Hugging Face Space Demo
aandyw opened this issue · 12 comments
Hi there, thank you for creating such a cool project/paper.
I have implemented your work as a Hugging Face Space Demo.
https://huggingface.co/spaces/Pie31415/rome
I have not had the chance to implement the video inference using ROME but will do so asap.
This is great @pie31415! Any chance to add way for users to download the generated 3d mesh?
That's a good idea I didn't think about that. I'll look on adding that as well. Do you have suggestions for how to retrieve the mesh?
@pie31415 I am not very familiar with the code but if you can access the vertices (position and texels) shouldn't be hard to generate an .obj
@Ha0Tang I'll try to get both the 3d mesh downloading and video inference done by the end of this week.
@pie31415 Please let me know if you need some assistant in getting the 3d mesh I have a particular interest on that.
@yuripourre For sure. Feel free to open a PR to the space demo if you have any suggestions.
I've gotten the video inference to work now. @Ha0Tang
You can try the video inference on the demo space.
@Ha0Tang I separated the inference into video and webcam inference tabs. You should be able to upload driver videos now to the 'Video Inference.'
Is there a way of downloading mp4 files instead of GIFs? Bit difficult to assess the quality of the output
@anwoflow For sure, GIF was mainly just so I could get something up and running quickly. I'll try to get a download video format ready as soon as possible.
@yuripourre Also, I've been having some issues with loading up the 3d downloadable mesh. If it's something you're still interested in I would love some help on that since I'm not too familiar with the graphics side of things.