LatentSync: Taming Stable Diffusion for Lip Sync!

[@taobabay]

magnifique as usual

[@StableAIHub]

Cheers!!

[@ChikadorangFrog]

Thanks

[@StableAIHub]

Cheers!!

[@trishul1979]

Looks interesting.

[@StableAIHub]

Cheers!!

[@ChikadorangFrog]

just curious. why do we have to remove the triton?

[@StableAIHub]

This need Triton 2.0.0 and newer wheel create issue. Installation failed for me if I don't remove it.

[@TheBlackbeansoup]

Can you share what you use for post processing?

[@StableAIHub]

Please refer this
ruclips.net/video/iVy2bXPQNKY/видео.html

[@trishul1979]

Triton wheel not gonna work?

[@StableAIHub]

Nope. This need Triton 2.0.0 and newer wheel create issue.

[@namanchawla710]

how does the inference steps and guidance value affect the output?

[@StableAIHub]

If the Guidance Scale is too low, the AI might produce images that bear little resemblance to the text prompt. If it's too high, the images might lack creativity and appear too literal or constrained. Increasing inference steps improves the quality.

[@WayneHendersonsr]

Why did you delete my comment in regards to viruses?

[@StableAIHub]

I didn't delete any comment. What viruses?

[@WayneHendersonsr]

@@StableAIHub I solved it.