Increases an image's resolution using a neural network that reconstructs plausible detail instead of simply stretching pixels — the result stays sharp even at 2x-4x enlargement.
For example: You only have a small old photo that needs to be printed on a large canvas — upscale it first to avoid a blurry print.
Who uses it most: Photographers, designers, print specialists, old-photo restorers
Regular enlargement just stretches existing pixels and blurs the image. The neural network reconstructs plausible detail instead, so the result looks noticeably sharper.
The first time you use it, the model (a few megabytes) downloads and gets cached in your browser. Every run after that uses the cache and is faster.