Photo SignalsAI Watermark Remover

Remove metadata, GPS and hidden files — without touching a pixel

The file is rebuilt from the parts we understood. Anything else — including whatever a tool we have never seen decided to store in it — does not survive the rebuild.

  • GPS and geotags
  • Serial numbers
  • Hidden thumbnails and video

Open an image to clean it

  • Nothing is uploaded
  • No account
  • Free, no limit
The method

Rebuild, not delete

Most strippers delete the fields they know about. That leaves the fields they do not know about, the bytes wedged between segments, and anything appended after the file's end marker. This one assembles a new file out of the picture data and the handful of structures that decide how it displays. Everything else is gone because it was never copied across.

What one photo file is made of, and what survives cleaning A photo file drawn as a row of blocks. Camera metadata, editing history, a hidden thumbnail, a Content Credentials manifest, an unknown tracker block and an appended video are all removed. The block holding the picture itself is kept byte for byte, along with a thirty-two byte orientation block. YOUR FILE BEFORE EXIF GPS, serial XMP names, history THUMBNAIL pre-crop copy THE PICTURE compressed pixel data C2PA AI provenance TRACKER unknown VIDEO appended MP4 rebuilt from the parts we understood YOUR FILE AFTER THE PICTURE byte for byte identical ORIENTATION 32 bytes, kept Everything else is gone because it was never copied across — including block types nobody has named yet.
Most strippers delete the fields they know about. This one keeps only what it needs and rebuilds the file, so an unknown block dies because it is unknown.
Exactly what goes

Every family, named

  • Watermarks you can see

    Removed — you mark it, it fills in

    • Stock-library tiling and corner logos
    • Burned-in dates and dashcam or CCTV timestamps
    • "Confidential", "Draft", "Sample" bars and diagonals
    • Channel bugs and app export badges
    • Screener and per-recipient overlays

  • AI generator marks

    Badges filled · manifests and fields removed

    • The Gemini and Nano Banana sparkle
    • Corner labels from Doubao, Jimeng, Qwen, Kling
    • Phone "AI edited" and Magic Editor badges
    • C2PA Content Credentials — Adobe, OpenAI, Cloudinary, camera makers, cloud image pipelines
    • Stable Diffusion's prompt, seed and model in the PNG text chunk
    • ComfyUI's whole workflow graph, filesystem paths included
    • China's TC260 AIGC labelling fields

  • Container metadata

    Removed — pixels untouched

    • EXIF — GPS with altitude, bearing and timestamp; make and model; body and lens serial numbers; shutter count; owner name from the firmware; timezone offset; how the fix was obtained; air pressure
    • XMP — editing history, document identifiers that link your files to each other, face regions naming the people in the frame, the original RAW filename, the pre-crop dimensions
    • IPTC — author, credit, city, contact details
    • MakerNotes — the manufacturer's private block
    • Photoshop IRB, JFIF, MPF and unknown APP segments
    • PNG chunks — tEXt, zTXt, iTXt, eXIf, tIME and anything unrecognised
    • GIF and WebP comment and application chunks

  • Whole files hidden inside the picture

    Extracted, shown to you, then removed

    • The embedded thumbnail — a copy from before you cropped or blurred
    • Motion photo video — an entire MP4 appended after the picture ends, with sound, covering the seconds either side of the shot
    • The MPF second full-size image some phones store
    • Ultra HDR gain maps and depth maps
    • Full-resolution previews inside RAW and DNG
    • PSD layers holding content you thought was deleted
    • Anything appended after the end marker, and the slack between segments
    • Payloads in fully transparent pixels, and messages in the low bits

  • Fingerprints of the software

    Removed by re-encoding at level 2

    • Compression tables that name the camera model or the program, with no metadata involved at all
    • Scan layout, subsampling and chunk ordering that identify the writer
    • ICC profiles named after a specific monitor calibration

  • Traces of the camera and the printer

    Printer dots removed · sensor noise reduced and measured

    • Printer tracking dots — the pale yellow grid colour printers add to every page, encoding the printer's serial number and usually the time of printing. Found in any scan of a printed page and removed at level 2, with the count and the lattice spacing reported
    • Sensor pattern noise — the signature of the individual camera body, not the model. Attenuated at level 4 and reported as a number, never claimed removed
    • Hot and dead pixel maps unique to one body

  • SVG, which is a document

    Removed by an allowlist

    • Beacons — references to an image on another server, contacted every time the file is opened
    • Executable script inside the image
    • Illustrator and Inkscape namespaces with layer names and local paths
    • Embedded raster images, which carry their own EXIF and are cleaned recursively

