ComfyUI workflow for realistic hair color and complex dyeing try-on

We need a specialist in ComfyUI / AI image generation to develop a quality workflow or pipeline for realistic hair color change on user selfies.

The goal of the product is to allow women and men to upload their photo and try on different hair colors, complex coloring techniques, and reference works from stylists. The main focus is not just a simple hair color change, but a quality try-on of complex colorings: balayage, ombre, highlights, money piece, shadow roots, color melting, and other similar techniques.

We need the system to be able to:

1. Accept user selfies.

2. Correctly identify hair, its shape, texture, length, lighting, and boundaries.

3. Change hair color realistically, preserving volume, shadows, shine, structure, and the naturalness of the photo.

4. Work not only with simple colors but also with complex coloring techniques.

5. Use reference images, such as a stylist's work from Instagram, and transfer a similar coloring style to the user's hair.

6. Ensure stable results, not random generation through a simple prompt.

7. Prepare the workflow so that it can be further used via API / endpoints, preferably on ComfyUI.

We already have a basic color try-on, but it is not sufficient for quality try-on of complex colorings. Simple prompts do not solve this task, so a more complex architecture is needed: segmentation / masking, ControlNet or other control mechanisms, reference-based generation, preservation of identity, hair texture preservation, color transfer logic, possibly custom nodes or several processing stages.

Expected results:

• A working ComfyUI workflow or pipeline;

• The ability to test different colors and complex colorings;

• Support for reference images for transferring coloring style;

• Realistic results without changing the face, background, and overall identity of the person;

• Documentation on how to run the workflow and how to integrate it via API.

It would be a plus if you already have experience with:

• ComfyUI;

• Stable Diffusion / Flux / SDXL;

• Hair segmentation;

• Inpainting;

• ControlNet / IP-Adapter / reference-based generation;

• Image-to-image pipelines;

• AI beauty / hairstyle / fashion try-on products;

• Preparing workflows for production API.

We want to start specifically with hair color try-on, especially balayage / ombre / highlights, but in the future, we plan to expand the logic to also include try-ons for hairstyles and cuts.