We’re sharing an open-source project on makeup transfer built on a Diffusion Transformer (DiT) backbone.
The goal is to transfer makeup from a reference face to a source face while preserving identity and background consistency. Unlike many previous approaches, this method does not rely on facial landmarks, 3D face models, or other face-control modules — it uses only a source image and a reference image as input.
Hey, pretty nice work!
Are you using any CLIP-like model for image retrieval? If so, would you try FG-CLIP 2 (https://360cvgroup.github.io/FG-CLIP) and see how it'll improve the search results.
We just open-sourced this model, which excels in fine-grained image-text understanding, and I met your post right before I post our work on HN.
The goal is to transfer makeup from a reference face to a source face while preserving identity and background consistency. Unlike many previous approaches, this method does not rely on facial landmarks, 3D face models, or other face-control modules — it uses only a source image and a reference image as input.
Paper (arXiv): https://arxiv.org/abs/2508.05069
Weights + comfyUI: https://github.com/360CVGroup/FLUX-Makeup
You can also give it a quick try at FLUX-Makeup agent: https://www.n.cn/tools/aiagent/chat/c8138adb99d04bac847c5574..., it's free to use, you might need web translation because the UI is in Chinese.
Interested in feedback from people working on diffusion-based editing and controllable image generation.