  • Writing inside the picture

    Found and pointed at — never read aloud

    • QR codes and barcodes on a badge, a boarding pass, a ticket or a parcel label. They usually encode a link with an identifier in it, a booking reference or a name — as readable to a stranger as the text beside them, and nobody thinks of them as writing
    • We say a code is there and where it is, so you can crop it or paint over it. We do not decode what it says: reading out what is on somebody's ticket is not this site's business

  • What nobody can remove

    Named and explained, never claimed

    • Imperceptible watermarks whose detector is held privately by its owner
    • Perceptual-hash databases platforms keep — not in your file, so nothing in your file changes them
    • A provider's own record of what it generated

    Why, and what our transforms actually do to them

Not every file hides things in the same place

What your file actually is

A tool written for JPEG finds nothing in a HEIC, because a HEIC does not keep its metadata in a segment — it keeps it as a separate item in a table of byte offsets. These are the four containers that hide things somewhere unexpected, and what happens to each.

  • HEIC, HEIF and AVIF

    What an iPhone shoots

    The serial number, the GPS fix and Apple's own dictionary live in mdat alongside the picture, reachable only through the offset table. They are overwritten where they lie and the file keeps its length on purpose: every offset in the file points at another offset, and a shorter file is one that opens as a grey rectangle.

  • RAW — DNG, CR2, NEF, ARW

    The format that gives most away

    A raw file carries a full-resolution preview of the picture as it looked before your edits. Crop a face out in the raw converter and that preview may still show it. It goes, along with the body serial and the original file's path on your machine. The sensor data is never touched: strips are cleared only in a directory that declares itself a reduced-resolution copy.

  • WebP

    Assumed clean because it is new

    It is not. WebP has dedicated EXIF, XMP and colour-profile chunks, and a phone converting a HEIC for upload copies the camera metadata straight across, GPS included. The file is rebuilt from its allowed chunks, and the header is corrected so it no longer advertises what it no longer has.

  • The video inside the photograph

    Offered back before it is removed

    A motion photo is a whole video file appended after the picture ends, with sound, covering the seconds either side of the shot — in JPEG and in HEIC alike. Most people do not know it is there. It is shown to you and offered as a download first, because deleting something you did not know you had is not the same as choosing to.

Read, but not rewritten: JPEG XL and ICO. They are identified and reported, and nothing is changed in them. A tool that claimed every format it could open would hand back a file it had not cleaned.

Old enough that nobody checks them

The three people assume are harmless

These formats predate the habit of looking. That is exactly why things end up in them — a picture nobody inspects is a good place to leave something.

  • GIF

    Two blocks that are pure payload

    A GIF is a stream of blocks, and two of them draw nothing. The comment block is where every conversion site, meme generator and export tool signs its work, and some write more than a name. The application block was designed to hold an animation's loop count and is routinely used to carry an entire XMP packet instead — editing history and the original file's name, inside a GIF. The loop count survives; the rest does not.

  • BMP

    No metadata section, so nobody looks

    There is nowhere in a BMP to put EXIF, which is why it gets treated as safe. What it has instead is slack: two header fields that must be zero and often are not, a gap before the pixels are declared to start, and whatever sits after the final row. None of it is drawn. And a BMP can declare its colour profile as a link — in which case the "profile" is a file path on the machine that saved it, account name and folders included.

  • PSD

    A working document sent as a picture

    Photoshop keeps IPTC, EXIF and XMP in a resource list, alongside something a JPEG has no equivalent for: a thumbnail of the flattened document, rewritten every time you save — the same before-the-edit leak a raw file has. Then there are the layer names, which are not metadata at all but something a person typed: client names, version numbers, and whatever the layer was called at two in the morning.

A PSD's layer names are overwritten in place, and the pass is allowed to refuse. The document is a chain of nested lengths and a wrong figure gives back a file that will not open, so the names are overwritten character for character with every byte count kept identical. If the layer section does not parse exactly, nothing is touched and you are told — a refusal is never quietly counted as a clean.

Two things kept on purpose

  • Which way up the picture goes

    Delete it and portrait photos display sideways. It is re-written as a block containing that single fact and nothing else — thirty-two bytes, with no room in them for a serial number.

  • The colour instructions

    Removing them shifts the colours. They can carry the name of a specific monitor calibration, so they are reported to you and removed one level deeper, where the colour change is a stated cost rather than a surprise.

Narrower jobs

To read the camera fields before deleting them, start with the EXIF page. For several images at once, the batch tool takes up to ten. If the file is an SVG it is not a picture but a document that can run code — it needs the sanitiser instead. And a PDF carries the same kinds of thing: Steady PDF does that job the same way, in the browser.

The pixels are untouched. At this level the compressed picture data is copied across byte for byte. What comes out is the picture that went in — not re-compressed, not re-saved, not quietly degraded.