Quick answer: Motion capture, or mocap, is the process of recording a real person’s or object’s movement and translating it into digital data that can drive an animated character or feed a biomechanical analysis. Traditional systems use physical markers or an instrumented suit to pin down exact positions in space; newer markerless systems estimate the same movement straight from ordinary video. It is the technology behind digital characters like Gollum and the Na’vi, much of the animation in modern video games, and a growing tool for analyzing an athlete’s technique.
This entry covers how the different flavors of motion capture work, where the technology shows up in film, games, and sports, how it relates to pose estimation, and where each approach runs into trouble.
What Is Motion Capture?
Motion capture records movement, not appearance. A camera or sensor rig tracks where key points on a body are over time, and that stream of positions becomes an animation skeleton that a digital character, a robot, or an analysis tool can use.
There are three broad approaches, each answering the same question with different hardware. Optical marker systems track physical markers with an array of cameras, while inertial suits track the wearer’s own body using onboard motion sensors instead. Markerless systems skip physical markers altogether and estimate a person’s pose directly from ordinary video.
All three answer the same underlying question, where are this body’s joints right now, but they trade off cost, portability, and precision differently enough that a film production, a game studio, and a sports lab tend to reach for different ones.
Whichever way the positions get captured, the output is usually a skeleton of joints and bones rather than a video. That skeleton can then be retargeted onto a completely different digital character rig, which is how one actor’s performance ends up driving a nine-foot-tall alien or a cartoon animal instead of a digital copy of the actor.
How Does Motion Capture Work?
Optical marker systems are the traditional standard for high-end film and game work. Reflective markers, or in active systems small LEDs, are attached at key points on a performer’s body and face, and a ring of specialized cameras triangulates each marker’s position, typically well over a hundred times a second and accurate to a fraction of a millimeter.
Inertial suits take a different approach, packing each body segment with a small sensor combining a gyroscope, an accelerometer, and a magnetometer. Because the suit measures its own motion rather than relying on external cameras, it needs no fixed capture volume and works outdoors or on a cluttered set, though small errors in each sensor reading can accumulate into positional drift over a long take.
Markerless motion capture drops physical hardware from the performer entirely. Software analyzes ordinary video, often from more than one camera angle, and estimates where each joint is using the same kind of neural network that powers pose estimation. Running several synchronized cameras lets the software recover that movement in three dimensions instead of flattening it into the image plane.
Where Is Motion Capture Used?
Film has used the technology for decades. Andy Serkis’s performance as Gollum in The Lord of the Rings, realized by Weta Digital starting with The Two Towers in 2002, combined motion capture data with traditional keyframe animation to build a character audiences read as fully alive. Weta pushed the approach further for Avatar in 2009, adding head-mounted cameras that captured each actor’s facial performance alongside full-body motion capture on a stage built specifically for the film.
Games have relied on the same pipeline since the 1990s, when early 3D fighting and action titles started using recorded human movement instead of hand-animated cycles to make characters move convincingly. Consumer motion sensing reached living rooms with Microsoft’s Kinect, which used markerless depth sensing to turn a player’s own body into a game controller.
Live performance is a newer use. A VTuber broadcast or a virtual concert act drives a digital avatar from a performer’s tracked body and face as it happens, which is the same pipeline with the rendering moved from post-production into the live stream.
Mocap has also become achievable outside a big-studio budget. Ninja Theory shot the lead performance for Hellblade: Senua’s Sacrifice with a small team in their own boardroom, using a marker suit and a lightweight helmet camera rig instead of a full commercial stage, and capturing the actress’s voice and physical performance together in the same take rather than recording them separately.
Sports is where markerless capture has opened doors that marker-based systems could not. A marker rig is impractical on a real field or rink, so coaches and researchers increasingly pull joint positions straight from broadcast or training footage instead. Figure skater Yuzuru Hanyu wore a sensor rig to study his own jump technique for a graduate thesis on skating biomechanics.
Clinics use the same markerless approach for gait analysis, screening for movement disorders such as Parkinson’s disease without attaching anything to the patient.
Motion Capture vs Pose Estimation: What’s the Difference?
The two terms describe overlapping but distinct things. Pose estimation is the underlying computer vision task, working out where a body’s joints are in a single image or video frame. Motion capture is the broader production pipeline built around recording movement over time, historically with physical markers or suits, and increasingly with pose estimation doing the core detection work frame by frame.
Put another way, every markerless motion capture system relies on pose estimation, but not every use of pose estimation is trying to produce motion capture data. A retail app counting how many people walk past a window is running pose estimation without capturing anyone’s motion for later use.
This same distinction separates motion capture from face tracking, which is really pose estimation applied to a face instead of a whole body; the two often run side by side on a single performance capture shoot, one rig covering the body and another covering the face.
What Are the Limits of Motion Capture?
Optical marker systems need a dedicated capture volume, specialized cameras, and markers placed correctly on every performer, which puts the highest-end setups out of reach for smaller productions and rules out any location without controlled lighting and space.
Inertial suits solve the space problem but not accuracy: drift accumulates over a long take, and the sensors alone cannot tell where a foot is relative to the actual floor, so many productions periodically recalibrate or blend in other tracking to correct for it.
Markerless capture inherits computer vision’s usual weaknesses. Fast motion can blur past a frame before the software can resolve it, one performer’s limbs can occlude another’s in a crowded scene, and poor lighting degrades every stage of the pipeline. None of the three approaches is strictly better than the others, which is why productions still mix them depending on the shot.