Instructions on the project

[xizaoqu]

Hi, thanks for sharing the great work. Could you provide more instructions on the project? (i.e. how to do inference, how to download checkpoints)

[G-U-N]

Thanks for the interest! I will write instructions asap.

[xizaoqu]

Thanks for the interest! I will write instructions asap.

Hi, could you help me to check the weird results?

I suppose the weights are loaded successfully?

[TomSuen]

Hi, could you tell me how to run the gradio？How much memory will it use? @xizaoqu

[TomSuen]

Hi, could you help me to check the weird results?

I have this error

how to fix it?

[xizaoqu]

how to fix it?

Hi, try setting the CUDA_HOME to your cuda dir. I do not produce the right results so I'm not sure if I do it right. I just run app.py.

[TomSuen]

Hi, try setting the CUDA_HOME to your cuda dir. I do not produce the right results so I'm not sure if I do it right. I just run app.py.

Thank you, for your drag result, I think you can add a brush mask on the face. It may work well.

[xizaoqu]

Thank you, for your drag result, I think you can add a brush mask on the face. It may work well.

In my case, adding mask can't fix the problem.

[TomSuen]

In my case, adding mask can't fix the problem.

Maybe you should remove the prompt and adjust the Flow unit, it seems to affect range of motion.

[Fancy93]

In my case, adding mask can't fix the problem.

Maybe you should remove the prompt and adjust the Flow unit, it seems to affect range of motion.

Hi, I would like to ask if you have encountered the following error?

[TomSuen]

Hi, I would like to ask if you have encountered the following error?

Sorry, i haven't meet with this error yet. It seems that something incompatible between CUDA and pytorch, if you use linux, you can run which nvcc to check the CUDA path, and run nvcc --version to check current CUDA version.

I use docker, and my nvcc --version is above 11.8, pytorch==1.13.1+cu117