Guide
What markerless capture is, how to shoot for it, and how to tell a usable take from a convincing one.
A traditional stage measures markers with dedicated cameras. Markerless capture infers the same skeleton from ordinary images: it finds the person, estimates where each joint sits in three dimensions, and produces joint rotations over time. The output is the same kind of animation data — the measurement is inferred rather than observed.
Keep the whole body in frame, keep it in focus, and light it so limbs separate from the background. A steady camera helps more than a high resolution does. Loose clothing, heavy motion blur and limbs leaving frame are the three things that cost you the most accuracy.
Look at feet first: sliding and floating are the clearest signs the reconstruction is guessing. Then look at the hips through a turn, and at any joint whose confidence is low. A take that reads well at playback speed can still be wrong frame by frame, which is why Motion Studio shows you both.
Cleanup trades detail for stability. Light smoothing removes noise without touching the performance; maximum smoothing will also remove the small accelerations that make a movement look human. Start light, and stop as soon as the problem you were fixing is gone.
Motion is captured on a human's proportions and applied to a rig with its own. Differences in limb length, reference pose and axis convention are where retargeting goes wrong. Check the reference pose first, then the scale, then the root orientation — in that order.
A calibrated optical stage remains the most accurate way to capture a performance, and some work will always need it. Video capture wins where a stage is impractical: existing footage, remote performers, quick iterations, volume, and reference for generative video.
Turn real performance into editable motion for animation, AI video, games, virtual production, robotics and more.
No suit. No markers. Just motion.
