Your editing software deleted the face from the photo. The EXIF thumbnail might still have it.

IFD1 in the EXIF APP1 block (tags 0x0201/0x0202). Same structure in WebP EXIF chunks and PNG eXIf blocks — thumbnail travels with the EXIF if you preserve it across conversions.

The nuance: most professional editors (Photoshop, Lightroom, GIMP) do regenerate the thumbnail. Many simpler tools, batch processors, and scripted workflows don't. Even tools that usually do may skip it in certain scenarios.

A mismatch is a finding. Clean check ≠ unedited. Photoshop also writes a separate APP13 resource 0x040C thumbnail alongside IFD1 — disagreement between them is its own signal.

snapWONDERS extracts IFD1, JFIF JFXX, Photoshop APP13, C2PA, and more — labels each by provenance, runs automated mismatch comparison across JPEG, WebP, PNG, HEIC, AVIF, and RAW.

Full piece: https://kennethbspringer.au/the-exif-thumbnail-that-gave-the-edit-away/

#OSINT #digitalforensics #imageforensics #JPEG #EXIF #infosec #metadata

You stripped the EXIF. Did you check the MakerNote?

Standard EXIF strippers remove GPS, timestamp, and the public tag block. They don't touch the manufacturer's proprietary MakerNote — which is a separate binary structure per brand, undocumented, and contains internal camera state at capture: AF mode, lens firmware version, scene classification, internal processing flags.

Canon, Nikon, Sony, Panasonic, Fujifilm — all different formats, all reverse-engineered by the forensics community. The data in there goes well beyond what the camera's own display ever shows.

Forensic value: MakerNotes survive many workflows that strip standard EXIF completely. You can confirm camera origin from a MakerNote when the public tags are gone.

Full article: https://kennethbspringer.au/2026/06/10/makernote-forensics-what-camera-manufacturers-hide-inside-every-photo/

#OSINT #digitalforensics #imageforensics #privacy

Another nice #ImageForensics for you. Which panels look more similar than expected?
#ImageForensics challenge of the day, and I made it a bit harder. Two panels overlap (tip: they are adjacent panels). Tell or show me which ones, and you might win an Emoji Award!
#ImageForensics Spot the overlap! Win an Emoji Award!
Oh, come on, @[email protected] Science Translational Medicine, correcting a paper for a photoshopped image? That is absolutely bonkers, and you should do much, much better. #ThisImageIsFine #ImageForensics
*How *not* to handle image manipulation*. Editors, please do not address this level of photoshopping with a correction. In 2020, I flagged Figure 3D of this 2017 BBRC paper. Multiple regions appeared reused within or across panels. pubpeer.com/publications... #ImageForensics /
Wiley's Food Science & Nutrition clearly not checking for image duplication - this should have been caught in 2025. #ImageForensics
Can you spot something unexpected in these four panels? Hint: duplicated groups of fluorescent cells. This is an easy one! Tell or show me at least three sets of problems. The first three correct answers will win an Emoji Award. #ImageForensics
#ImageForensics Kidneys of four differently treated groups of animals. None of these photos should overlap. But is that true???? Tell or show me if you find at least 2 sets of overlaps.