Google's New Approach to Image Super-Resolution

The Google research team presented SR3, an approach to image Super-Resolution that is based on Repeated Refinement. SR3 uses denoising diffusion probabilistic models to conditional image generation and performs super-resolution with a stochastic denoising process.

The team noted that "inference starts with pure Gaussian noise and iteratively refines the noisy output using a U-Net model trained on denoising at various noise levels. SR3 exhibits strong performance on super-resolution tasks at different magnification factors, on faces and natural images."

The team shared a paper that showcases the performance of SR3 on the tasks of the face and natural image super-resolution. You can find more details and results here. 

Don't forget to join our new Reddit page, our new Telegram channel, our Discord, follow us on Instagram and Twitter, where we are sharing breakdowns, the latest news, awesome artworks, and more.