Despite what the “zoom and enhance” trope on film and TV wants you to believe, no amount of digital zooming or image enhancement can create information that wasn’t captured in the original photo. But with powerful artificial intelligence, we can at least create realistic-looking images even from low-quality or pixelated source material.

Enter Stability.AI and its newly released upscaling tool that enlarges images without compromising the sharpness. Utilizing the Real-ESRGAN model, this new image upscaling tool doubles the resolution of a 512-by-512 input image in less than one second.

The tool also makes use of the Stable Diffusion x4 upscaler model, which quadruples an image resolution while preserving or even enhancing the image texture through an advanced diffusion process. It is, however, more time-consuming that takes at least 20 seconds.

StabilityAI-4x-resolution-images

The Stability.AI upscaling tool can be used to upscale any input image, even those that were not created from the Stability API.

Besides this image upscaling tool, Stability AI is also known for its other APIs for generating and editing images powered by artificial intelligence. Its image model called Stable Diffusion is open source and has already been integrated into apps like Draw Things and PicsArt.

Leave a comment

Your email address will not be published. Required fields are marked *