Yuhao Xu
Yuhao Xu
I see. It seems to be an environmental error and the onnx file is not successfully loaded. Could you check your environment and files again?
https://huggingface.co/levihsu/OOTDiffusion/tree/main/checkpoints
Our training script will be released later. Before that you could refer to some unofficial implementations like https://github.com/lyc0929/OOTDiffusion-train
Please check if your checkpoints files are correctly downloaded, including several different ***config.json***
仅安装Pytorch是远远不够的 请参考readme配置环境
> `OSError: We couldn't connect to 'https://huggingface.co' to load this model, couldn't find it in the cached files and it looks like ../checkpoints/ootd is not the path to a directory...
Hi. Do you mean your own trained model or our official checkpoints? In our method the unmasked region won't be distorted. Please check the resolution of your input images, etc.
Any input will be automatically rescaled to 1024*768. So if the original image is too small, some features may be distorted.
这个功能限于数据和方法暂时不行哈 可能需要加入更强的语义控制
11.7 但是这个貌似不会太影响显存占用？