EXIF Viewer: See Every Field Hidden in a Photo

Every time a phone or a camera saves a photo, it writes a second, invisible record and tucks it into the same file. That record is the metadata, and the largest part of it follows a standard called EXIF, short for Exchangeable Image File Format.

The standard dates back to the mid 1990s, when digital cameras needed a way to store the settings a photographer would once have written in a notebook. Shutter speed, aperture, focal length, and the film speed equivalent all went into the file so they could be reviewed later. That original purpose still shows in the field names, many of which describe mechanical parts a phone does not physically have.

What changed is scope. If you have ever wondered what information is stored in a photo, the answer on a modern phone is: the device make and model, the specific lens used on a multi lens camera, the firmware version, the software that last saved the file, the orientation sensor reading, a small embedded thumbnail, a full colour profile, and, when location services were switched on, the latitude and longitude of the spot where the shutter fired.

EXIF is not the only standard in there. IPTC came from the news industry and holds captions, credits, and copyright lines. XMP is a newer, more flexible format that editing software uses to record its own history. ICC holds the colour profile. A single photo routinely carries fields from three or four of these at once, which is why a full count of entries in an ordinary phone photo often passes 100.

None of it is visible when you look at the picture. Reading it takes a tool that opens the container and walks through the blocks one at a time.

The privacy problem with photo metadata is not abstract. It is specific, and it follows a pattern.

Someone lists an item for sale and photographs it on the kitchen table. The listing shows a chair and a price. The file also carries the coordinates of the kitchen, accurate to a few metres, along with the timestamp of a weekday afternoon. Anyone who downloads the image rather than just viewing it has the seller's home address and a rough idea of when they are there.

The same pattern repeats in quieter ways. A photo posted to a forum to illustrate a point also dates and places the person who posted it. A picture sent to a stranger for a repair quote carries the workshop, or the house. A holiday photo shared while the trip is still happening announces that nobody is home.

A common assumption is that platforms handle this. Many services do strip metadata when they process an upload, because it saves them bandwidth as much as it protects anyone. Many others do not, and image hosts in particular often store and serve the original file untouched. Behaviour varies between services, between the web and the app version of the same service, and between this year and last. Treating it as solved is the mistake.

There is a second gap that catches people who do think about this. Stripping on upload protects the copy that platform serves. It does nothing for the copy sitting in a chat thread, in a shared album, in an email attachment, or on the memory card. Those files still carry everything.

The practical response is to check whether a photo has location data before you post it, rather than guessing. Once you know what is in the file, removing the metadata is a separate and much shorter decision.

Most online metadata readers show a dozen rows and stop. Camera, date, exposure, maybe a location. That covers the common questions and hides everything else, which is a problem when the field you care about is one of the ones that got hidden.

This tool shows the output in three layers, and the third layer is the point.

The first layer is a summary card: dimensions, total field count, camera and lens, capture date, exposure, aperture, focal length, and the software that last touched the file. These are the rows that answer a question in one glance.

The second layer is one card per metadata block, in the order worth reading them. GPS comes first because it matters most. IFD0 holds the basic image directory, including make, model, orientation, and resolution. EXIF holds the capture settings proper. IPTC and XMP hold descriptive and editorial fields. ICC holds the colour profile, which on a phone photo is often the single largest block by field count. IFD1 describes the embedded thumbnail, which is why a second, smaller set of dimensions sometimes appears. Interop holds a couple of compatibility flags.

The third layer is a single table listing every field found, labelled with the block it came from. Nothing is filtered out for being obscure or awkward to display. Values that are genuinely unreadable, such as the numeric curves inside a colour profile, are shown by their size in bytes rather than as a wall of digits, but their presence is still recorded.

The same field name can appear in more than one block, and the values do not always agree. A date can sit in IFD0 and in EXIF and in XMP, written by three different pieces of software at three different moments. Seeing all three is more useful than seeing whichever one a tool decided to trust.

