Detection capability
Deepfake Detection
Facial structure, expression, and motion are analysed over time alongside texture, boundary, reflection, and lighting cues to identify signs of manipulated or synthetic video.
FaceOff’s Deepfake Detection capability analyses a video by looking at facial structure, expressions, and movement over time. It also considers visible texture, boundaries, reflections, and lighting cues when assessing signs of manipulated or synthetic content. Instead of relying only on individual pixels or a single frame, the capability looks at how facial features behave together over time. Rhythm, symmetry, and overall coherence are considered as part of the analysis, helping identify signs that a face or video may have been digitally manipulated. This provides a broader way to assess video authenticity when the integrity of digital content needs to be checked.
Deepfake Detection is one of eight signals used by the Trust Factor Engine. It is considered alongside the other seven signals rather than being treated as the only basis for the overall result. The engine considers multiple signals together when assessing whether the video, voice, or face being evaluated can be trusted. A single signal does not independently determine the final verdict. How the weighting works
Next
See the whole engine
Capabilities are components. The platform page shows how the eight combine into one score.
See it score a real video
Bring your own footage. We will run it through the Adaptive Cognito Engine and walk you through the findings.