viewhack

EXIF viewer: where, when and with what camera a photo was taken

Drop a photo to see its camera, lens, date and exposure, and whether it carries a GPS location, before you send it to anyone. JPEG, PNG, WebP, AVIF and TIFF, and camera RAW files (DNG, NEF, CR2, ARW, CR3), whose built-in JPEG preview you can see and save.

The photo is read by this tab on your own device. It is never uploaded, so checking it for a location does not hand the location to anyone.

What you get

How much a GPS tag gives away

Phones write the position into every photo unless location access for the camera app is switched off. Four decimal places of a degree, as shown here, pin a spot to about 11 metres north–south (one ten-thousandth of a degree of latitude is 11.1 m), which is one house on a street. 55.6761° N, 12.5683° E, the example above, is the square in front of Copenhagen City Hall.

Whether the location survives sharing depends on how the file travels. The big social networks generally remove it when they re-compress an upload. E-mail attachments, cloud-drive share links, AirDrop, and a chat app's "send as file" or "document" option pass the original bytes on unchanged, GPS included. On an iPhone, the share sheet's Options › Location switch leaves it out.

What "without metadata" removes and keeps

A JPEG is a list of segments followed by the compressed image data. The copy is made by leaving segments out, never by decoding and re-encoding the picture, so it loses no quality and the image data from the start-of-scan marker to the end-of-image marker is identical to the original's, byte for byte.

Removed
APP1 Exif: camera, lens, serial numbers, dates, GPS, and the small thumbnail inside Exif, which some editors leave showing the photo as it was before cropping. APP1 XMP: edit history, ratings, face-region names, sometimes location. APP13: IPTC captions, keywords, byline and copyright. COM comments. APP2 Multi-Picture data and anything after the end-of-image marker, where some phones store a second image such as a depth map. Every other APPn segment.
Kept, because the picture depends on them
APP0 JFIF (pixel density and aspect, 16 bytes, nothing personal). APP2 ICC_PROFILE: without it a Display P3 or Adobe RGB photo shows duller colours. The 12-byte APP14 Adobe marker, which says whether the three colour channels are YCbCr, RGB or CMYK: without it some Photoshop-saved JPEGs decode with wrong colours. None of these hold a camera, date, name or place.
Sideways photos
Phones usually store a portrait photo lying on its side and add an Orientation tag (value 6: turn 90° clockwise). The Orientation tag lives in Exif, so the stripped copy shows sideways. When the tag is not 1, the page says so and offers an upright copy instead. That copy is turned on a canvas and re-encoded as a new JPEG at quality 92, so it is not byte-identical, and the page labels it that way.

A photo with the camera's credit or copyright in it loses that too: stripping cannot tell your GPS from someone's byline.

RAW files: the preview inside

A RAW file holds the sensor's unprocessed readings, which need a raw developer (Lightroom, Capture One, darktable, RawTherapee, the camera maker's own software) to turn into a picture. Next to them, cameras store a finished JPEG so the camera's screen and file browsers can show something fast. This page finds that JPEG and shows it. It does not develop the RAW sensor data: what you see and save is the camera's own rendering, with its white balance and picture style baked in.

DNG, NEF, CR2, ARW (TIFF-based)
The file's image directories (IFD0, the ones after it, and their SubIFDs) are walked and every JPEG they name is collected, through the JPEGInterchangeFormat tags or a single strip compressed as JPEG (compression 6 or 7). Lossless JPEG streams, which DNG uses for the raw data itself, are skipped. Canon CR2 files keep a full-size JPEG in IFD0; Adobe DNG Converter writes none, a 1024-pixel one or a full-size one, as you chose when converting.
Canon CR3 (ISO base media, like MP4)
The THMB box holds a 160 × 120 thumbnail, the PRVW box a 1620 × 1080 preview, and the first track usually a full-size JPEG. The largest one that decodes is shown.
When nothing is named
The page falls back to the largest complete JPEG stream (FF D8 … FF D9) it finds anywhere in the file, and says so.

What it cannot do