Cafca Synthetic Face Dataset
\-
Cafca Synthetic Face Dataset
A large-scale, diverse multiview face dataset for avatar reconstruction and animation, face parsing, synthetic-to-real transfer learning, animation, relighting, camera estimation, and more.
Please use the links in the navigation bar to quickly jump to sections.
Project Page Download (GitHub) Download (Hugging Face) Quickstart Colab
Dataset Facts
- 1,755 Mio. images
- 1,500 diverse identities x 13 expressions x 3 environments x 30 views
- Resolution 512x512
- Annotations for 14 semantic regions
- Perspective OpenCV cameras
- Pseudo-GT keypoints and FLAME 2023 annotations & meshes (fitted with VHAP)
- CC BY-NC-SA 4.0 License
- DAViD (ICCV 2025): a synthetic dataset with 100K images for each of these body regions: face, upper-body and full-body.
- BEDLAM (CVPR 2023): a synthetic dataset with 10K image full-body sequences.
- IMAvatar (CVPR 2022 oral): a synthetic face dataset with 10 identities.
Dataset Samples
Identity
Environment
Expression
Show Image
I'm feeling lucky
###### Image
###### VHAP Fitting Result
Related Datasets
The following synthetic datasets could also be useful:
Acknowledgements
We thank Shenhan Qian for providing the photometric optimization pipeline VHAP. --- If you find this dataset useful, please consider citing:
@incollection{buehler2024cafca, title={Cafca: High-quality Novel View Synthesis of Expressive Faces from Casual Few-shot Captures}, author={Marcel C. Buehler and Gengyan Li and Erroll Wood and Leonhard Helminger and Xu Chen and Tanmay Shah and Daoye Wang and Stephan Garbin and Sergio Orts-Escolano and Otmar Hilliges and Dmitry Lagun and Jérémy Riviere and Paulo Gotardo and Thabo Beeler and Abhimitra Meka and Kripasindhu Sarkar}, year={2024}, booktitle={ACM SIGGRAPH Asia 2024 Conference Paper}, doi={10.1145/3680528.3687580}, url={https://doi.org/10.1145/3680528} }