Mingi Kwon

@dreamingtulpa Because of this, we’re following our advisor’s guidance and working on implementing safety rather than releasing the model openly right now. We’re planning to release the checkpoint for academic purposes with appropriate safeguards. Please stay tuned a little longer!

but... but...

in case you didn't know, jam-flow matches speech to video:

@dreamingtulpa Hi! Thanks for sharing! We created a simple example to demonstrate how our model could be misused, and we confirmed that it’s alarmingly easy to make any celebrity say whatever we want in their own voice.

@proff_mahdi @2ptmvd I had the same issue before, and we claimed it to ECCV. ECCV decided to change our paper to be accepted. You SHOULD claim it.

@2ptmvd My reviewers rejected my paper only for an ArXiv paper published after the CVPR deadline!

What a solid justification. CVPR 2025.

Got accepted after appeal! 🤣 "The PCs discussed the appeal in detail and arrived at the final decision of accepting the paper." Two papers got accepted to #ECCV2024 😇

One paper is accepted to #ECCV2024 BUT another is rejected with following meta review; "The reviewers who gave accept all said there is significant overlap with A paper's approach." and A paper is also accepted to ECCV2024... WHAT SOULD WE DO?

A horse trotting along a beach at sunset. 😊

Okay. I can send them wherever.

cute

Generated with frozen StableDiffusion + temporal layers by @Mingi_Kwon_ 🥳

A teddy bear walking down 5th Avenue, beautiful sunset. 🥰