fenghe12
fenghe12
感觉这种FF++视频中获取图片的都很难复现
me too
maybe you can refer to their model code，at least we have that，training code seems just like animate anyone
yes,i also think reader /writer is used to implement The image of the target character is inputted into the ReferenceNet to extract the reference feature maps outputs from the self-attention layers....
ok ，i'll take a look. we can use pretrained model,in fact Alibaba also uses pretrained model (from hugging face Stable Diffusion v1.5) reference-net and backbone inherit weights from the original SD...
yes,maybe just a combination of existing methods and modules. I'll take a look later. ------------------ 原始邮件 ------------------ 发件人: "johndpope/Emote-hack" ***@***.***>; 发送时间: 2024年3月27日(星期三) 上午10:56 ***@***.***>; ***@***.******@***.***>; 主题: Re: [johndpope/Emote-hack] take a look at this (Issue...
Have you run through the entire process? Congratulations! Let me take a look at the repo and code！ ------------------ 原始邮件 ------------------ 发件人: "johndpope/Emote-hack" ***@***.***>; 发送时间: 2024年3月29日(星期五) 中午1:36 ***@***.***>; ***@***.******@***.***>; 主题: Re: [johndpope/Emote-hack] take a...
VASA adopts DiT as backbone denoising network，but it lacks more details about how to integrate conditons into dit. I attempted to replace the unet in Moore animateanyone with DIT (Latte,...
I guess DiT will be mainstream video generation architecture，because of SORA
Maybe I can offer some help > hey @fenghe12 > > i had some success with megaportraits https://github.com/johndpope/MegaPortrait-hack > > and now attempting to integrate into VASA on this branch....