Yuhao Xu comments

Results 63 comments of


                                            Yuhao Xu

Model garment affect the end result

Hi, Thanks for asking. It is a common failure case for cross-category VTON due to the imperfect mask. We plan to solve this in our OOTDiffusion V2.

When selecting the dc model, it works fine, but when selecting the hd model, it reports an error.

Please download the model checkpoints from https://huggingface.co/levihsu/OOTDiffusion/tree/main

After completion, where is the last updated file saved?

Final output is saved in `./run/images_output` and no such intermediate result should be saved Please check the correctness of your code and files

INVALID_PROTOBUF

Please download the checkpoints from https://huggingface.co/levihsu/OOTDiffusion/tree/main/checkpoints

Is it possible to display an intermediate image of the training process, or to restart training again from the intermediate process?

Yep. Just save checkpoints and set resume_from_checkpoint in your training script to restart training.

How to run code on our own machine

Please refer to [README](https://github.com/levihsu/OOTDiffusion/blob/main/README.md) for installation and inference

Gradio下是不是有点小问题

请检查你的环境配置我们自己的gradio demo都是几秒出图 https://huggingface.co/spaces/levihsu/OOTDiffusion https://ootd.ibot.cn/

Discrepancy between Local Gradio and Online Demo Results

Hi all, please check if your downloaded checkpoints (including several different ***config.json***) are correct. Our online demo uses the completely same checkpoints. See the files at https://huggingface.co/spaces/levihsu/OOTDiffusion

torch.cuda out of memory

Hi. It takes around 3 seconds and 6GB memory for 1 sample and 20 steps on our RTX 4090 GPU. Maybe a 6GB 3060 is not completely enough.