/Coiffeur

[ICCV 2023] HairCLIPv2: Unifying Hair Editing via Proxy Feature Blending

Primary LanguageJupyter Notebook

Coiffeur - Interactive UI fork of HairClip v2

HairCLIPv2 HairCLIPv2 supports hairstyle and color editing individually or jointly with unprecedented user interaction mode support, including text, mask, sketch, reference image, etc.

Getting Started

Prerequisites

$ pip install torch==1.9.0+cu111 torchvision==0.10.0+cu111 -f https://download.pytorch.org/whl/torch_stable.html
$ pip install ftfy regex tqdm matplotlib jupyter ipykernel opencv-python scikit-image kornia==0.6.7 face-alignment==1.3.5 dlib==19.22.1
$ pip install git+https://github.com/openai/CLIP.git

Pretrained Models

Download and put all the downloaded pretrained weights into the pretrained_models directory.

Path Description
FFHQ StyleGAN StyleGAN model pretrained on FFHQ with 1024x1024 output resolution.
Face Parse Model Pretrained face parse model taken from Barbershop.
Face Landmark Model Used to align unprocessed images.
Bald Proxy Bald proxy weights from HairMapper.
Sketch Proxy Sketch proxy weights trained on hair-sketch dataset using E2style.