How is motion capture used in games?
To record real human movement and apply it to a digital character — and modern productions capture body, face and voice, frequently at the same time, which is why game acting has become a distinct profession.
How body capture works. A performer wears a suit with reflective markers. An array of infrared cameras tracks each marker in three dimensions, producing the motion of a skeleton over time. That skeletal data is then retargeted onto the character rig, which rarely has human proportions.
Markerless systems using computer vision and machine learning are increasingly viable, and inertial suits with accelerometers and gyroscopes work outside a camera volume, at the cost of drift.
Facial capture is a separate discipline — a head-mounted camera records the performer's face, with markers or markerless tracking, driving a facial rig. Facial performance is where uncanny results most often appear, because humans are extraordinarily sensitive to small errors in expression, and to eyes in particular.
Performance capture is the term for capturing body, face and voice simultaneously, so the performance is unified rather than assembled from separately recorded parts. The difference is easily perceptible.
Why raw capture is never enough. The data always needs cleaning: occluded markers, jitter, foot sliding where the digital ground does not match the stage. Animators then adjust it, exaggerate where needed, and build transitions — because a game character must blend between hundreds of animations in response to input, which film never requires.
This is the fundamental difference from film. Film plays a fixed sequence. A game must respond, so animation must be interruptible, blendable and responsive, and pure captured realism can feel unresponsive if it takes too long to turn or stop.
Where motion capture is not used: highly stylised games, where hand-keyed animation gives control and exaggeration capture cannot; impossible movements; and creatures with no human referent, though performers still act these out.
Voice acting may be recorded separately in a booth, or as part of performance capture. Industrial disputes over AI voice replication have made consent and compensation for recorded performance a live issue.