Tools · Motion capture

Motion Capture Rig Planner

Plan a multi-camera markerless motion capture setup before you carry a single tripod into the room. Part 1 places the cameras so every one sees the whole body; part 2 checks that the frame rate, exposure and network can deliver every frame. Built from the planning tools inside MoCap6, and set up for six Basler acA720-290gc cameras by default.

Camera spacing, height and tiltFull-body coverage mapFrame rate and bandwidth checkPrintable setup checklist

1 · Camera layout

Floor plan, top view

Side view

Saved rooms

Placement sheet

2 · Capture settings

GigE Vision cameras stream over UDP with no congestion control. Ask for more than the wire can carry and nothing errors: frames simply arrive with holes in them, and you find out later when the 3D is noisy. This check is the same model MoCap6 runs before it allows a recording.

Pre-flight check

How it works

How the layout is chosen · expand
  • The walking direction runs along X, the room's length. The ring is turned 30° off the walking line so no camera stands in the subject's path.
  • For points across the capture area, the planner checks whether each camera sees a box around the body from floor to head (plus 0.15 m) inside 92% of its field of view, at the best tilt between −10° and 30°. The dots on the plan show how many cameras see the whole body at each point.
  • "Reset to the smallest layout" searches for the smallest ring, or for an oval the smallest that fits the room, where every point is covered by the number of cameras you require.
  • A circle is the most even layout and suits turning, squats and jumps. An oval suits a straight walkway: it stretches along the walk and narrows across it, so it fits long, thin rooms, but usually needs "at least 4" cameras rather than all of them.
  • Calibration needs a board the cameras can recognise at these distances: with cameras 3–4 m away, a one-page board is too small. MoCap6 uses a 5 × 4 ChArUco board with 175 mm squares, recognisable to about 6.8 m.
How the capture check works · expand
  • Bandwidth on the wire: each frame is split into packets. Every packet carries an 8-byte GVSP header and is wrapped in 46 bytes of Ethernet/IP/UDP and a 20-byte preamble and gap, plus a leader and trailer packet per frame. That is what actually goes on the wire.
  • 85% rule: above 85% of a link there is no room for packet resends, so a single lost packet becomes a lost frame. The fastest safe frame rate keeps every camera link and the shared link to the PC under 85%.
  • Exposure: it must be shorter than the frame period, or cameras skip frames and drift out of sync. Exposure also sets motion blur: a limb moving at 5 m/s travels 10 mm during a 2,000 µs exposure.
  • Jumbo frames: at 1,500-byte packets each frame needs about six times as many packets, and an untuned network adapter starts dropping above about 400,000 packets per second.
  • Disk: the encoder has to accept the raw sensor rate even though the saved video is compressed far smaller.
Using this in classWeek 12 of the Biomechanics Laboratory uses motion capture. Have students plan the rig for your room here first, then compare their plan with the real setup. For the full system, see MoCap6.