Embedded Camera Integration Checklist: Interfaces, Drivers, Carrier Boards and Cables

Embedded camera integration should be evaluated as a complete system rather than as a camera-only decision. The camera, interface, processor, carrier board, connector, cabling, driver and software environment must support the required configuration.

For embedded vision, the integration path also changes by interface. MIPI CSI-2 uses a direct camera-to-processor architecture, while GMSL2 and FPD-Link III use a serializer/deserializer architecture for remote camera placement. Although GMSL2 and FPD-Link III solve a similar system-level problem, their hardware and software are not interchangeable.

This checklist covers the main points to verify before selecting and validating an embedded camera configuration.

Already know your processor, required camera distance and interface? Compare the available MIPI CSI-2 cameras, GMSL2 cameras and FPD-Link III cameras, or discuss your embedded vision configuration if you need help matching the camera, serializer/deserializer hardware, carrier board and software environment

The distance between the camera and the processing platform is a key consideration when selecting a camera interface:

Camera close to the processor: start with the 36S Series MIPI CSI-2 camera modules.
Remote camera in a rugged IP67 environment: evaluate the Acuva GMSL2 camera series.
Remote board-level or IP67 camera: evaluate the 36C Series FPD-Link III cameras.

Selecting the appropriate camera also requires consideration of sensor characteristics, resolution, shutter type, frame rate, processing platform, and software compatibility.

A carrier board provides the physical and electrical interface between the compute module and embedded cameras, including connectors, power, signal routing and system interfaces.

Its configuration therefore depends on the processor, camera interface, camera count, cable distance and software requirements.

For MIPI CSI-2, the path can remain relatively direct. For GMSL2 and FPD-Link III, the complete serializer/deserializer path must also be compatible with the camera and processing platform.

Step 1 — Define the imaging and system requirements

Start with the application. Define the required resolution, frame rate, shutter type, field of view, camera count and synchronization requirements.

Also establish system constraints such as camera position, required cable distance, available installation space and environmental conditions.

These requirements already help narrow the camera family:

36S MIPI CSI-2: compact integration with the camera positioned near the processor.
Acuva GMSL2: rugged remote-camera configurations using a serializer/deserializer architecture.
36C FPD-Link III: remote board-level or IP67 configurations using FPD-Link III serializer/deserializer hardware.

Step 2 — Define the processing platform and interface

Identify the processor or SoM early in the project and determine which camera architecture fits the system.

For compact camera-to-processor integration, MIPI CSI-2 cameras are suitable. For remote camera placement, GMSL2 or FPD-Link III may provide the required serialized architecture.

The processor, interface and camera should be evaluated together because they determine the available carrier-board, serializer/deserializer and software paths.

For example, projects based on NVIDIA Jetson Orin or NXP i.MX should be evaluated against the exact camera interface, camera count and carrier-board configuration rather than processor support alone.

Step 3 — Verify the complete hardware path

Confirm the complete connection from the camera to the processor.

For MIPI CSI-2, this includes the connector, pin assignment, lane configuration and compatible CSI-2 input.

For GMSL2 and FPD-Link III, verify the complete serialized path:

Camera → serializer → cable → compatible deserializer → carrier board → processor

Do not verify only the connector or cable. The serializer and deserializer must be compatible with each other and with the selected camera architecture.

Mechanical requirements such as camera dimensions, connector position, lens mount and cable routing should be checked at the same stage.

Step 4 — Confirm driver and software support

Verify software support for the exact camera and processing platform before finalizing the hardware.

Confirm the driver, BSP, operating-system version, required image formats, camera controls and acquisition environment.

Electrical compatibility alone does not establish that a camera configuration is supported in software.

Step 5 — Validate the complete configuration

Test the final system under the intended operating conditions.
Validation should use the required resolution, frame rate, pixel format, camera count, triggering, synchronization, cabling and software configuration. For production projects, component availability, software maintenance, documentation and repeatability should also be reviewed.

MIPI CSI-2 integration

Confirm the camera connector and pin assignment, required CSI-2 lanes, carrier-board input, processor CSI-2 resources and sensor-driver support.
A physically matching connector does not confirm compatibility. Lane configuration, pixel format, device-tree configuration and the required resolution and frame rate must also be supported.

For projects that fit this architecture, start with the 36S Series MIPI CSI-2 camera range and narrow the selection by sensor, shutter type, resolution and frame rate.

GMSL2 integration

Evaluate the complete GMSL2 signal path:

Camera / serializer → coaxial → GMSL2 deserializer → carrier board → processor

The Acuva GMSL2 camera series is the relevant camera family for this integration route. For systems requiring several camera inputs, compatible hardware such as the MDSER-GMSL2-4CH GMSL2 deserializer board should be evaluated as part of the complete architecture.

FPD-Link III integration

Evaluate FPD-Link III as its own serializer/deserializer configuration:

Camera / serializer → cable → FPD-Link III deserializer → carrier board → processor

Confirm serializer/deserializer compatibility, cable and FAKRA configuration, power requirements, camera count, trigger and synchronization support, processor compatibility and the required driver/BSP version.

The 36C Series FPD-Link III cameras provide board-level and IP67 options for this architecture. Compatible multi-camera configurations can use hardware such as the MDSER-FPD3-4CH FPD-Link III deserializer board.
GMSL2 and FPD-Link III hardware and software should not be treated as interchangeable.

Hardware and software compatibility should be validated separately.

The software path must support the exact camera or image sensor, selected processing platform and required operating mode. Depending on the platform, this can include the camera driver, BSP, operating system, device-tree configuration, image formats, camera controls and acquisition framework.

Carrier-board integration can also depend on software components such as camera drivers and board support packages. 

Before hardware is fixed, verify the exact camera + processor + carrier board + BSP / operating-system combination.

MIPI CSI-2 camera integration

The 36S Series provides MIPI CSI-2 single-board camera modules for compact direct camera-to-processor integration.

GMSL2 camera integration

The Acuva Series provides GMSL2 IP67 cameras for rugged applications where the camera must be positioned remotely from the processor.

FPD-Link III camera integration

The 36C Series provides FPD-Link III cameras in board-level and IP67 configurations for remote embedded camera integration.

Carrier boards and deserializer hardware

The Carrier Boards portfolio supports the connection between camera interfaces and embedded processing platforms. MIPI CSI-2 uses a direct camera connection in supported configurations, while GMSL2 and FPD-Link III generally require the corresponding deserializer hardware. 

For initial selection, define the camera interface, processor platform, camera count, cable distance and software requirements before selecting the carrier-board configuration.