Digital replica of an actor built via motion capture or 3D modeling for VFX-heavy sequences. Required for impossible stunts no human survives.
Digital Body
When you need to drop an actor from a bridge, have them burst into flames, or press them through a glass dome — that's where the digital body comes into play. This is the digital double that replaces the real person on screen, without insurance, safety, or logistics collapsing. Unlike a classic stunt double, who still has flesh and blood, the digital body is a pure data package: a 3D geometry with a skeletal structure (rigging), shader information, and — if necessary — motion capture movement data from the original actor.
Production typically proceeds along two parallel paths: On one hand, the direct 3D scanning process — the actor sits in a Light Stage or scanner setup, and a photogrammetry team captures their face and body in high-resolution geometry. On the other hand, motion capture data from set sessions (or from stunt doubles) is loaded into the digital body, bringing it to 'life'. In post-production, you typically need four to eight weeks until a production-ready version is available — depending on how detailed the skin, hair, and clothing need to be. For extreme close-ups (during a fall), the effort is brutal; for medium to wide shots, less texture resolution is sufficient.
The most common use cases: Superhero films, where 300-meter jumps or explosions are involved; science fiction, where bodies are digitally deformed or torn into other dimensions; historical action, where mass battle scenes with a hundred digital doubles run simultaneously. The crucial point is real-time integration — the digital body must match under real-time lighting (volumetric lighting, reflections), otherwise, it's immediately obvious that the human and the digital are separate. That's why VFX supervisors and DPs work closely together: the lighting must be precisely defined for both realities — camera and CGI element.
Typical source of error: The digital doppelgänger moves too perfectly, too smoothly. Real humans have irregularities — tremors, weight shifts, micro-movements. Without these details, the digital body appears like a puppet. Professional VFX teams therefore deliberately inject small inaccuracies (noise) into the motion capture to avoid the uncanny valley. Another practical point: The storage requirement is enormous. A highly detailed digital body animation for a two-minute sequence can easily mean 50–200 GB of raw data — rendering farm resources must be planned accordingly.