aigc-apps / aigc-apps/VideoX-Fun
图生视频训练问题
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 2.3k
- Forks
- 188
- Avg merge
- 2d 22h
- Merged PRs (30d)
- 3
Description
我基于pose模型，加了一些额外的图片输入条件进行了重新训练。训练结果如下：
2000轮：
https://github.com/user-attachments/assets/b54ef88a-e0d6-45e2-8f64-4ad4c5ac77c9
3000轮：
https://github.com/user-attachments/assets/db879a45-9c81-4c03-809d-eff557a58951
为什么训练时间越久，人物的嘴部刚开始反而出现了奇怪的亮斑导致嘴形崩坏？而且不仅仅这个case，几乎所有训练的结果都在视频一开始有嘴部崩坏，然后逐渐消失？
（我没有对嘴部添加任何监督或者别的什么特殊处理，除了数据集和增加参考图输入外，没有改动其他任何训练设置和代码）
Contributor guide
No contributing guide indexed for this repository
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by reproducing the pose-model training with the added reference-image input and compare the 2,000- and 3,000-step outputs, focusing on the bright mouth artifacts at the beginning of videos. Review the dataset and the changed training inputs against the unchanged settings and code; done means identifying a reproducible cause or documenting the conditions that trigger the artifact.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- computer-vision, machine-learning
- Issue type
- Bug
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100