Clarity AI Upscaler is an open-source AI image upscaler and enhancer that began life on GitHub as a free alternative to Magnific, and grew into the hosted Clarity AI service. The repository, maintained by its creator Philip, has earned more than 5,100 stars and remains the public, self-hostable core of the product.
Clarity rebuilds images at higher resolution using diffusion-based image-to-image techniques: instead of merely interpolating pixels, it regenerates detail while trying to stay faithful to the source. Controls for creativity and resemblance let you decide whether the output should invent rich new texture or hug the original closely, which makes it useful both for restoring small or low-quality photos and for polishing AI-generated art. The hosted Clarity AI service builds on the same ideas with additional closed-source modes, including the high-precision Crystal upscaler and Clarity Pro, plus a production API offered through platforms such as Replicate and Fal.
The open-source release supports local workflows through Stable Diffusion WebUI as well as scripted use, and its fidelity controls made it a popular finishing step in AI art pipelines, where faces, hands and fine textures are exactly the details that naive upscalers tend to destroy.
Freemium. The GitHub project is free and open source (AGPL-3.0) for self-hosting. The hosted Clarity AI service is commercial, with subscription and pay-as-you-go API pricing for the managed web app and the additional Crystal and Clarity Pro modes.
Clarity suits AI artists refining generated images, e-commerce teams sharpening product photography, and designers or agencies that need small visuals enlarged to clean, print-ready resolution.