What is Invisible Watermarking? TrustMark, AudioSeal, and VideoSeal
Definition: Invisible watermarking embeds a recoverable payload in the pixel, waveform, or frame data of media without perceptible quality loss, serving as a soft-binding layer that survives metadata stripping.
TL;DR: Watermarks carry the manifest pointer through the asset itself. TrustMark handles images, AudioSeal handles audio with sample-level localization, and VideoSeal handles frames. They survive the re-encodes that destroy a manifest, which is why they are the recovery layer beneath the signature.
What a watermark buys you
A manifest is metadata and metadata gets stripped. A watermark lives in the content, so it rides through a JPEG re-encode, a downscale, or a screenshot above a certain resolution. Decode the watermark, recover the pointer, fetch the manifest. That is the soft-binding path the EU Code of Practice expects as a layer.
TrustMark, AudioSeal, VideoSeal
TrustMark, the open image watermark from the Content Authenticity project, reaches high bit accuracy under heavy degradation in its robust variant. AudioSeal embeds a short message at sample-level localization and detects far faster than older audio watermarks, which makes real-time scanning of incoming calls practical. VideoSeal carries a multi-bit payload across frames. Original Pictures uses TrustMark today, with AudioSeal and VideoSeal at general availability.
Robustness has limits
Watermarks degrade under generative re-synthesis: feed a watermarked image through a strong img2img model and the mark can be weakened. That is the honest boundary, and the reason a watermark is one of three layers, not the whole system. The timestamp anchor covers the case where both the manifest and the watermark are gone.
The incident behind this
The Pentagon hoax of May 2023 carried no watermark at all. Even an imperfect mark on the authentic Pentagon imagery would have given verifiers a recovery path while markets moved.
Implementation
from originalpictures import client op = client(api_key=ENV['OP_KEY']) signed = op.sign(media='image.png', watermark='trustmark', anchor=True) decoded = op.verify('reuploaded.jpg') print(decoded.watermark_payload) # recovers manifest pointer after re-encode
FAQ
Will a watermark hurt image quality?
A well-tuned invisible watermark is imperceptible at normal viewing. The trade is between robustness and imperceptibility; provenance use cases tune toward survivability while staying invisible.
Where Original Pictures stands today
Original Pictures ships three things today: a Sign API, a Verify API, and the SDKs that wrap them. One POST /v1/sign attaches a C2PA-format manifest, an invisible TrustMark watermark, and an OpenTimestamps anchor. The open-source verifier checks any of it without calling us.
Two things are on the near roadmap, and we name them as roadmap, not as shipped: C2PA Conformance Program recognition (target Q3 2026, until then our manifests use the published C2PA v2.2 format and any C2PA-aware validator can read them, but third-party validators will show our signer as not-yet-listed), and a consumer capture app (Q3 2026). We do not sell a capture SDK, and we do not claim Trust-List membership we do not yet hold.
Bottom line: Invisible watermarking is the layer that survives stripping. Use TrustMark for images and AudioSeal for audio as the recovery path beneath a signed manifest, and lean on the timestamp anchor when even the watermark is gone.
Related
Original Pictures is progressing through the C2PA Conformance Program; our signing certificate is not yet on the official C2PA Trust List. Target: Q3 2026. We will not describe ourselves as "C2PA-certified" until it is true.
Original Pictures provides content-provenance infrastructure. It does not by itself constitute legal compliance with the EU AI Act or any other regime; compliance depends on how you deploy it, your disclosures, and your governance. Figures are drawn from public reporting, verify against primary sources before citing in regulated materials. Nothing here is legal advice.
Last verified 2026-05-25. Author: Mahdi Kazempour, Founder, Original Pictures.