality 4D reconstruction from sparse and uncalibrated camera inputs.

[**Project page**](https://inspatio.github.io/sparse-cam4d/) | [**Paper**](https://arxiv.org/abs/2603.26481) | [**GitHub**](https://github.com/inspatio/sparse-cam4d)

## Data Layout

The expected data layout for the dataset is as follows:

```
balloon1/
├── depth/
│   ├── cam01/              # per-frame depth maps for training camera cam01 (*.npy)
│   ├── cam06/              # per-frame depth maps for training camera cam06 (*.npy)
│   ├── cam10/              # per-frame depth maps for training camera cam10 (*.npy)
│   ├── cam01.mp4           # depth video visualization
│   ├── cam06.mp4
│   └── cam10.mp4
├── images/                 # all input images, named as <cam>_<time>.png
├── preprocess/
│   ├── time_0000/
│   │   ├── diffusion/      # pseudo-view images generated by ViewCrafter at t=0
│   │   └── sparse/0/       # COLMAP sparse reconstruction at t=0 (cameras.bin, points3D.ply, ...)
│   ├── time_0001/
│   │   └── diffusion/      # pseudo-view images at t=1
│   └── ...                 # time_0002 ~ time_0099, each with diffusion/
├── sfm_transforms_extend.json   # camera intrinsics + extrinsics for all views and timestamps
├── vc_roma_sfm_300.ply          # initial point cloud (SfM + RoMa dense matching)
├── transforms_train.json        # camera poses for training split
└── transforms_test.json         # camera poses for test split
```

**Depth maps** are estimated by [Video Depth Anything](https://github.com/DepthAnything/Video-Depth-Anything) on the training-camera videos.

**Pseudo-view images** under `preprocess/time_*/diffusion/` are synthesized by [ViewCrafter](https://github.com/Drexubery/ViewCrafter) from training cameras to cover additional viewpoints at each timestamp with sparse camera poses estimated by [VGGT](https://github.com/facebookresearch/vggt).

## Sample Usage

### Training

To train the model on this dataset, edit the `source_path` and `model_path` fields in the config file, then run:

```shell
python train.py --config configs/nvidia/balloon1.yaml
```

### Rendering and Evaluation

After training and performing pose alignment, you can render and evaluate using:

```shell
python render.py --config configs/nvidia/balloon1.yaml --skip_train --iteration 30000
```

## Citation

```bibtex
@article{pan2026sparsecam4d,
  title={SparseCam4D: Spatio-Temporally Consistent 4D Reconstruction from Sparse Cameras},
  author={Pan, Weihong and Zhang, Xiaoyu and Zhang, Zhuang and Ye, Zhichao and Wang, Nan and Liu, Haomin and Zhang, Guofeng},
  journal={arXiv preprint arXiv:2603.26481},
  year={2026}
}
```