Cleaning options
Hidden characters and AI tags are always removed. These are the extras, and they are already set the way most people need them.
Drop your files in
Text, Markdown, HTML, PDF, Word, ODT, SVG and code files.
Drop your images in
PNG, JPEG, WebP, AVIF and HEIC. The picture is never re-encoded, so nothing about how it looks changes.
Removed from every image
Content Credentials and C2PA records, XMP and EXIF generator fields, PNG text chunks, JPEG app segments and WebP metadata.
Drop your videos in
MP4, MOV, M4V, 3GP, WebM and MKV.
Metadata cleaning
Content Credentials, C2PA records, XMP packets and generator tags come out of every video. Nothing is re-encoded, so the file keeps its exact size and the picture is untouched.
Erase a logo
Drop your audio in
MP3, FLAC, WAV, OGG, Opus and M4A.
Tags come out, sound stays
Nothing is re-encoded, so the audio is bit for bit what it was. Only the tag blocks are cleared, using each format's own padding.
- MP3 — ID3 frames blanked, trailing ID3v1 and APE tags removed
- FLAC — comment, picture and application blocks turned into padding
- WAV — LIST, INFO, broadcast and ID3 chunks replaced with JUNK
- OGG / Opus — comment header blanked and the page checksum repaired
- M4A — handled by the video engine, same underlying boxes
See everything inside a file
Drop anything in — a photo, video, song, PDF or document — and read every piece of information it carries. Camera and lens, the date it was taken, where it was taken, who is named as the author, editing history, track titles, the lot.
Clean text directly
Paste anything in. Invisible characters, hidden tag blocks and zero-width carriers are found and taken out.
Process a whole folder
Reads files straight from disk, so the upload limits do not apply. Your originals are never modified — cleaned copies go to the output folder.
If you just want to clean a file
Drop it in and press Clean. The boxes are already set the way most people need them, so you can ignore everything below.
Some things happen no matter what you tick. Every file gets its hidden characters pulled out and its AI tags wiped. There is no switch for that, because there is no good reason to skip it.
The rest of this page explains the tick boxes, in case you want to know what you are changing.
The tick boxes, one by one
Normalise odd characters
Some letters can be typed more than one way. The "fi" in "file" can be two normal letters, or one single character that just looks like two. A wide Ａ and a normal A are the same letter to you, but different characters to a computer. Swapping a few of them around is one of the quiet ways text gets marked.
Ticked: everything gets rewritten the normal way, so those tricks have nowhere to hide.
Unticked: the odd characters stay exactly as they are.
Untick it if you need the file back byte for byte, like source code, or something you are going to compare against the original.
Fix disguised letters
The Russian letter о and the English letter o look identical on screen. They are not the same character. Change a handful of letters in a paragraph and you have a fingerprint nobody can see, one that survives copy and paste.
Ticked: those lookalikes get turned back into ordinary English letters.
Unticked: they are left alone.
Keep ordinary file details
Photos, videos and music files carry a hidden information panel. Camera model, the date, the software used, song titles, and often the exact spot on the map where it was taken.
Unticked (the default): all of it goes. Nothing is left behind.
Ticked: the ordinary camera and song details stay, and only the AI-related parts are removed.
Leave it unticked unless you have a reason. The location buried in a holiday photo is usually the thing you would least want to post in public.
Clean inside documents
A document has two separate places to hide something. The properties box, where the author and title live, and the actual words on the page.
Ticked: both get cleaned. The invisible characters sitting between your visible words are
pulled out too.
Unticked: only the properties box is cleaned, and anything hidden in the text itself stays.
There is not much reason to turn this off. Leave it on.
Pixel watermark removal
Some AI image tools hide a pattern in the pixels themselves. You cannot see it, and deleting the file's hidden information does nothing to it, because it is part of the picture.
The only known way to disturb that is to feed the image through another AI model and have it redraw the picture. That is what this option is for. It is missing on a normal install because the model is a separate download of roughly ten gigabytes, it needs a proper graphics card to run in any sensible time, and it repaints your image, so the result is never quite the original.
It has nothing to do with video. Both available backends work on still images only.
Erase a logo
This one is different from all the others. Everything else edits hidden information and leaves your file otherwise untouched. This paints over part of the picture.
Ticked: you pick a corner, and that rectangle gets smudged over using the colours around
it. The whole video has to be rebuilt to do it, so it takes a while, uses a lot of processing
power, and the picture quality drops a little. You will usually still see a soft blurry patch
where the logo was.
Unticked: the picture is not touched at all.
It only works on a logo that sits still in one corner. If the logo drifts around the screen, a fixed rectangle cannot follow it, and this will not help.
What this actually does
Everything runs on this machine. Nothing is uploaded anywhere.
- Hidden characters — zero-width spaces and joiners, Unicode tag blocks, variation selectors, direction controls and disguised letters.
- Metadata — C2PA and Content Credentials, XMP and EXIF generator fields, PNG text chunks, document properties, and the metadata boxes inside MP4, MOV and WebM.
- Pixel watermarks — optional, needs a separate model and a graphics card.
Text also gets an AI writing-style score, based on sentence rhythm, word variety and stock phrases. That is a hint, not proof, and nothing is rewritten because of it.
Video: what comes out and what stays
Metadata is stripped without re-encoding. Each metadata box is overwritten in place with padding of exactly the same size, so the file keeps its byte length, every internal offset stays valid, and the picture is untouched.
- Removed: C2PA and Content Credentials manifests, XMP packets, generator and encoder tags, and the WebM equivalents.
- Optional: a logo sitting in a fixed corner, via the tick box in the Video tab.
- Not removed — Sora's moving logo. It drifts around the frame, so a fixed rectangle cannot cover it. Following a moving mark needs frame-by-frame tracking, which is a different kind of problem and is not built here.
- Not removed — SynthID. Google Veo spreads its watermark across every frame and the audio, and it is built to survive re-encoding, cropping and compression. There is no public way to detect or strip it. Nothing here claims to, because nothing here could.
Optional helper tools
Three well-known command-line tools can widen what this app reads. None of them is required — the built-in readers already handle the common formats — and nothing is uploaded to them. They simply do a more thorough job on unusual files when present.
- exiftool — reads and writes the hidden information panel in a very wide range of file types. Extends coverage beyond the formats handled here natively.
- c2patool — the official reader for Content Credentials. With it, a provenance record can be opened and described rather than only detected and removed.
- qpdf — rebuilds the internal structure of PDFs, which helps clean the more awkward ones properly.
Service details
…