Now you can use your low resolution photos. Thanks to the Google tool, pixelated images can be converted into good quality photographs
In recent years, several tools have been launched to improve photographs. Nevertheless Google launched a new feature that allows you to convert pixelated photos on HD images.
This will be possible thanks to the Artificial intelligence. In this way, people can recover their old photos that do not have the quality that may currently exist.
Through a blog, Google developed two models to generate high-quality photographs: super-resolution imaging (SR3) and cascade diffusion (CDM).
“Super resolution has many applications that can range from restoring old family portraits to improving medical imaging systems,” said Chitwan Saharia, software engineer at Google.
How do Google’s AI systems work?
Google He explained that his SR3 or Repeated Refinement model allows you to remove digital noise from a photograph. This is what causes the images to look pixelated.
Chitwan Saharia explained that his tool first adds digital noise until it is reduced to pixels, at this time the artificial intelligence It works to scale it and remove all noise from the photo until it becomes fully visible.
“SR3 is a super-resolution diffusion model that takes a low-resolution image as its input. And create a corresponding high-resolution image from pure noise »
The result is a high quality photography. Google revealed that so far they have had good results and even people have opted for his tool over photographs taken with a camera.
Google is also working on another AI tool that allows images from 32 X 32 pixels to 1,024 X 1,024.
CDM is “a class conditional diffusion model trained with ImageNet data to generate high resolution natural images”
“With SR3 and CDM, we have brought the performance of broadcast models to the state of the art. In the class conditional and super-resolution ImageNet build benchmarks »
So far it is unknown how Google could implement these improvements, since it is not known if they will be within your search engine or will be part of the Google Photos tools.