Awesome
<img src="./draggan.png" width="750" alt="Architecture of DragGAN"/>DragGAN
Implementation of DragGAN: Interactive Point-based Manipulation on the Generative Image Manifold.
# gui
pip install dearpygui
# run demo
python gui.py
<img src="./sample/start.png" width="200" alt="Demo Start"/><img src="./sample/end.png" width="200" alt="Demo End"/><img src="./sample/multi-point.png" width="200" alt="Multi Point Control"/>
TODO
- GUI
- drag it
- load real image
- mask
StyleGAN2 Pre-Trained Model
Rosinality's pre-trained model(256px) on FFHQ 550k iterations [Link].