Saved in:
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2506.10639 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866908405799583744 |
|---|---|
| author | Bao, Xiaoyi Lv, Jindi Wang, Xiaofeng Zhu, Zheng Chen, Xinze Zhou, YuKun Lv, Jiancheng Wang, Xingang Huang, Guan |
| author_facet | Bao, Xiaoyi Lv, Jindi Wang, Xiaofeng Zhu, Zheng Chen, Xinze Zhou, YuKun Lv, Jiancheng Wang, Xingang Huang, Guan |
| contents | Recent progress in diffusion models has greatly enhanced video generation quality, yet these models still require fine-tuning to improve specific dimensions like instance preservation, motion rationality, composition, and physical plausibility. Existing fine-tuning approaches often rely on human annotations and large-scale computational resources, limiting their practicality. In this work, we propose GigaVideo-1, an efficient fine-tuning framework that advances video generation without additional human supervision. Rather than injecting large volumes of high-quality data from external sources, GigaVideo-1 unlocks the latent potential of pre-trained video diffusion models through automatic feedback. Specifically, we focus on two key aspects of the fine-tuning process: data and optimization. To improve fine-tuning data, we design a prompt-driven data engine that constructs diverse, weakness-oriented training samples. On the optimization side, we introduce a reward-guided training strategy, which adaptively weights samples using feedback from pre-trained vision-language models with a realism constraint. We evaluate GigaVideo-1 on the VBench-2.0 benchmark using Wan2.1 as the baseline across 17 evaluation dimensions. Experiments show that GigaVideo-1 consistently improves performance on almost all the dimensions with an average gain of about 4% using only 4 GPU-hours. Requiring no manual annotations and minimal real data, GigaVideo-1 demonstrates both effectiveness and efficiency. Code, model, and data will be publicly available. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2506_10639 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | GigaVideo-1: Advancing Video Generation via Automatic Feedback with 4 GPU-Hours Fine-Tuning Bao, Xiaoyi Lv, Jindi Wang, Xiaofeng Zhu, Zheng Chen, Xinze Zhou, YuKun Lv, Jiancheng Wang, Xingang Huang, Guan Computer Vision and Pattern Recognition Recent progress in diffusion models has greatly enhanced video generation quality, yet these models still require fine-tuning to improve specific dimensions like instance preservation, motion rationality, composition, and physical plausibility. Existing fine-tuning approaches often rely on human annotations and large-scale computational resources, limiting their practicality. In this work, we propose GigaVideo-1, an efficient fine-tuning framework that advances video generation without additional human supervision. Rather than injecting large volumes of high-quality data from external sources, GigaVideo-1 unlocks the latent potential of pre-trained video diffusion models through automatic feedback. Specifically, we focus on two key aspects of the fine-tuning process: data and optimization. To improve fine-tuning data, we design a prompt-driven data engine that constructs diverse, weakness-oriented training samples. On the optimization side, we introduce a reward-guided training strategy, which adaptively weights samples using feedback from pre-trained vision-language models with a realism constraint. We evaluate GigaVideo-1 on the VBench-2.0 benchmark using Wan2.1 as the baseline across 17 evaluation dimensions. Experiments show that GigaVideo-1 consistently improves performance on almost all the dimensions with an average gain of about 4% using only 4 GPU-hours. Requiring no manual annotations and minimal real data, GigaVideo-1 demonstrates both effectiveness and efficiency. Code, model, and data will be publicly available. |
| title | GigaVideo-1: Advancing Video Generation via Automatic Feedback with 4 GPU-Hours Fine-Tuning |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2506.10639 |