Unofficial implementation of DragGAN with StyleGAN2/3 pretrained models

This repo is mainly to re-implement DragGAN based on stylegan2/3

• DragGAN: Drag Your GAN: Interactive Point-based Manipulation on the Generative Image Manifold

Trying out the Web Demo for dragging your own image:

• Linux or macOS
• NVIDIA GPU CUDA CuDNN
• Python 3

• Clone the repository:

git clone https://github.com/MingtaoGuo/DragGAN_stylegan3.git
cd DragGAN_stylegan3

• Dependencies:
  We recommend running this repository using Anaconda or Docker. All dependencies for defining the environment are provided in environment.yaml and Dockerfile .

Downloading the stylegan2 pretrained models:

• stylegan2-ffhq-512x512.pt
• stylegan2-afhqwild-512x512.pt
• stylegan2-afhqdog-512x512.pt
• stylegan2-afhqcat-512x512.pt
• stylegan2-human-1024x1024.pt

Drag generated image:

python draggan_stylegan2.py

Drag generated human image:

python draggan_stylegan2_human.py

Drag real image:

python draggan_stylegan2_realimg.py

In the draggan_stylegan2.py, src_points (red point in image) will be dragged to the tar_points (blue point in image), so just revise the points in src_points and tar_points.

FFHQ1	FFHQ2

Human1	Human2

AFHQ1	AFHQ2

AFHQ_Cat1	AFHQ_Cat2

AFHQ_Dog1	AFHQ_Dog2

Real image	Projected image	Drag Result

Mingtao Guo E-mail: gmt798714378 at hotmail dot com

stylegan3 stylegan-human cutout team

[1]. Pan, Xingang, et al. "Drag Your GAN: Interactive Point-based Manipulation on the Generative Image Manifold." arXiv preprint arXiv:2305.10973 (2023).