It's already July, where are codes and dataset?

[KleinXin]

looking forward

[guahunyo]

https://github.com/KwaiVGI/LivePortrait There's a similar one here

[johndpope]

just an observeration - Megaportrait paper dropped by samsung ai - russia in 2022-07 (2207.07621 then MetaPortrait https://arxiv.org/abs/2212.08062 by microsoft in december - 2212 - used DenseMotionNetwork then VASA paper says ( 2404.10667) - says dont use metaportrait - use megaportrait (doesnt use densemotionnetwork) for faster fps - more details here https://github.com/johndpope/MegaPortrait-hack/issues/16 (livePortrait is using densemotionnetwork)

it will be good when this EmoPortrait code drops (as @neeek2303 worked on the original megaportraits code) - as we can clear up details on warping with resnet50...

but..... some researchers in china released paper in 2405 may - which claims to achieve simpler / cleaner faster disentanglement - there's no warp function - i've been coding this - and wasnt getting anywhere - then switched to gan architecture - believe after a few hours - it managed to disentangle the pose see the turned faces on the right. https://github.com/johndpope/speak-hack (once i get architecture right, will throw it at a gpu farm)

[neeek2303]

@johndpope @guahunyo @KleinXin @JZArray Hey guys. Thank you for commenting and sorry for the long wait. At the moment I’m working on a project that is very time-consuming; I literally don’t have time to sit down, put things in order in the code, prepare the data and put everything in the public domain. This work was done as part of my work at Imperial College London, not as part of my work at Meta. So I can't do this during work hours. But rest assured, I'm still going to do it!

I will try to post the code in July as promised (and the dataset too). I will also have a vacation in August, and if I don’t have time to do everything in July, then at least I’ll post it in the first half of August. Thank you for your understanding!

[JZArray]

https://github.com/KwaiVGI/LivePortrait There's a similar one here

@guahunyo thx，already tried

[JZArray]

@johndpope @guahunyo @KleinXin @JZArray Hey guys. Thank you for commenting and sorry for the long wait. At the moment I’m working on a project that is very time-consuming; I literally don’t have time to sit down, put things in order in the code, prepare the data and put everything in the public domain. This work was done as part of my work at Imperial College London, not as part of my work at Meta. So I can't do this during work hours. But rest assured, I'm still going to do it!

I will try to post the code in July as promised (and the dataset too). I will also have a vacation in August, and if I don’t have time to do everything in July, then at least I’ll post it in the first half of August. Thank you for your understanding!

thx, actually we have already successfully reimplement MegaPortraits, and problems(such as ID leakage, closing eyes, lip movement and so on) have been solved. We only want to compare our reimplementation with the original one, and to see where we can further refine our self-built model.

[johndpope]

https://x.com/purzbeats/status/1812287664240107969?s=46&t=-tkSIrsyNobBjIvQ2IAQwQ

God speed @neeek2303

[KleinXin]

https://x.com/purzbeats/status/1812287664240107969?s=46&t=-tkSIrsyNobBjIvQ2IAQwQ

Got speed @neeek2303

Amazing! Is the face on the bottom left corner yours？It seems not the one from the paper.

[johndpope]

Hi @KleinXin - no - thats not me - https://www.youtube.com/watch?v=zLSfnO0Ob-I

[FurkanGozukara]

i used my face you can see my live portrait tutorials

https://youtu.be/FPtpNrmuwXk

Animate Static Photos into Talking Videos with LivePortrait AI Compose Perfect Expressions Fast

81.) Free & Paid - Cloud - RunPod - Massed Compute - Kaggle

https://youtu.be/wG7oPp01COg

LivePortrait: No-GPU Cloud Tutorial - RunPod, MassedCompute & Free Kaggle Account - Animate Images