Face reenactment is a computer vision technique that transfers the facial expressions, head position, and lip movements from a 'driver' video to a 'target' face, creating a video that appears to show the target person performing the driver's actions. This technology can create fake intimate videos of real people without their consent. Unlike static deepfake generation, face reenactment produces temporally consistent video output. All AI-generated intimate video content, including face reenactment, has significant ethical concerns.

Key facts about this term

  1. Recognize face reenactment output as concerning content Any video that realistically depicts an identifiable person in intimate situations without consent raises significant ethical issues regardless of the generation technique.
  2. Use ScanErase to identify reenactment video content The biometric scan identifies face reenactment video content through frame-level facial geometry analysis.
  3. Consider reporting to appropriate authorities for investigation Face reenactment intimate video creation can involve serious ethical and legal implications.

Frequently asked questions

How is face reenactment different from a standard deepfake?

Standard deepfakes often replace a static face in existing content. Face reenactment specifically transfers facial movements and expressions from a live or recorded source to a target face, enabling fake videos of the target person 'performing' arbitrary actions.

Can I tell if a video was created with face reenactment technology?

Face reenactment creates characteristic artifacts including blinking inconsistencies, edge artifacts at the face boundary, and lighting mismatches. Detection tools can identify these artifacts in most cases.

Related glossary terms