Invisible, verifiable watermarks for your images
You mark an image with an invisible watermark and can later prove that it is from you. The calculation is done by the marking service: the page submits your order, the service picks it up and reports the result back. Depending on the load, this can take up to half a minute.
Try it out right away
The entire process in three steps, directly here in the browser. No registration required, nothing is uploaded or downloaded. It’s a small simulation that shows you how it works.
Step 1: Select and mark an image at the top.
Step 2: Change the image and see how much the mark withstands.
Step 3: Check if the character is still found.
For your own images and real, verifiable watermarks, continue below.
· Mark
Marking your own images is possible with a free account (1000 tb-coin free). Checking and the demo above do not require registration.
Log in or create an account· Check
Upload any image. The marking service reports whether a watermark was detected, which ID, and how secure.
JPG, PNG, or WEBP · up to 25 MB · multiple allowed (batch)· Test Lab
First mark an image, then enlarge it here, crop it, change the color and have it checked whether the watermark survives this. Each check costs 1 tb-coin.
· Overpaint and check
Load marked image, paint over with the pen and check at the push of a button whether recognition still works. This way you can see how much overpainting the watermark can withstand.
How it works
- Mark embeds an invisible signal plus your ID. The picture looks the same.
- Check reads out the signal again, even after a picture has been passed on, newly compressed or slightly edited.
- The marking service calculates on its own computer. If it is currently off, your order waits until it is back online. You can see the status above.
- Limits, to be honest: no watermark is indestructible. Strong cropping or painting over can remove it. The test lab shows where the limit lies.
- Later: also audio, video and text. Start is image.
Prices
- Create account000 tb-coin for free
- Mark image10 tb-coin
- Check image1 tb-coin
tb-coin is the currency unit of TB-Software (from Artificial-Payment). If a job fails, you automatically get your tb-coin back.
Invisible watermarks, origin tracking and related methods
Bilder entstehen, verbreiten sich und verändern sich heute in Sekunden. Damit wird die Frage nach der Herkunft wichtiger: Stammt ein Bild wirklich von der angegebenen Quelle? TraceInkMarksOrigin beantwortet das mit einem unsichtbaren, prüfbaren Wasserzeichen. Es steckt direkt in den Bildpunkten, trägt eine kurze ID und übersteht übliche Weitergabe, erneutes Speichern und leichte Bearbeitung. Anders als reine Metadaten bleibt es auch dann erhalten, wenn ein Bild ohne Zusatzinformationen weitergereicht wird.
The same goal of making the origin of images traceable (often referred to as “Content Provenance”) is pursued in different ways by several methods:
TraceInkMarksOrigin intentionally uses a robust, pixel-based watermark with verifiable ID and a registry. Metadata methods like C2PA complement this well: one survives distribution, the other carries rich provenance data. Those who want to prove authorship or label AI content should ideally combine both.
Frequently asked questions
What is an invisible watermark?
A signal not visible to the eye that is embedded into the image pixels. It does not perceptibly alter the image but can be read out later and assigned to an ID. This way, you can prove the origin of an image.
How does it differ from Google SynthID or C2PA?
SynthID primarily marks AI-generated images from specific generators. C2PA and Content Credentials work with metadata that can be lost during sharing. TraceInkMarksOrigin robustly embeds the mark into the pixels, so it remains verifiable even after recompression and sharing.
Does the watermark survive editing?
Typical sharing, re-saving, and light editing yes. Strong cropping or painting over can remove it. In the test lab above, you can see live where the limit lies. No watermark is indestructible, and we say that openly.