*******************************

This workflow can assist you in identifying and replacing clothing (applicable to other objects as well) in videos, utilizing the SAM and groundingdino models for recognition and generating corresponding masks.

Additionally, it incorporates OpenPose and Animatediff for creating motion videos.

To begin, all you need is a video with a simple background, minimal movement, and no messy hair, along with a reference image such as a T-shirt.

The workflow draws inspiration from communities like Civitai，Banodoco channel of Discord and so on.

*******************************

这个工作流可以帮助你对视频中人物衣服（同时也可以应用于其他物体）的识别和替换，其使用SAM和groundingdino模型作为识别，并且生成对应的蒙版。

另外，也引入openpose和Animatediff作为动作视频的生成。

你只需准备好一条（没有复杂背景，动作幅度不大，没有凌乱头发）的视频，以及一张参考 的图片，比如T恤。

工作流灵感借鉴于civitai、banodoco等社区。

微信：pymo_AI

抖音/B站/小红书/视频号等平台：AI大师工作流