PictureMatters Editorial TeamPublished
GUIDE

How to Check If an Image Has Been Edited

There is no single "edited" switch inside a file — but edits leave traces. Here are the honest methods, what each can tell you, and where they run out.

The short answer

You can rarely prove an image was edited from the file alone. You can often find signals: editor software tags, missing original capture data, re-encode evidence, structural oddities. Weigh them together and, usually, a picture emerges — literally.

Signal 1: Software traces

Editors and apps often leave their name in metadata: Photoshop, Lightroom, Snapseed, Canva in EXIF's Software field or XMP's history. This is one of the strongest hints — but only for the last save. A strip-and-export workflow leaves nothing, and platforms overwrite it. PictureMatters surfaces these as "processing trail" markers.

Signal 2: Missing original metadata

A "photo from a camera" with no camera EXIF at all is suspicious in contexts where originals are claimed. Careful though: sharing strips metadata too. The useful question is whether the story around the file matches its metadata survival — a news photo posted as "original" with zero EXIF deserves questions.

Signal 3: Compression and structure

  • Progressive JPEG — cameras shoot baseline; progressive encoding typically means a re-save or platform re-compression.
  • Double compression — quality-estimate patterns can indicate the file was JPEG-compressed more than once.
  • Appended data — bytes after the image's end marker suggest someone modified the file container; sometimes editors, sometimes smuggled payloads.
  • Wrong dimensions for the claim — a "RAW original" at 1080×1350 is a repost, not an original.

PictureMatters reports all of these in the structure map, encoding card and trailing-data card of every report.

Signal 4: Hashes and history

If you have a claimed original, compare SHA-256: different hash means the file changed at the byte level (though not necessarily visually — that's what the perceptual fingerprint is for). For published images, reverse-image search finds earlier versions to compare against.

What does not work

  • "It looks AI/edited" — human judgment of pixels is unreliable and increasingly moot.
  • Missing metadata = fake — absence is normal; see above.
  • Single-signal certainty — one weak signal proves nothing; weight comes from convergence.

Run a check now

Upload the image to PictureMatters image forensics — you'll get the software trail, structure map, encoding facts, entropy profile and a journey classification in one report, with facts separated from interpretation. Free, no account.

Use the tool

Want to inspect an image? The PictureMatters analyzer reads metadata, structure, entropy and provenance signals in seconds — free, without an account.

Analyze an Image

Frequently asked questions

Can forensics prove who edited an image?

No. Files can show which software likely saved them last — not who operated it. Attribution needs sources and context beyond the file.

What about photos shared on WhatsApp — can I check those?

WhatsApp strips metadata and re-compresses, so an edited photo and an untouched photo look structurally identical after transit. Compare against an original source instead.

Related guides