DragDiffusion
DragDiffusion copied to clipboard
[CVPR2024, Highlight] Official code for DragDiffusion
我离线下载了huggingface的模型,也训好了lora模型,然后在gradio界面想用模型拉伸图片,可是给我报“网络连接”的错误,我挺纳闷的
My GPU is 12g and still out of memory.
用了最新版本的代码,在A100服务器上从训lora到drag一张图大概要50秒的时间。如果想处理多个图片,速度显得有点慢了,有没有方法可以提速呢?
To those who are struggling to replicate this. Below is the google colab version which can run on python3.10 linux google colab !git clone https://github.com/Yujun-Shi/DragDiffusion.git %cd /content/DragDiffusion !pip install accelerate==0.17.0...
I would like to ask if the training part of lora can be continued to speed up through some means to make DragDIffusion faster?
When i click on Train Lora button. Windows 11 Nvidia 3060
cudatoolkit
ResolvePackageNotFound: - cudatoolkit=11.7 Any chance of supporting a current version?
i noticed that your training DragPipeline class will save checkpoint incompatible with StableDiffusionPipeline,StableDiffusionImg2ImgPipeline and (https://github.com/cloneofsimo/lora/blob/master/scripts/run_img2img.ipynb). Is there a way for me to make this dragpipeline checkpoint compatible with the original...
I really love on your paper webpage the 4 MP4 videos of the "Editing Trajectory" that show an animation of the intermediate steps of the sculpture head turning, for example....