Four numbers describe how almost any photo was taken, and they are worth understanding even if you never touch a manual camera. Together with the make, model, and lens fields, they answer the question of what camera took a photo and how it was set up at that moment.

Exposure time, also called shutter speed, is how long the sensor was collecting light. It is written as a fraction of a second: 1/250 is fast enough to freeze someone walking. Anything slower than roughly 1/60 starts to show hand shake. A value of 1 second or longer means the camera was almost certainly resting on something.

Aperture is the size of the lens opening, written as an f number such as f/1.8 or f/8. The counterintuitive part is that a smaller number means a wider opening. A wide aperture lets in more light and throws the background out of focus, which is why portrait modes report low f numbers.

ISO is how much the signal from the sensor was amplified. Low values such as 50 or 100 mean there was plenty of light. High values such as 3,200 mean there was not, and the picture will carry visible noise as the price.

Focal length is written in millimetres and describes how much of the scene fits in the frame. Low numbers are wide, high numbers are telephoto. On a phone with several rear cameras, the focal length is the most reliable way to tell which lens actually took the shot.

Read together, they reconstruct the moment. A short exposure at a high ISO with a wide aperture describes an indoor photo of something moving. A long exposure at ISO 100 describes a tripod and patience.

None of this changes the image itself. If the file needs to be smaller or a different shape, that is what the image compressor and the resizing tool are for.

A photo does not store its location as a place name. It stores two numbers and two letters.

Inside the file, latitude and longitude are written the old fashioned way, as degrees, minutes, and seconds, with a separate field holding the hemisphere reference: N or S for latitude, E or W for longitude. Those pieces have to be combined before they mean anything, which is why a raw metadata dump can show a latitude of 41 alongside a lonely letter and give no hint that the two belong together.

Converted to a single decimal pair, the same location becomes something like 41.008° N, 28.978° E. That is precise. The sixth decimal place of a degree is roughly a tenth of a metre, and while a phone's GPS is nowhere near that accurate in practice, it is routinely good to within a few metres outdoors. Indoors, or in a dense city, it can drift by a block or fall back on nearby wireless networks instead.

Two numbers still tell most people nothing, so when a photo carries coordinates this tool puts them on a map with the point marked. The same location can be opened in OpenStreetMap, Google Maps, Apple Maps, Bing Maps, or Yandex Maps, because the map people trust and have installed differs from country to country.

One honest note about how that works. Drawing a map means asking a map service for the tiles around those coordinates, so that service learns the coordinates. The photo itself never goes anywhere: it is read on your device and stays there. If even the coordinates are too much to share, the pair is printed as text beside the map and you can read it without loading anything.

Altitude, heading, and speed sometimes ride along in the same block. A photo taken from a plane or a moving car often carries all three.

Inspecting a photo and cleaning it are two different jobs, and they deserve two different tools. This one only reads. It never rewrites the file, never re-saves it, and never touches a single pixel.

That restraint is useful more often than it sounds. A photographer checking how someone else exposed a shot has no reason to alter the file. A researcher establishing when and where an image was made needs the original intact, because a re-saved copy is weaker evidence. Someone sorting an old archive wants the capture dates preserved, not stripped, because those dates are what makes the archive sortable at all.

Reading first also changes what people do next. The abstract advice to remove metadata before sharing tends to be ignored. Seeing that a specific photo carries a specific address, a specific timestamp, and the name of the phone that took it tends not to be.

When the answer is that the file should not go out as it is, the next step is a separate cleaning tool that strips every EXIF, XMP, and IPTC field and hands back a clean copy. For JPEG files it does that without re-encoding, so the visible image is identical byte for byte and there is no quality argument to have.

A reasonable habit for anything published in volume: look at one file from the batch to learn what the camera writes, clean everything that goes out, and keep the untouched originals archived somewhere private. The metadata that is a liability in public is genuinely useful at home.

Files that arrive in a phone specific format can be brought over first with the HEIC conversion tool, since a converted copy carries the original metadata along with it and needs the same inspection.