Teledyne FLIR Firefly Camera Series: Complete Setup & Usage Guide

When it comes to machine vision in space-constrained environments, few cameras have made the impact that the Teledyne FLIR Firefly series has. Whether you are integrating cameras into handheld diagnostic tools, autonomous robotics, industrial inspection rigs, or cutting-edge deep learning pipelines, the Firefly family delivers serious imaging performance without the bulk or power demands of larger systems. This guide walks you through every stage of the setup process — from understanding the hardware to configuring the Spinnaker SDK and deploying neural networks — so you can start capturing high-quality images with confidence.

What Is the Teledyne FLIR Firefly Series?

The Teledyne FLIR Firefly is a family of ultra-compact, USB3-connected machine vision cameras designed for applications where size, weight, and power consumption are critical constraints. The series includes two principal variants — the Firefly S and the Firefly DL — each optimized for distinct use cases within industrial, scientific, and intelligent edge computing environments.

Both variants share the same compact physical footprint. According to Teledyne FLIR, the camera measures just 27 mm x 27 mm x 14 mm, weighs 20 grams, and consumes only 2.2 W of power. This makes it genuinely suited for applications where traditional industrial cameras simply cannot fit or would create unacceptable power demands.

Key Specifications

Understanding the technical specifications of the Firefly series is essential for matching the camera to your application requirements. Below are the key specifications for the Firefly S 1.6 MP model (FFY-U3-16S2), one of the most widely deployed variants in the lineup.

The Firefly S uses a global shutter CMOS sensor, which is a critical advantage for industrial and motion-sensitive applications. Unlike rolling shutter sensors that scan the image progressively, a global shutter captures the entire frame simultaneously, eliminating the motion artifacts and distortion that can compromise image quality in fast-moving scenes. This makes the Firefly S particularly well-suited for conveyor belt inspection, robotics, and any application where subjects or the camera itself may be in motion.

It is also worth noting that the sensor in Firefly S board-level models is open to the camera's electronics. According to Teledyne FLIR, this design offers "greater design flexibility and cost-effectiveness for embedded and compact vision applications," though it means these models may not be ideal where high optical clarity under magnification or collimated light is a priority. Cased versions address this for those use cases.

GPIO Connector

The Firefly S is equipped with a 6-pin JST GPIO connector on the rear of the camera body. This connector supports trigger input, strobe output, and serial input/output communication. GPIO is an essential feature for synchronized multi-camera systems and production-line triggering, allowing external hardware — such as PLCs, light controllers, or other cameras — to coordinate precisely with image acquisition events. The Firefly S ships with or supports a GPIO Cable with 6-pin JST Connector, compatible with the Dragonfly S and Blackfly S Board-level cameras as well.

What You Need Before Installation

Getting your Firefly camera ready for operation requires a few components and a properly prepared system. Taking time to gather everything before you start will save significant troubleshooting time later.

Recommended System Configuration

Teledyne FLIR specifies a recommended system configuration for optimal Firefly camera operation. Your host computer should run Windows or Linux (32-bit or 64-bit), with an Intel Core i5 processor or higher, a minimum of 4 GB of RAM, and at least one available USB3 host controller port. For software development and running SDK example code, Microsoft Visual Studio 2010, 2013, or 2015 is recommended on Windows platforms.

USB3 performance is heavily dependent on your host controller hardware. Teledyne FLIR publishes a list of Recommended USB 3.1 System Components on their support center, and consulting that list before purchasing additional hardware can prevent common bandwidth and enumeration issues.

Required Components

To complete the physical installation of a Firefly S USB3 camera, you will need the following: a USB3 cable compatible with the camera's micro-B connector; a GPIO cable if you plan to use triggering, strobe, or serial I/O features; a compatible lens for the camera's CS mount; and optionally a tripod adapter for stationary setups. Teledyne FLIR sells all necessary accessories through their official Accessories page, ensuring compatibility with your specific model.

Create a Teledyne FLIR Account

Downloading the Spinnaker SDK software and firmware updates requires a registered Teledyne FLIR downloads account. The registration process is straightforward: visit www.flir.com/account, enter your email address, complete the account creation form, and activate your account via the confirmation email. Once your account is active, you have access to the full Firefly resources page, which includes the Spinnaker SDK with drivers, firmware packages and release notes, dimensional drawings and CAD models, and all technical documentation.

Physical Installation Step by Step

Installing the Spinnaker SDK

The Spinnaker® SDK is Teledyne FLIR's primary software development kit for machine vision cameras. It provides GenICam-compliant controls to create applications for the camera and is the backbone of all Firefly camera software interaction. According to Teledyne FLIR, the Spinnaker API "is used to interface with Teledyne's USB3 Vision and GigE Vision cameras" and provides access to both the acquisition engine and the configuration engine. For optimal driver performance with the Firefly camera series, Teledyne FLIR recommends installing Spinnaker version 4.0 or later.

