Jianzhu Guo comments

Results 126 comments of


                                            Jianzhu Guo

how to remove the source image in the middle from the concat?

@bgomiss A temporary solution is to remove the `source_image_resized` variable below: https://github.com/KwaiVGI/LivePortrait/blob/7f755d93b6e120d5ab6bdc2f08d814b5b6935ad7/src/utils/video.py#L99

An interesting result in v2v mode

Thanks for your feedback @iloveOREO We update the main branch to fix a small bug, and the last frame no longer has pursed lips. We will update more details about...

ISSUE: Latest version Face Mask show BOX EDGES (no blur)

Thanks for your feedback, @AlonDan . We will carefully consider these demands, but it may take a considerable amount of time to evaluate and implement them.

Use JoyVASA to support audio-driven video generation

> s9.jpeg-s9.jpeg-865042464-1-192.wav.pkl-org-audio-ff.mp4 > New version of JoyVASA, better lip sync @warmshao The nose is a little strange 😂

No CUDA GPUs are

It's better for LivePortrait to run on Nvidia GPU.

Warning: data is not aligned! This can lead to a speed loss

It's a known issue, we will dig into it lately.

(fixed) Issue on animal inference with CUDA 12.8

Great thanks. I will update `ms_deform_attn_cuda.cu`. @rgaricano It seems that `weights_only` is set to `False` by default (e.g., torch 2.1.1)

(fixed) Issue on animal inference with CUDA 12.8

Fixed: https://github.com/KwaiVGI/LivePortrait/pull/484

(fixed) Issue on animal inference with CUDA 12.8

> (CUDA 12.8 - torch 2.6.0 & python 3.12.3 with updated requirements libs) (ubuntu 24.4 Ryzen 3-16Gb RTX3060-12Gb) Got.

(fixed) Issue on animal inference with CUDA 12.8

> > Great thanks. I will update `ms_deform_attn_cuda.cu`. [@rgaricano](https://github.com/rgaricano) It seems that `weights_only` is set to `False` by default (e.g., torch 2.1.1) > > On torch 2.6 they change to...