Document robustness laboratory
How far does your watermark resist?
Import an image or PDF, identify the mark to test, run a local reconstruction and observe what resists, what disappears and what is degraded.
Processed locally in your browser
The document, mask, fingerprint and result are not sent to the After Algorithm analysis API.
Drop the document to test
or click to select a file
JPEG · PNG · WebP · multi-page PDF2.3.1 test release: multipage PDFs are supported when every page is a full-page JPEG image. Vector or mixed PDFs are explicitly rejected rather than partially analysed.
The file stays on this device during the test.
Ready. Add a file to begin.
Identify the watermark to test
Sample a watermark colour for assisted selection, then refine the mask with the brush. The red overlay shows pixels included in the test.
Page 1 / 1
Reference original (optional)
Add the unwatermarked version only if you have it. It is used to measure reconstruction, never to reconstruct the document.
Filigrane Lab
Comparison
Indicative reading, not a universal security score.
The reconstruction is an estimate generated from neighbouring pixels. It does not recover the original pixels hidden by the watermark.
Blue team
Strengthen the watermark
A visible watermark can deter misuse but is not an integrity proof. To increase removal cost:
- Personalise the mark with purpose, recipient and date when relevant.
- Avoid a perfectly periodic pattern and slightly vary position, density or opacity.
- Cross informative zones without making the document unreadable.
- Combine visible marking with an integrity mechanism or digital signature when authenticity must be verified.
- Test several document families: flat colours, text, photography, logos and fine detail.