Navigating SpinView

SpinView is described by Teledyne FLIR as "a generic, easy-to-use streaming image viewer" that allows users to access all camera controls without writing any code. The SpinView interface is divided into several resizable, dockable panes: the Devices pane lists all connected cameras and adapters; the Features pane provides access to all camera parameters by category; the Display pane shows the live image stream; and the Log Viewer records system events, saved by default to C:\ProgramData\Spinnaker\Logs.

By default, SpinView displays all features in Guru mode, giving access to every available camera parameter. You can limit visibility to Expert or Beginner mode via the context menu in the Features pane — a helpful approach when onboarding operators who do not need access to advanced controls.

Updating Camera Firmware

To update firmware through SpinView: right-click the camera in the Device List and select Update Device Firmware, or click the Update Device Firmware button from the Device pane menu. Streaming stops automatically during the update. Browse to select the downloaded firmware file, click Open, and confirm to proceed. The camera will reboot automatically once the update is complete.

Configuring Camera Settings

Once you have the camera streaming in SpinView, you can begin configuring acquisition parameters to suit your specific application. The Firefly S provides a rich set of controls for managing image quality, timing, and data output.

Exposure and White Balance

The Firefly S supports both automatic and manual control over exposure and white balance. In automatic mode, the camera's onboard algorithms continuously adjust these parameters to maintain optimal image quality. In manual mode, you have full control over exposure time in microseconds and gain values. For most machine vision inspection applications, manual control is preferred — it ensures consistent, repeatable imaging conditions regardless of changes in scene lighting.

Be aware that when the camera is first powered up or power-cycled, the auto-exposure and auto-white balance algorithms do not run during the power-down period. It may therefore take several captured frames before these algorithms converge on a satisfactory image. Planning your application's startup sequence to account for this settling period avoids processing poor-quality frames during initialization.

Triggering Modes

Triggering is one of the Firefly S's most powerful features for production and inspection applications. The camera supports both software triggering (initiated by a command from the host application) and hardware triggering via the GPIO connector. Hardware triggers allow the camera to synchronize image capture precisely with external events — conveyor position sensors, robotic arm positions, light pulses, or any other electrical signal indicating when a frame should be captured.

To configure triggering in SpinView, navigate to the Trigger category in the Features pane. Set the Trigger Source to Line 0, Line 1, Line 2, or Software depending on your wiring. Set Trigger Mode to On (waits for trigger signal) or Off (free-running). Trigger Activation specifies whether the camera responds to the rising edge, falling edge, or level of the incoming signal.

Partial Image Modes

The Firefly S supports partial image (region of interest) modes that allow you to capture only a defined rectangular portion of the full sensor area. By reducing the number of pixels transferred per frame, you can achieve substantially higher frame rates than the camera's maximum full-resolution rate of 60 fps. In inspection applications where only a narrow band of the image contains relevant data, partial image modes can dramatically improve throughput without any hardware changes.

User Configuration Sets

The Firefly S provides two user configuration sets — on-camera non-volatile memory spaces where you can store complete camera configurations including exposure, gain, trigger settings, region of interest, and more. These settings persist across power cycles without requiring the host application to re-apply them each time the camera starts. This is especially valuable in embedded deployments where minimizing startup complexity is important. To save settings, use SpinView's User Set controls or the Spinnaker API's UserSetSave command.

The Firefly DL: On-Camera Deep Learning

The Firefly DL represents a significant leap beyond traditional machine vision. While the Firefly S delivers excellent image capture and GPIO-driven inspection workflows, the Firefly DL integrates inference directly on the camera. This means a trained neural network can make classification or detection decisions without ever sending the full image to a host computer — reducing latency, eliminating data transfer overhead, and dramatically simplifying system architecture.

FLIR Neuro Technology and the Intel Myriad 2 VPU

At the heart of the Firefly DL is the Intel Movidius Myriad 2 Vision Processing Unit (VPU). Teledyne FLIR describes the VPU as "a new class of processor" that "combines high-speed hardware image processing filters, general-purpose CPU cores, and parallel vector processing cores." This architecture is purpose-built for running neural network inference at low power — far more efficiently than a general-purpose CPU or GPU in a larger system.

The FLIR Neuro technology provides an open platform supporting both TensorFlow and Caffe machine learning frameworks. This is a meaningful advantage over proprietary smart camera platforms, which typically lock developers into manufacturer-specific tools that can lag behind the broader machine learning ecosystem. With Neuro, your existing model training workflows can produce networks directly deployable to the Firefly DL.

