Vision explainer
A cutaway look at how an industrial camera turns a physical part into a decision — from the light that creates contrast, through the optics and imager, to onboard processing and the interface that hands the result to a controller.
Explore the light-to-data pathEach block in the chain changes the signal. In the scene below, adjust the controls to see where contrast, resolution, and decisions are actually made.
Creates contrast before the sensor ever sees a pixel. Type, angle, and color decide what the camera can and cannot see.
The lens maps a working distance and field of view onto the sensor. Focal length and aperture set magnification and depth of field.
The imager samples light into pixels; exposure and gain affect signal quality. Each pixel is one intensity measurement.
Onboard processing can reduce raw image data to usable features or pass/fail decisions before it ever leaves the camera.
Interface choice affects bandwidth, cable length, topology, and controller integration — from GigE to discrete pass/fail I/O.
A multi-camera master can simplify wiring by concentrating camera connections before the controller — one upstream link instead of many.
The controls and sensor preview still work. For the 3D cutaway, open this page in a current desktop browser with hardware acceleration enabled.
A monochrome imager records only intensity, so a colored light lets you engineer contrast the eye can't. A color (Bayer) imager captures hue but trades resolution and light. Beyond the visible, UV, near-IR and SWIR expose material properties — moisture, fill level, and plastics that look identical in color.
The imager's physical size and pixel pitch, together with the lens's focal length and the working distance, set how many pixels cover the part — and how small a feature you can resolve.
Shares focal length (25 mm) and working distance (200 mm) with the scene above. A larger sensor widens the field of view; smaller pixels raise resolution but gather less light; the lens image circle must cover the sensor or the corners vignette.
Resolution and frame rate multiply into the data rate the interface must carry; shutter speed decides whether a moving part is sharp or smeared. Raise resolution or FPS and you outgrow slower links.
A fatter pipe carries the data rate that resolution, color and frame rate demand — but the tradeoff is cable reach and how well it suits the factory floor. Each row below is checked against the rate you built up above; pick the interface that fits.
A global shutter exposes every pixel at once, so a fast-moving part freezes in its true shape. A rolling shutter reads each row a moment later, so the image warps — the classic "jello" skew and shattered-propeller effect.
Every pixel samples the same moment — true geometry.
Each row samples a later instant — it bends or shears.
Under the hood they share the same chain — lighting, lens, imager. What differs is where the processing lives: inside the device, in a dedicated controller, or in an external PC or the cloud. That choice sets how much the system can do, how fast it deploys, and how flexible it is.
Self-contained — a sensor that can see.
Processing on the camera or the controller.
A component — you bring the brain.
How to choose: pick a vision sensor for a fixed pass/fail with almost no setup; a vision system — a smart camera or a camera plus controller — for complex inspection, where the vision software is the most complex of the three and a vendor black box you depend on for tools and support; an industrial camera when you want to run your own algorithms or ML on an industrial PC or in the cloud, keeping the processing open and under your control.