There are two honest ways to move a 3D body: pose it yourself frame by frame, or record a performer and map the data onto a rig. Both work. Both fail, and they fail in opposite directions, which is the useful part. If you can name the failure you are looking at, you know which route produced it and what will fix it.

What hand-keying buys you

Keyed animation gives you exact control over emphasis. You decide which beat lands, how long a look holds, and where the audience should be looking. Nothing arrives by accident. For a scene built around two fictional adult women reading each other, that control is the whole point, because the meaning sits in when a hand pauses rather than in the fact that it moved.

The cost is time and skill. A few seconds of convincing two-character contact can take days, and the failure mode is stiffness: poses that are individually good but connected by curves that nobody in the room believes. Keyed work also tends toward symmetry and repetition, because a tired animator reuses the last thing that worked.

What capture buys you

Motion capture gives plausibility almost for free. Micro-adjustments, breathing, the constant small corrections a standing person makes: all of it comes in the data without anyone deciding on it. That texture is very hard to key by hand and it is the reason capture keeps winning on schedule-driven productions.

The cost is cleanup and loss of intent. Marker noise, foot slide, and limbs passing through each other are normal. Worse, the data flattens emphasis. A performer's clear choice can turn mushy after retargeting onto a rig with different proportions, and you end up keying over the top of it anyway, which is the expensive middle you were trying to avoid.

Where close two-body scenes break

Intimacy is contact, and contact is where capture struggles most. Two performers touching produce occluded markers, and occlusion produces guesses. The result is limbs that intersect, hands that sink into a shoulder, and pressure that never appears because a capture volume records position rather than force. Skin does not deform on its own in the data.

Hand-keyed contact has the opposite problem: it looks clean but weightless. The hand sits on the shoulder without pressing into it. Both routes need the same repair, which is a deformation pass and a set of pinned contacts. Plan for that pass rather than treating it as a surprise at the end of the schedule.

Choosing a route for a real schedule

A rough rule: capture for locomotion and for anything long and continuous, keys for anything the audience is meant to read as a decision. Most small teams blend the two on the same shot, with captured body and keyed hands and face. That is not a compromise; it is the normal shape of the work.

Picking between keyframes and capture in practice

One more thing decides the route more often than craft does: how many shots you need. A single hero shot rewards hand-keying, because the whole budget can sit on it. Forty shots of two people moving around a flat do not, and trying to key them all is how a small team ends up with four finished scenes and no film. Count the shots before you argue about method, then divide your available weeks by that number and see whether the answer is a day per shot or an hour.

Decide before you build the rig, because retargeting requirements change how a rig is made. Then write down which shots are keyed and which are captured, and stick to it, because switching halfway is where budgets die. Everything else follows from that one early call.

Adult work adds one condition that is not negotiable: performers, if you use them, are consenting adults doing a job, and characters are fictional and unmistakably adult. Our reader agreement spells out the line we hold, and the editorial principles explain why we publish craft notes rather than files. For weight problems that survive both routes, start with the piece on timing and settle.