Hello. I’m using the IDM-VTON Virtual Try-On with a consistent human and dress image. Even with using a green screen with the dress image, and clearly marking masking the area in the human image where the dress should appear, the only result is an unexplained error. This even occurs with default settings and using examples. Is it offline or am I using it wrong?
It seems that the CUDA version of Zero GPU has changed and is no longer compatible with older versions of PyTorch. The following fix worked.
#requirements.txt
#torch==2.0.1
torch==2.4.0
torchvision>=0.15.2
torchaudio>=2.0.2
huggingface_hub==0.25.2
pydantic==2.10.6
Okay, great, I’ll try it again after work. Thanks.
1 Like
I’ve tried it again and it still encounters errors, so I’m guessing that it’s been superseded by one for which you provided the link with the fix. Still, when I click on that link, it generates an error too, specifically " runtime error ## Memory limit exceeded (16Gi)".
1 Like
Um, the space is quite heavy, so it won’t work unless you copy it to the Zero GPU space.
You mean there are too many users using it at the same time? Also, where do I find the Zero GPU space if not via the link you provided? By “space” do you mean webpage?
1 Like
Also, copy what to there exactly?
1 Like
It’s like this. I’m not a staff member or anything special, so I can fix the code myself, but I can’t provide paid space for free…