Veo moving-mark detector

Veo can attach two independent signals to its generated video: a visible mark whose screen position shifts across the clip, and SynthID, a sub-perceptual signal embedded in the frame data. These require entirely different verification approaches, which is the entire analytical basis for this page, and conflating them produces a weaker conclusion than treating them separately.

The upload control on this page sends the file to our own media workflow, which scans multiple points across the video for the visible mark rather than trusting a single frame. SynthID's presence cannot be confirmed with certainty by this or any independent tool; only Google's own detector rules on that authoritatively, and no page on this network claims otherwise.

A fixed-position mark reduces detection to one repeatable check that generalizes across a file. A mark whose position varies breaks that generalization: a result on one frame doesn't extend to the next, so a single glance at one point in the clip carries materially less evidentiary weight than a systematic scan across several points spread through the footage.

SynthID sits entirely outside this framework. It isn't a visual signal a scan can locate, since it's engineered to be sub-perceptual, which is precisely why verifying it requires the issuer's own tool rather than any form of independent visual inspection, however careful or systematic that inspection might be.

The scan runs through the upload control on this page.

Step 1: source the original clip

Not a re-share, since re-encoding can shift or degrade the visible mark's appearance across the clip.

Step 2: run the scan above

Our media workflow checks multiple points across the clip, since the mark's position varies and a single frame can miss it entirely.

Step 3: scope the conclusion correctly

The result addresses the visible mark specifically and only that. It carries no implication about SynthID either way.

Treating a clean visible-mark check as evidence against SynthID is a category error: the two signals are engineered independently and carry no relationship to each other whatsoever. A clip can present the visible mark clearly while also carrying SynthID, or show no visible mark while still carrying it, since the two systems were never designed to correlate. Precision here isn't pedantic, it's the difference between a defensible conclusion and an unfounded one that happens to look confident.

The visible mark exists specifically to be seen, which is exactly why a systematic scan can confirm or rule it out with reasonable confidence. SynthID exists specifically to resist perception, embedded in the statistical structure of the frame data rather than the visible content, which is why inspection alone, however thorough or systematic, cannot settle the question without the issuer's own verification tool. That asymmetry is a designed property of the two systems, not an inconsistency in how this page treats them.

No. They are independent by design. A positive or negative result on the visible mark carries zero evidentiary weight regarding SynthID's presence, whatever it happens to show. Only Google's own detector can confirm SynthID authoritatively, and this page states that limitation directly rather than implying a broader capability than what's actually offered here.

Recurring cases share a need for a precisely scoped conclusion.

Publication verification

Confirming a clip's visible-mark status against a documented standard before it goes to publication.

Moderation policy

Treating the visible mark and SynthID as two distinct, independently verified inputs rather than one.

Provenance research

Understanding precisely which signal a given tool can and cannot independently confirm on its own.