🗺️Roadmap: The future of HivisionIDPhotos

[Zeyi-Lin]

🔥Thanks for the open-source community's attention to HivisionIDPhotos, this is the best motivation for us.

🚀Here are the possible future directions of HivisionIDPhotos:

1. Change Suit: A set of suit can greatly improve the quality of ID photos, but people may not have suitable formal wear at home - but don't worry, HivisionIDPhotos' mission is to allow you to create exquisite ID photos anytime and anywhere. The solution for changing clothes considers using traditional CV methods and AI automatic clothing changes.

2. Facial fill light: When taking ID photos with a mobile phone camera, there is often a problem of "yin-yang face" - half of the face has shadows while the other half does not, and in photo studios, fill lights correct this. But through AI technology, this can be intelligently filled in.

3. AIGC: Through technologies such as Stable Diffusion and DreamBooth, it is possible to achieve generative ID photos. The goal is to achieve effects close to professional photo studios like Haimati Studio.

🔥非常感谢开源社区对HivisionIDPhotos的关注，这是对我们最好的激励

🚀以下是HivisionIDPhotos未来可能的迭代方向：

1. 换正装：一套正装可以让证件照品质得到跃升，但人们的家中也许没有合适的正装——但不用担心，HivisionIDPhotos的使命就是让你随时随地制作出精美的证件照。换装的方案考虑使用传统CV方法与AI自动换装。

2. 面部补光：通过手机摄像头拍摄证件照，时常会出现“阴阳脸”问题——面部一半有阴影一半没有，而在照相馆中有补光灯修正了这一点。但通过AI技术，可以智能的补上这一点。

3. AIGC：通过如Stable Diffusion、DreamBooth这样的技术，可以实现生成式证件照，目标是达到接近如海马体照相馆这样专业照相馆的效果。

[1nFrastr]

赞。第3点的AIGC功能其实就是AI写真应用。比如妙鸭相机app，也有证件照功能，是基于AI换脸实现的。AIGC换脸生成比较消耗GPU算力资源，私有部署成本高。如果考虑第三方API的话可以参考这个项目 https://github.com/loxi-opensource/luna-swapping