Intel RealSense T265

Demo and values

Tracking camera from Intel with two global-shutter fisheye cameras and a Bosch BMI055 IMU, solved here on its own IMU from one recording of our VI-Rig.

The 3D scene MVIS recovered from recording 1.

Reprojection median0.21 px

cam0–cam1 baseline63.8 mm

cam0 td−7.99 ms

cam0 lever arm15.0 mm

One recording, with no independent reference on this page. These values show what MVIS solved, not how close they are to the truth.

Calibration process

The solved path of the rig around the board. Press Play or drag the bar to follow it, and use the buttons to change the view.

▶︎ Playcam0cam1Loading

Board cornersRig path, darker is laterStartCurrent position

Solved rig

The solved rig in 3D, to scale. Tap a sensor for its values. Full screen lets you rotate and zoom. Grid squares are 20 mm.

3DFrontTopSide+−

To scale, from above and the front

From above

From the front

Dashed: past the fit, drawn to the corners' reach, so the lens sees at least this much.

Camera	Lever x, y, z (mm)	Distance (mm)	log(R) (rad)	td (ms)	Field of view
cam0	+11.7, +4.3, +8.4	15.0	[+0.017, −0.000, −3.127]	−7.99	past the fit
cam1	−52.1, +4.1, +7.5	52.8	[+0.027, −0.005, −3.126]	−10.65	past the fit

Solved by MVIS. Lever arms are in the IMU frame. Rotation is log(R), camera in the IMU frame.

Specifications

Published by the maker
Cameras	2 × OV9282, 848×800 at 30 Hz
Shutter	Global
Lens	Fisheye, about 163°
Measured in these recordings
IMU	Bosch BMI055, gyro at 200 Hz, accel at 62.5 Hz
Calibrated field of view	cam0 past the fit · cam1 past the fit

Dataset

Recording	VI-Rig, recording 1, 80 s window chosen by motion
Format	ROS bag
Board	AprilGrid 6×6, tag 88 mm, spacing 0.3, t36h11
Configs and results	Calibench community, intel-t265/ (access by automatic approval)
Licence	CC BY-NC 4.0, for the data and for this page's clip and figures; commercial use by request
Hosted	Calibench, virig-4imus-3cams/ (access by automatic approval)

AprilGrid 6 × 6, 660 × 660 mm. Standard pattern, to scale

Detailed results

Lens Calibrated field of view

Calibrated field of view, horizontal × vertical. cam0: past the fit, corners reach 70° from the axis · cam1: past the fit, corners reach 69° from the axis

cam0 cam1

Read from the calibrated lens fit at the image edges, through the principal point. ~ marks an axis past the range the corners reached, where the fit is extrapolated. Past the fit means an edge lies beyond where the fit holds, so no value is given.

IMU Gyro-to-accel offset −6.59 ms

Scale as deviation from 1. Model: calyx. Gyro-to-accel time offset t_ga: −6.59 ± 0.43 ms.

Gyro scale x−0.45 %

Gyro scale y−0.29 %

Gyro scale z+0.48 %

Accel scale x+2.08 %

Accel scale y+2.18 %

Accel scale z+1.31 %

−8 %0+8 %

Timing cam0 td −7.99 ms

Offsets to scale, with t_imu = t_cam + td.

Full results Every solved parameter

Rotation is log(R) in rad, camera in the IMU frame.

Parameter	MVIS
cam0
fx px	286.44
fy px	285.86
cx px	415.36
cy px	403.21
k1	−0.00258
k2	0.03660
k3	−0.03556
k4	0.00595
Lever x mm	+11.69
Lever y mm	+4.33
Lever z mm	+8.41
log(R) rad	[+0.0173, −0.0003, −3.1271]
td ms	−7.992
Reprojection px	0.415
cam1
fx px	286.30
fy px	285.56
cx px	414.91
cy px	397.49
k1	−0.00356
k2	0.03876
k3	−0.03782
k4	0.00702
Lever x mm	−52.15
Lever y mm	+4.07
Lever z mm	+7.52
log(R) rad	[+0.0268, −0.0045, −3.1260]
td ms	−10.649
Reprojection px	0.406
IMU
t_ga ms	−6.59 ± 0.43

Limits

• One recording, with no independent reference on this page. These values show what MVIS solved, not how close they are to the truth.
• The T265 gyro-to-accel offset isn't a pure time shift. We measured about −6.8 ms at 2–4 Hz and −1.2 ms at 30–40 Hz, so a single t_ga is a compromise that depends on the motion.

How it was run

• Fisheye stereo plus the T265's own IMU, camera intrinsics free, t_ga on. Batch of 2026-11-03.