Skip to main content


Photo restoration can be a long and tedious process, especially when the degradation artifacts are complex and the poses and expressions are diverse. Fortunately, the researchers from ARC Tencent developed GFP-GAN – a brand-new technique for real-world blind face restoration that makes use of a pre-trained GAN and spatial feature transform to restore facial features in only one forward pass.

This new method is significantly faster than existing methods, and it produces results that are more consistent with human perception. In addition, GFP-GAN is capable of handling a wide variety of degradation artifacts, making it an essential tool for anyone who wants to restore old photos.

GFPGAN is a great tool for anyone who wants to develop a practical algorithm for real-world face restoration. It uses a pre-trained face GAN (StyleGAN2) to help you encode rich and diverse priors in your algorithm. This makes it easy to create a blind face restoration that looks natural and realistic.

By continuing to use the site, you agree to the use of cookies.