Researchers in China have created a new AI system that assures the restoration of old and damaged images to greater quality. The photographs may be smeared, damaged, or of low resolution. The open source program can recover these photographs in a relatively short period of time.
GFP-GAN is the name given to it (which in translation is something like Pre-Generative Facial Generative Adversarial Network). One of its strengths is that it can do restoration without affecting photo components, which results in better results than other apps with similar capabilities.
GFP-GAN employs an NVIDIA AI model called StyleGAN-2, which is primarily responsible for eliminating scratches and noise from photos, as well as repairing markings and creases that emerge when a photo has been kept and folded for an extended period of time. It may also improve the color level and highlight dark elements in the image using a guessing approach.
The purpose is merely to clean up the image, and it can even fill certain gaps, but it does not alter the image. This is a benefit when compared to other tools of this type.
She works with both old and low-quality photographs that have been transferred to a computer and digital photographs that have been damaged in some way. However, the researchers claim that the tool still has certain limitations, and that as a result, there may be some distortions in the final outcome.
To try out the tool, go to the website by clicking here. The researchers have already informed whom a more advanced model will be released in the near future, bringing improvements and a more precise detection with a higher ndice of acertos.