GUIDE · 4 MIN READ

What invisible watermarking actually does.

A practical guide to embedding image identity, recovering it later, and interpreting the result responsibly.

An identifier inside the pixels

A visible watermark adds a logo or text to an image. An invisible watermark encodes a signal within image pixels, with the aim of preserving the viewing experience. In Etchv, the signal represents a 256-bit digest of the reference data supplied with the image.

Embedding is only half the workflow

When you embed, save the returned watermark ID and link it to your own asset or delivery record. The identifier has little practical value if you do not retain the information needed to interpret it later.

Detection attempts to recover that signal

The detection endpoint returns whether a watermark was detected, a confidence value and a watermark ID when available. The confidence is the mean certainty of decoded bits. It is not a calibrated probability of ownership or the likelihood that a named person leaked an image.

Watermarks have limits

Compression, cropping, resizing and other transformations can affect recovery. No detection result should be treated as an unconditional guarantee. Test representative images, include unwatermarked controls and evaluate performance against your own requirements.

What Etchv supports today

Etchv exposes image embedding and detection endpoints. The embed response is a PNG, and uploads are limited to 20 MB and 40 megapixels. The API does not maintain your delivery-to-recipient mapping or scan external websites on your behalf.

Ready to evaluate with your own content? Talk to the Etchv team.

MAKE EVERY COPY ACCOUNTABLE

Your content has a story.
Keep it connected.

Let’s find the right watermarking workflow for your platform.

Talk to the team