Enlarge an image without losing quality
Double the size of a photo while keeping edges sharp, using a model that reconstructs detail rather than simply stretching the pixels you already have.
or tap to browse — JPG, PNG or WebP
Best on small or low-resolution images, where there is most to recover.
Why ordinary resizing looks bad
Making an image bigger in an ordinary editor means inventing new pixels between the existing ones, and the traditional methods do it by averaging neighbours. Averaging is exactly the operation that produces blur: a hard edge between two colours becomes a gradual ramp, and the picture looks soft everywhere it used to be crisp. Nothing has been lost, but nothing has been added either.
What this does differently
The model here was trained on millions of pairs of small and large versions of the same image, so it has learned what the detail in a downscaled photograph usually looked like before it was shrunk. Given a small image it predicts that detail rather than averaging towards it, which is why edges stay defined and textures survive. It is reconstruction based on likelihood, not recovery of information that was genuinely destroyed.
What to expect, honestly
This works best on photographs with real detail to build on. A clean 500-pixel photo doubled to 1000 looks genuinely good. A tiny, heavily compressed thumbnail will improve but will not become a high-resolution photograph, because the information was never there and the model is guessing. Faces are the case to check most carefully, since we are all extremely sensitive to small errors in them.
Frequently asked questions
How is this different from resizing in Photoshop?
Standard resampling interpolates between the pixels you already have and cannot add detail. This predicts what the missing detail most likely was. On photographs the difference is obvious at edges and in textures like hair, fabric and foliage.
How long does it take?
A few seconds for a small image, longer as the picture grows, since the work scales with the number of pixels. Large images are processed in overlapping sections and reassembled, so the wait grows steadily rather than suddenly.
Will it fix a blurry photo?
It sharpens the softness that comes from an image being small, but it cannot fix a photo that was out of focus or shaken when it was taken. Those are different problems: the detail was never captured, so there is nothing for the model to reconstruct from.