How to fix a blurry image after compression

The honest answer first: mostly you cannot. Detail destroyed by compression is not stored anywhere. What you can do is find the original, or make the damage less obvious.

What is and is not possible

Recover the lost detail
No. It was discarded and is not stored anywhere.
Find the original
Yes, and this is the only real fix. Check cloud backups and sent folders.
Sharpening
Masks softness by increasing edge contrast. Does not restore anything.
AI upscaling
Invents plausible detail. Useful sometimes, but it is invention, not recovery.
Re-saving at high quality
Achieves nothing except a larger file.
Prevention
Compress once, from the original, at quality 80 or above.

What to try, in order

  1. Step 1: Look for the original everywhere

    Cloud photo backups, your camera roll, the Sent folder of your email, the chat you received it in, a memory card you have not formatted. This is the only route that genuinely restores quality, and it succeeds more often than people expect.

  2. Step 2: Work out what caused it

    Over-compression looks blocky and blotchy in flat areas. Upscaling looks uniformly soft with no fine detail anywhere. Motion blur or missed focus is not a compression problem at all. The cause determines whether anything is worth trying.

  3. Step 3: Try mild sharpening, and stop early

    A small amount of unsharp mask or a sharpening slider increases contrast at edges, which reads as crisper. It adds no detail, and pushed too far it produces halos and amplifies the very compression artefacts you are trying to hide.

  4. Step 4: Consider displaying it smaller

    Softness is relative to display size. An image that looks poor at full width often looks perfectly acceptable in a smaller column or as a thumbnail. This is unglamorous and it works.

Understanding what happened

Why the detail cannot come back

Lossy compression discards information deliberately. The encoder decides certain fine detail is unlikely to be noticed, rounds those values to zero, and writes a file that does not contain them.

The file that reaches you has no record of what was removed. There is nothing to decode, nothing to undo and nothing to recover — which is why re-saving at quality 100 achieves precisely nothing except a larger file containing the same degraded image.

Any tool promising to restore a compressed image is either sharpening it or inventing new detail. Both can be useful; neither is recovery.

Diagnosing the cause from the symptom

Blocky squares in flat areas
Over-compression. Quality was set too low, or the image went through several lossy saves.
Uniformly soft, no fine detail anywhere
Upscaling. Someone enlarged a small image, and there was no detail to enlarge.
Grey haze around text or lines
Wrong format. Line art in a lossy encoder produces ringing at every quality.
Muddy, smeared, blotchy overall
Generation loss from repeated re-saving, typically via chat apps.
Sharp in places, blurred in others
Not compression. Missed focus or subject movement, and nothing here will help.

What sharpening actually does

Sharpening finds edges and increases the contrast across them, making a light side lighter and a dark side darker. Your eye reads that heightened contrast as crispness.

No new information is created. And because compression artefacts are themselves edges — the boundaries of 8 × 8 blocks, the ringing beside a letter — sharpening amplifies them along with everything else. Applied heavily to a compressed image it makes the damage more visible, not less.

Use a light touch, view at display size, and stop as soon as it looks acceptable rather than pushing for how it used to look.

AI upscalers, described accurately

Modern upscalers are genuinely impressive. Trained on enormous numbers of images, they produce a plausible high-resolution version by generating detail consistent with what they have seen before.

That word — generating — is the whole point. The model is not recovering your image's lost detail; it is inventing detail that looks right. On landscapes and textures the result is often excellent. On faces it can subtly alter features, and on text it can produce letters that were never there.

This makes them useful for presentation and unsuitable for anything where accuracy matters: evidence, documentation, identification, or any image someone will rely on as a record of what was actually photographed.

Not making it happen again

  • Keep originals. Compress copies, never the master.
  • Compress once. Each lossy save compounds; three passes at 80 is worse than one at 70.
  • Resize before compressing. Fewer pixels at good quality beats many pixels at poor quality, every time.
  • Never enlarge. Upscaling cannot add detail; it only makes the softness bigger.
  • Match format to content. Text and graphics in PNG or lossless WebP, photographs in JPEG or WebP.
  • Send originals as documents through chat apps if quality matters, which bypasses their re-compression entirely.

Do it here instead

These run in your browser on any device, so nothing is uploaded and there is nothing to install.

  • Lossless compressor

    Zero quality loss, every pixel preserved exactly. Honest about how modest the saving is.

    Open the tool
  • Compress JPEG

    Shrink JPG and JPEG photographs with a quality slider or an exact target size.

    Open the tool
  • Resize image

    By pixels or percentage, aspect ratio locked, across a whole batch.

    Open the tool

Questions

Frequently asked questions

Every answer below is present in the page source, expanded, so it can be read without opening anything.

Can I restore an over-compressed image?

Not genuinely. The discarded detail is not stored in the file, so there is nothing to decode back. The only real fix is to find the original — check cloud backups, your camera roll and email sent folders before trying anything else.

Does re-saving at 100% quality help?

No. It produces a larger file containing exactly the same degraded image. The quality setting controls how much is discarded on this save; it cannot recover what a previous save already removed.

Do AI upscalers really fix blurry photos?

They generate plausible detail rather than recovering the original — which often looks very good, particularly on landscapes and textures. Be cautious with faces, where features can shift subtly, and with text, where letters can be invented. Never use one for anything that must be an accurate record.

Why did my photo get blurry after sending it?

Messaging and social apps re-compress everything you send. If the image had already been compressed once, that is a second lossy pass on top, and the damage compounds. Sending as a document bypasses it on apps that offer the option.

Will sharpening make it look better?

Sometimes, in moderation. Sharpening raises contrast at edges, which reads as crisper, but it adds no detail and it amplifies compression artefacts along with everything else. Apply lightly, judge at display size, and stop early.

How do I stop this happening again?

Keep your originals and compress copies. Compress once rather than repeatedly. Resize before compressing rather than instead of it. And never enlarge an image — there is no detail to enlarge.

From the blog

Related reading

More background from the blog.

  • Image Compression

    JPEG Quality Settings Explained: What the Number Actually Means

    Quality 80 is not 80% of anything. It is an index into a quantisation table. Measured on one photograph: dropping from 100 to 90 removes 77% of the file and 4.2 dB of fidelity; dropping from 40 to 30 removes 17% and costs almost as much.

    8 min read

  • Practical Guides

    How to Compress Images for a Website Without Wrecking Them

    Resize before you compress. Measured on one photograph: half the width at quality 80 produced a 16.4 KB file, while full width at quality 40 produced 20.6 KB and looked far worse. The order of operations matters more than the settings.

    8 min read

  • Image Formats

    PNG vs JPEG: When to Use Each, With the File Sizes to Prove It

    PNG stored the same photograph in 1274.1 KB where JPEG needed 67.5 KB — nineteen times larger. On a flat graphic PNG won. On a smooth gradient PNG lost to JPEG by a factor of two, which surprises most people.

    9 min read