Motion capture records a body, and the character is built afterwards

This is less a set of instructions about performance capture than an argument, and it is worth saying so at the start.

The volume is not a set

A capture stage is a marked area surrounded by cameras that track reflective markers or optical patterns, and it contains almost nothing a performer can use. Sets are represented by scaffolding, boxes and tape lines, so an actor climbing a cliff is often climbing a step unit with an eyeline marked on a stand.

Costume is replaced by a fitted suit, which removes the weight, restriction and silhouette that normally do a great deal of an actor’s work. Everything an environment would ordinarily supply has to be imagined, and that imaginative load is the reason the technique is harder than it looks. Some productions shoot capture on location or alongside live action, which restores context at the cost of technical complexity.

Data is not a performance yet

The system records the position of joints over time, which is a skeleton in motion rather than anything anyone could watch. That skeleton is then retargeted onto a character whose limb lengths, mass and proportions may be entirely different from the performer who supplied it. Retargeting introduces errors wherever proportions diverge, so a gesture that read as delicate on a human frame can read as clumsy on a larger creature.

In the edit, fixing those errors is animation work, and it involves judgement about what the performance meant rather than simple correction. This is the step that most reporting on the subject skips, and it is where a substantial part of the final character is decided.

The face is a separate problem entirely

Facial capture typically uses a head-mounted camera and its own solving pipeline, because the face moves in surfaces rather than in joints. Skin slides over muscle and bone in ways a skeleton model does not describe, so the underlying rig is built from controls that approximate anatomy.

Watched twice, transferring an expression to a face with a different structure changes its meaning, and a smile can become a grimace through geometry alone. Eyes are the hardest element, since audiences read gaze direction with great precision and tolerate almost no error in it. Much of the perceived quality of a digital character rests on a few square centimetres around the eyes and mouth.

Animators are co-authors, not technicians

Between the captured data and the finished shot sits a team making decisions about timing, weight, contact with the ground and exaggeration. Those decisions are the same ones hand animation has always made, and they can strengthen or flatten what the performer originally did.

The industry argument about credit follows from this, because two crafts contributed to one visible result and the boundary between them is genuinely blurred. Describing the process as an actor simply appearing in another shape understates the animation work, and describing it as pure animation understates the performance.

The accurate description is collaborative authorship with an unusually long gap between the two contributions.

Integration with live action is where it shows

A digital character sharing a frame with a photographed one must match lighting, lens distortion, grain and the exact perspective of the plate. Contact is the usual failure point, since a hand resting on a shoulder or a foot pressing into mud requires interaction the plate never contained. Eyelines are agreed on set with markers and tennis balls, and any drift between what the actor looked at and where the character ends up is visible.

In practice, shadows cast by the digital element onto real surfaces are frequently the detail that convinces or fails to convince an audience. This is why a creature shot in a plain environment is far cheaper than the same creature walking through a real crowded street.

What the technique changed about casting and rehearsal

Because physical resemblance is no longer a constraint, performers can be cast for movement quality and voice rather than for how closely they look the part. Rehearsal becomes more important, not less, since the environment offers no support and the physical logic has to be established before capture begins.

What the cut is doing: multiple performers sometimes contribute to one character, with body, face and voice recorded separately and assembled later. That flexibility is genuinely useful and it complicates any simple account of who gave the performance. The technology did not remove acting from the process; it moved it further away from the finished image.

The takeaway

The capture stage records movement. The character is authored in the months that follow.

Craft is the part that keeps working after the surprise has gone.