ConsID-Gen Dataset

Summary

This dataset is released with our paper:

ConsID-Gen: View-Consistent and Identity-Preserving Image-to-Video Generation
Mingyang Wu, Ashirbad Mishra, Soumik Dey, Shuo Xing, Naveen Ravipati, Hansi Wu, Binbin Li, Zhengzhong Tu (2026)

This release contains processed videos and hierarchical captions used in ConsID-Gen experiments across:

• public_synthesis (MVImgNet)
• public (CO3D, OmniObject3D)

The dataset is designed for image-to-video generation research with an emphasis on identity preservation and view consistency.

Paper and Project

• Paper (arXiv): https://arxiv.org/abs/2602.10113
• Project page: https://myangwu.github.io/ConsID-Gen/

Directory

DATA_PATH/
`-- datasets/
    |-- public_synthesis/
    |   |-- processed_videos/
    |   `-- hierarchy_video_caption/
    `-- public/
        |-- processed_videos/
        `-- hierarchy_video_caption/

Current Release Notes

• Current snapshot includes the folders listed above.
• More data is currently under internal review and will be released after approval.

Download

Use the Hugging Face CLI:

hf download mingyang-wu/ConsID-Gen \
  --repo-type dataset \
  --local-dir DATA_PATH

Or download specific subsets:

hf download mingyang-wu/ConsID-Gen \
  --repo-type dataset \
  --include "datasets/public/**" \
  --local-dir DATA_PATH

Usage

Captions are stored as .txt files in hierarchy_video_caption/{source}/{s1,s2}.
Videos are stored as .mp4 files in processed_videos/{source}.

Example:

from pathlib import Path

root = Path("DATA_PATH/datasets")
video = root / "public/processed_videos/co3d"
caption_s1 = root / "public/hierarchy_video_caption/co3d/s1"
caption_s2 = root / "public/hierarchy_video_caption/co3d/s2"

print("videos:", len(list(video.glob("*.mp4"))))
print("captions_s1:", len(list(caption_s1.glob("*.txt"))))
print("captions_s2:", len(list(caption_s2.glob("*.txt"))))

Citation

@misc{wu2026considgenviewconsistentidentitypreservingimagetovideo,
  title={ConsID-Gen: View-Consistent and Identity-Preserving Image-to-Video Generation},
  author={Mingyang Wu and Ashirbad Mishra and Soumik Dey and Shuo Xing and Naveen Ravipati and Hansi Wu and Binbin Li and Zhengzhong Tu},
  year={2026},
  eprint={2602.10113},
  archivePrefix={arXiv},
  primaryClass={cs.CV},
  url={https://arxiv.org/abs/2602.10113},
}