Stereo Camera Modules: Synchronization, Calibration, and Depth Output Explained
Understand synchronization, calibration, and depth output as separate requirements when evaluating a stereo camera module.
Last updated: Nov 6, 2026
A dual-lens camera module is not automatically a complete depth camera. Before ordering, separate three questions: when are the two images captured, how is the camera pair calibrated, and what output does the system actually deliver?
Synchronization concerns timing. Calibration describes imaging geometry. Depth output requires a processing path that uses suitable images and calibration data. Confirm each requirement for the exact module, software, and configuration you intend to use.
Three requirements to confirm separately
Requirement
What it addresses
Evidence to request
Synchronization
The timing relationship between the two captures
Supported mechanism, timing definition, tolerance, and applicable capture mode
Calibration
Each camera's imaging parameters and the geometry between the cameras
Applicable calibration data, configuration details, and instructions for using it
Depth output
The processing that produces distance information from the images
Output format and units, required processing hardware or software, and documented operating conditions
These requirements are connected, but approval of one does not establish the others. A synchronized pair still needs an appropriate calibration and processing workflow if your application requires stereo depth.
Synchronization: check capture timing, not only frame rate
Two streams running at the same frame rate do not, by that fact alone, establish matched exposure timing. Ask what “synchronized” means in the proposed system: exposure start, a trigger event, frame timestamps, or software pairing of delivered images.
Also ask where timestamps are generated. In its MIPI stereo-rig documentation, StereoLabs distinguishes its hardware synchronization from software timestamps assigned when the driver receives frames. That is an example of why a timestamp comparison needs a documented interpretation, rather than a timing guarantee for other cameras.
Request the following details:
How the captures are coordinated, and whether additional hardware or control signals are required.
Which event the timing specification describes and what tolerance is documented.
Which resolution, frame rate, and exposure settings the specification covers.
How the application identifies corresponding frames and detects missing frames.
Even an external trigger requires a timing explanation. Basler's acquisition timing documentation separates trigger detection, exposure start delay, sensor readout, and transmission timing. Use the documentation for your chosen camera to understand its behavior; do not transfer another model's delay figures into your design.
Calibration: identify the data and its configuration
For a stereo workflow, calibration commonly includes the intrinsic parameters and lens distortion of each camera, plus the relative rotation and translation between them. A physical baseline measurement alone does not supply that complete description. StereoLabs' calibration documentation illustrates the distinction between per-camera parameters and the stereo transform.
Ask whether calibration data is supplied, which physical unit or camera pair it belongs to, and which image configuration it describes. If data is not supplied, establish who will perform calibration and how the results will be checked before relying on the pair for measurements.
Keep calibration and rectification distinct. Calibration estimates parameters; rectification uses camera geometry to transform the image pair for stereo processing. OpenCV's calibration and reconstruction reference documents separate operations for stereo calibration, rectification, and reconstruction.
Record the resolution and any crop, scaling, or image transformation applied before processing. Calibration parameters must correspond to the images being used. If the lenses, focus, or relative camera mounting change, check whether the existing calibration remains applicable before continuing with the old files.
Depth output: establish where the processing happens
Ask the supplier to name the output explicitly. Two camera images, a rectified pair, a disparity image, and a depth map are different deliverables.
Stereo matching estimates correspondences between the views. Disparity describes their positional difference; converting that information into depth uses the camera geometry. OpenCV's depth-from-stereo tutorial explains the relationship and demonstrates disparity computation. A colored visualization alone does not specify a usable distance interface.
For a system advertised as providing depth, confirm:
Whether processing runs on the module, an additional processor, or the host.
Which driver, SDK, or application is required and which versions are supported.
The output format, distance units, coordinate convention, and handling of invalid values.
The documented range and accuracy conditions, including the relevant image mode and scene.
How your application obtains images and depth together, if it requires both.
Request evidence for your intended working distance, lighting, and subject conditions. Do not infer a depth specification from the module's resolution, lens count, or stated baseline alone.
Decide what your application needs
These hypothetical examples describe requirements, not tested configurations or customer projects.
You need two views for recording
If the application only records two views, specify the stream formats, simultaneous capture requirements, and any timing relationship you need. A ready-made depth map may not be part of the requirement. Confirm that the host can receive the required streams before selecting a module.
You need distance measurements of moving objects
Specify the required distance output, working range, and acceptable error under your scene conditions. Then evaluate timing, calibration, and processing as a complete path. Ask for validation that addresses movement as well as a static demonstration; a clear pair of pictures does not establish measurement performance.
Request evidence before approving a purchase
Keep a short record for the proposed module:
Identity: Exact model, hardware revision, and camera-pair identification where relevant.
Capture: Required modes, synchronization mechanism, timing specification, and frame-pairing method.
Geometry: Calibration availability, applicable image configuration, and responsibility for calibration or revalidation.
Processing: Required host, software versions, processing location, and delivered output.
Validation: Application conditions and the checks needed before purchase approval.
Included items: Confirmed files, cables, adapters, software, and support scope.
Mark anything unconfirmed as an open question. If evaluation hardware is needed, ask about availability and terms rather than assuming a sample or evaluation kit is included.
When reviewing our MIPI dual-lens camera module, use its listed specifications and confirm any missing information before relying on it in a design. This guide does not establish synchronization, calibration files, depth output, or platform compatibility for that product.