Google researchers in the company’s Brain Team highlight new accomplishments in image super-resolution in an article titled “High Fidelity Image Generation Using Diffusion Models” published on the Google AI Blog (and spotted by DPR).
A machine learning model is taught to convert a low-resolution photo into a detailed high-resolution photo in image super-resolution, and potential applications range from recovering old family images to improving medical imaging.
The approach is called SR3, or Super-Resolution via Repeated Refinement.
“SR3 is a super-resolution diffusion model that takes as input a low-resolution image, and builds a corresponding high resolution image from pure noise. The model is trained on an image corruption process in which noise is progressively added to a high-resolution image until only pure noise remains.”
![](https://www.nucleiotechnologies.com/wp-content/uploads/2021/09/image-36-1024x587.png)
“It then learns to reverse this process, beginning from pure noise and progressively removing noise to reach a target distribution through the guidance of the input low-resolution image.” Google added.
![](https://www.nucleiotechnologies.com/wp-content/uploads/2021/09/image-37-1024x582.png)
![](https://1.bp.blogspot.com/-QM0e0S3uxAo/YS9CcfEuIzI/AAAAAAAAcLU/m6efgEA6Cykm_RgBwEKvYxzpph0xiW3MQCLcBGAsYHQ/s1296/g-ai.png)
The new technology is incredible, and Google is proud of its team for doing such a fantastic job. We can’t wait to see what else they can do in the world of diffusion models.