FLUX-Makeup: One-Click Makeup Transfer for Creative Face Painting

360 AI Research
2026-10-08 3367 views
FLUX-Makeup: One-Click Makeup Transfer for Creative Face Painting

As digital identity becomes increasingly important, makeup is no longer confined to the dressing table. It is becoming a form of visual expression that can be generated instantly and changed freely.

FLUX-Makeup is a makeup-transfer model and product built for this purpose. There is no need to apply physical makeup or perform professional retouching: provide a source face photo and a reference makeup image to transfer the look in one click with high quality.

Technical foundation: a custom architecture without auxiliary facial controls

FLUX-Makeup is built on the Diffusion Transformer (DiT) architecture and is the first model in the field to apply DiT to makeup transfer.

It takes a source image and a reference image directly as its native conditioning inputs. It does not rely on auxiliary facial-control modules such as landmark detection or 3D mesh alignment. This avoids the additional errors that auxiliary cues can introduce in conventional methods, makes transfer more stable and direct, and substantially reduces model complexity.

Core capabilities

Preserve the original facial features

While transferring makeup, FLUX-Makeup preserves the source person's facial features, expression, and appearance. You can change the makeup without changing the face. Whether trying a subtle everyday look or creative face painting, the facial structure remains intact.

Natural color transitions

Our lightweight RefLoRAInjector module efficiently extracts the colors, textures, and layers of the reference makeup and injects those features precisely into the target regions. Highlights, shadows, blush, and lip colors blend smoothly, producing a natural finish comparable to the work of a professional makeup artist.

Support a wide range of makeup styles

FLUX-Makeup handles everyday makeup, professional office looks, traditional Chinese forehead ornaments, festive face painting, theatrical makeup, and creative special-effects looks.

Trained on diverse, high-quality makeup pairs, the model generalizes across challenging lighting, subtle or dramatic styles, and people with different ethnic backgrounds and skin tones.

Technical highlights

A pioneering use of DiT. Moving beyond the limitations of conventional GANs and U-Net diffusion models, FLUX-Makeup introduces a new approach to makeup transfer.

Lightweight injection with RefLoRAInjector. The reference-makeup pathway is decoupled from the model backbone, substantially accelerating transfer while precisely directing makeup-feature injection.

A high-quality paired-data pipeline. Our robust, scalable data-generation process produces diverse, precisely aligned makeup pairs, providing training supervision of substantially higher quality than existing datasets.

No auxiliary control modules. A simpler training and inference pipeline reduces accumulated errors and makes transfer more stable and controllable.

Try it now

FLUX-Makeup is available online:

Try FLUX-Makeup

Open the mobile web experience, or scan the QR code below.

FLUX-Makeup mobile experience QR code

Whether you want to preview a new look, create festive face painting, or produce makeup images in batches for professional work, FLUX-Makeup helps bring the look you imagine to a face in seconds.

FLUX-Makeup illustration from the original article