利用 AI 模型重建静态图片,使其同步目标视频中的表情 (表情迁移)
利用 AI 模型重建静态图片,使其同步目标视频中的表情(表情迁移)。
用人话说:
上传一个人像图片,再上传一个你想复制表情的参考短视频(几秒钟),点击“重建表情”按钮,就会把静态图片转换成视频。表情与你上传的参考视频一致,音频也是复制参考视频的。
软件信息
- 开源软件主页: https://github.com/KwaiVGI/LivePortrait
- 硬件要求: 需要使用 NVIDIA 显卡(老黄家),显存占用约 3GB。
- 性能参考: 4060 显卡执行转换 7 秒钟视频,大约消耗 35 秒。
版本优化说明
- 已重新打包,删除了官方示例视频与图片。
- 精简了代码,并完成了汉化。
- 版本拆分: 表情迁移和表情重塑已拆分为两个程序(方便追求简洁界面的用户)。
- 所有功能均已通过测试。
使用说明
1. 表情迁移使用说明
- 上传图片
- 上传参考视频
- 点击 开始同步表情
- 注意: 所有设置全部默认即为最佳设置。目标图片和参考视频必须使用正面、正脸,否则效果不佳。
2. 面部表情重建使用说明
- 上传图片
- 调整参数(例如向左偏头、眼睛调整大些、张嘴等等)
- 执行重建
- 几秒钟即可看见效果。
注意事项(避坑指南)
- 路径问题: 软件不要放在有空格的文件夹,也不要放在中文文件夹。
- 文件名问题: 载入的图片、视频文件名不能有中文,不能有空格。
- 如果报错,请优先检查上述路径与命名问题。
下载链接
https://pan.baidu.com/s/19orN6bPtPkRU4AHHMEzKYg?pwd=gpcb
效果展示

(注:该图为 AI 生成)

播放器载入中...
播放器载入中...
完整界面:
底下的面部重建也非常有趣:
{!{data:image/webp;base64,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}!}