Supported Neural Network Types

The Firefly DL supports two categories of neural networks: classification networks (which determine which category an image belongs to) and object detection networks (which identify and locate specific objects within an image). For object detection, Teledyne FLIR recommends SSD MobileNet. Notably, TensorFlow object detection is not supported — detection workflows should use Caffe-based SSD MobileNet. The Firefly DL imposes a size limit of approximately 15 MB for neural network graph files, so networks must be pruned or quantized accordingly before deployment.

Deploying a Neural Network to the Firefly DL

The deployment workflow involves training your model using TensorFlow or Caffe, converting it to an OpenVINO Intermediate Representation format compatible with the Myriad 2 VPU, and then uploading the converted graph file to the camera using SpinView's File Access tab. Teledyne FLIR provides detailed application notes for this process for both Windows and Linux, available from the Firefly DL resources page.

Once deployed, the camera runs inference continuously in the background. At 60 fps, inference results are updated every four frames captured. The Inference Frame ID chunk data feature — available through SpinView or the Spinnaker API — allows your application to correctly associate classification outputs with the specific frames that generated them.

Building Applications with the Spinnaker API

Once you have validated your camera configuration in SpinView, you can integrate the Firefly S or Firefly DL into your custom application using the Spinnaker SDK. The SDK supports C, C++, C#, VB.NET, and Python, with precompiled example applications included in the installation to accelerate your development start.

Spinnaker is built on the GenICam standard, which provides a generic programming interface for machine vision cameras. Every GenICam-compliant camera — including the entire Firefly series — exposes an XML description file that describes available features, their interdependencies, and access controls. This means applications written for the Firefly S can often be adapted to other Teledyne FLIR cameras with minimal code changes, protecting your development investment as your system scales.

QuickSpin and the Configuration Engine

Spinnaker includes QuickSpin, a wrapper API that makes interacting with the GenICam node map significantly easier. Rather than constructing feature access calls manually, QuickSpin provides named properties that correspond directly to camera features — enabling IDE autocomplete to guide feature discovery while you write code. This is particularly valuable for developers new to machine vision who need to access features quickly without memorizing node map structures.

Trigger-to-Image Reliability (T2IR)

For production inspection systems where missed images or incorrect trigger-image associations are unacceptable, Teledyne FLIR's Trigger-to-Image Reliability (T2IR) framework provides a software and hardware architecture for reliable acquisition. T2IR combines optimized connection management, built-in exception handling for common acquisition errors, system performance monitoring via Device and Stream Diagnostics, and configurable debug logging to ensure quick troubleshooting when issues arise.

Multi-Camera Synchronization

For multi-camera deployments, the Firefly's GPIO connector enables hardware-synchronized acquisition across multiple units using a primary and secondary (master/slave) configuration. The primary camera generates the trigger signal — either from software or an internal counter at a fixed frame rate — which is physically wired to the GPIO input of all secondary cameras. This ensures all cameras capture frames at the same instant, which is essential for stereo vision, 3D reconstruction, and wide-area inspection applications where temporal alignment matters.

Where the Firefly Series Excels

The Teledyne FLIR Firefly series has found widespread adoption across industries where its combination of compact size, USB3 connectivity, and imaging performance opens up deployment scenarios that were previously impractical or cost-prohibitive.

In biomedical and life science instrumentation, the Firefly S's compact form factor allows integration into microscopy systems, flow cytometers, and diagnostic devices where space inside the instrument enclosure is at a premium. Its global shutter sensor eliminates the motion artifacts that would otherwise compromise cell imaging or particle counting accuracy.

In 3D scanning and structured light applications, the precise triggering capabilities of the Firefly S allow it to synchronize exactly with projected light patterns, capturing each phase of a structured light sequence with the timing precision that accurate 3D reconstruction demands.

In autonomous robotics and drone platforms, the Firefly DL's on-camera inference eliminates the need for dedicated onboard processing hardware for computer vision — reducing system weight, complexity, and power consumption simultaneously.

In factory automation and automated optical inspection, the Firefly S's GPIO triggering, partial image modes, and user configuration sets combine to create a reliable, easily configurable inspection station that can be deployed and reconfigured quickly as production requirements change.

Support, Documentation, and Next Steps

Teledyne FLIR provides a comprehensive support ecosystem for Firefly camera users. The Firefly resources page (accessible after logging into your FLIR account) is the central hub for all downloads, documentation, and technical guidance. The Teledyne FLIR support center contains over 300 Knowledge Base articles, and a Community Forum where developers share solutions and discuss implementation challenges.

At Lore+ Technology and Wilco Imaging, our team works directly with the Teledyne FLIR product line and can help you identify the right Firefly model for your application, source the accessories and lenses you need, and provide guidance on integration and SDK development. Whether you are evaluating the Firefly series for the first time or scaling an existing deployment, we are here to support every stage of that journey.