JiauZhang/DragGAN
Implementation of DragGAN for interactive point-based manipulation of images using StyleGAN2 generative models.

Velocity · 7d
+1.9
★ / day
Trend
→steady
star history
This repository implements DragGAN, enabling users to drag points on an image to deform it while staying on the generative manifold learned by StyleGAN2. The system optimizes latent codes to move handle points toward target positions while enforcing mask constraints. It includes a GUI built with DearPyGUI for interactive manipulation and supports pre-trained FFHQ models.