How to unblur an image, and when it cannot be done

Zoom in to 100 % and compare what you see with these four cases. Most disappointing results come from using the wrong fix for the kind of blur.

• Compression softness: the photo looks smudged and blotchy, with faint square blocks or halos around edges. Typical for photos sent through WhatsApp or saved from social media. Fixable.
• Phone processing: night and indoor phone photos where faces and textures look waxy or painted, because noise reduction smoothed them. Largely fixable.
• Too few pixels: the image is small and only looks blurry when shown large, such as a thumbnail, a crop of a distant subject or a web copy. Fixable by upscaling.
• Motion blur or missed focus: edges are doubled or streaked in one direction, or everything is evenly out of focus even at a small size. The detail was never recorded. Not truly fixable.

Enhance is trained on phone photos damaged by noise, noise reduction and repeated compression. It removes the compression blocks, ringing and colour blotches and restores natural fine texture, at the same size. Edges that were smeared by compression come back crisp because the information about where they are is still in the photo.

This is the most common kind of blur in shared photos, and it is the one where an AI repair makes the biggest visible difference. For photos from chats see how to fix WhatsApp photo quality; for grainy low-light shots see grainy night photos.

A small image looks blurry when a screen or a printer stretches it. Sharpening filters only raise contrast along edges, which adds halos without adding detail. The real fix is more pixels: Upscale 4× or Upscale 8× rebuild a larger image with clean edges.

If an image was already enlarged badly, for example saved at a large size from a small original, it is often better to start from the small original. See how to fix a pixelated image.

When the camera moves during the exposure, every point of the scene is smeared along a path. When focus is missed, every point spreads into a disc. In both cases the fine detail is mixed together before the photo is ever saved. Mild cases can be improved a little; strong ones cannot be reversed.

Some apps produce sharp-looking faces or text from such photos. They do it by generating a new face or new letters that fit the blur, which can change who the person looks like or what the text says. PhotoRescue does not do that, and its models leave strong motion blur blurred. For why this matters, see faithful vs generative upscaling.

Before giving up, look for another frame. Burst shots, Live Photos, video stills and the other copies in a shared album often include a sharper moment.

Text and faces are where people most want to unblur a photo, and where guessing is most harmful. Our models protect detected text instead of redrawing it, and they do not swap in generated faces. The results are not suitable for identifying a person or reading a number plate as evidence. For screenshots and documents see blurry text in an image.

You get 3 free credits a day, with no sign-up and no card. One photo with one tool uses 1 credit; Upscale 8× uses 2. Free uploads can be JPG, PNG or WebP up to 3 MB and 8 MP, and free results carry a watermark. Pro ($9.99 a month, 200 credits) and Pro+ ($19.99 a month, 500 credits) remove the watermark and accept uploads up to 20 MB and 30 MP. Uploads and results are deleted within 24 hours and are never used to train AI.