Calibrate a whole rig at once

From one recording to a full sensor model.

MVIS is the calibration library behind the Open MVIS benchmark. It solves camera intrinsics and distortion, extrinsics, time offsets, rolling-shutter readout, board poses, camera–IMU transforms, IMU intrinsics and biases in a single GTSAM factor graph.

ROS1 and ROS2 bags are read through rosbags, alongside EuRoC/ASL folders, image folders and video, on a machine that has never had ROS installed.

# cameras only
$ mvis cam -c camera.yaml

# cameras + IMUs, jointly
$ mvis cam-imu -c cam_imu.yaml

# apply the result and recalibrate; PASS / FAIL
$ mvis roundtrip -c cam_imu.yaml

One self-contained HTML report per run.

What the graph estimates

The state, not a sequence of stages.

Quantity	Symbol	Per	Unit	Notes
Intrinsics + distortion	f, c, k	camera	px	Six models, mixable within one rig
Camera–IMU extrinsics	T_CtoI	camera	m · log(R) rad	Every camera to the base IMU
Camera time offset	t_d	camera	ms	Recovers USB lag of −124 ms on the ELP stereo
Rolling-shutter readout	t_r	camera	ms	Each corner projected at its own row time
IMU–IMU extrinsics	T_IstoI	aux IMU	m · log(R) rad	Plus a per-IMU time offset
IMU intrinsics	D_w, D_a	IMU	—	kalibr, rpng or calyx convention
Gyro–accel offset	t_ga	IMU	ms	The T265 reads −6.2 ms, stable to ~0.2 ms across bags
Biases	b_g, b_a	IMU	rad/s · m/s²	Along the whole trajectory
Board pose + trajectory	T_BinW, T_ItoW(t)	recording	m · log(R) rad	Chessboard, AprilGrid or ArUco targets

Camera models

Six projection models.

• radtanPinhole, radial–tangential
• equidistantKannala–Brandt KB4 fisheye
• fovField-of-view model
• double_sphereWide-angle fisheye
• eucmExtended unified camera
• omni_radtanOmnidirectional

Validated

Checked against published calibrations.

228

regression runs across 65 dataset families, with zero failures

<0.6%

focal-length difference from the published TUM-VI and Intel T265 calibrations

≤2mm

lever arm vs Kalibr on TUM-VI raw IMU bags, with ≤0.15 mm stereo baseline and ≤1 ms time offset

VI-Rig: 4 IMUs and 5 cameras, one of the rigs in the Calibench roster.

It checks its own work

A low residual is not the test.

mvis roundtrip applies a calibration to its own data and recalibrates. PASS needs the IMU intrinsics to collapse to identity and the stereo baseline to hold. See Round-trip validation.

Inputs

ROS1 and ROS2 bags, EuRoC/ASL, image folders, video

Targets

Chessboard, AprilGrid with Kalibr 2-bit border, ArUco

Interop

Reads and writes Kalibr camchain and camchain-imucam

Export

mvis apply-calib re-records a raw bag with the calibration applied

Cite

If you use the toolbox or the benchmark.

Benchmark

@misc{openmvis,
  title  = {{Open MVIS}: A Benchmark for Multi-Visual-Inertial Sensor Calibration},
  author = {{Open MVIS Contributors}},
  year   = {2026},
  url    = {https://openmvis.com}
}

Method

@article{yang2024mvis,
  title   = {Multi-visual-inertial system: Analysis, calibration, and estimation},
  author  = {Yang, Yulin and Geneva, Patrick and Huang, Guoquan},
  journal = {The International Journal of Robotics Research},
  year    = {2024},
  doi     = {10.1177/02783649241245726}
}

Cite leaderboard results as Open MVIS leaderboard, vX.Y.