Text-To-Image with Outfit Transfer Reference Image

Like with anything related to Krea Edit this is very much experimental but it works surprisingly well so I decided to share it with the community.

Notes:

• Requires https://github.com/lbouaraba/comfyui-krea2edit or https://github.com/ostris/ComfyUI-Krea2-Ostris-Edit (See notes in Workflow)

• Requires input outfit reference images in specific format. See example dataset: https://huggingface.co/datasets/AliveAi/outfits

• Use trigger "transfer the outfit"

Workflow:

• Two workflow options are attached. Let me know which one you prefer!

  • t2i_outfit_reference.json requires https://github.com/lbouaraba/comfyui-krea2edit - better accuracy but much slower

  • Krea2_Ostris_Edit_outfit.json requires https://github.com/ostris/ComfyUI-Krea2-Ostris-Edit - less perfect outfit reference adherence but much faster

Known limitations / issues:

• Only trained on female outfits (probably also works for men)

• Can sometimes produce images with two people. Re-try with different seed and/or update prompt to force "single person"