AniPortrait

Why we included this project

AniPortrait turns one photo of a face plus an audio clip into a talking-head video where the lip movements line up with the speech. You can also drive the animation from a video instead, copying the expressions and head motion of the source footage onto your reference image, which is useful when you already have recordings and want to reuse them rather than generate every gesture from scratch. The Gradio demo and documented inference commands make it easy to judge output quality before wiring it into a larger pipeline, whether your project is a virtual presenter, a game character, or localized voiceover content. It is research code from Tencent rather than a finished product: expect to download the pretrained weights yourself and keep an eye on GPU memory, but the workflow stays approachable for anyone comfortable with Python and diffusion models.