Thread by @_Epoching_

I made my first @huggingface Spaces with @Gradio: 3D Photo Inpainting! Converts your photos into short trajectory-following videos. May be old news to some folks, but I didn't see it on Spaces yet!

Check it out here: huggingface.co/spaces/Epochin…🙂

Inference times are long though😭

I do apologize for the long runtimes, definitely feels like a computationally-heavy task. The queue time is estimated around 400-700 seconds.

(It's CPU-only, but will run with a GPU if there is one, feel free to clone the Space and run the gradio app locally!)

Thanks a lot @ak92501 for always creating lovely examples to follow, and @mervenoyann + @osanseviero for sharing awesome resources & polished spaces🙂

Can't wait to make more👀

I do also recommend playing with the original repo, as it's argument.yaml lets you customize the behavior in much depth (pun-intended)!

Original Paper: arxiv.org/abs/2004.04727
Project Page: shihmengli.github.io/3D-Photo-Inpai…
Github Repo: github.com/vt-vl-lab/3d-p…

@huggingface @Gradio cc @juaniters1 for helping build this space!🙂

Great summary of the paper by @casualganpapers
casualganpapers.com/single_view_la…

@huggingface @Gradio Apparently my smiling emoji got attached to the link, so clicking on it doesn't work, whoops.

Here it is with nothing attached 😅

huggingface.co/spaces/Epochin…

@ak92501 @mervenoyann @osanseviero Speaking of making more, here's #2 😅

https://twitter.com/_Epoching_/status/1473661081369210887?s=20