Industrial camera specifications: a buyer's guide

A camera can meet the megapixel requirement and still fail an inspection because its shutter distorts moving parts, its exposure needs more light, or its interface cannot deliver the selected image format. Industrial camera specifications become useful only when you connect them to the inspection conditions.

This guide helps you turn feature size, motion, contrast and timing requirements into a defensible camera shortlist. Start with measurable acceptance limits, then compare sensor, acquisition and optical specifications together. The aim is not the largest specification sheet; it is repeatable images under production conditions.

Start with the inspection limits, not the camera

Write down what the image must reveal before comparing area scan cameras. A useful requirement describes the smallest relevant feature, its contrast, the inspected area and the available acquisition time.

For example, detecting a dark mark against a pale surface is different from measuring a low-contrast edge. Both may involve the same feature width, but they need different evidence of image quality.

Specify the failure criteria too. Examples include missing a feature, saturating a reflective region, accepting an incomplete frame or exceeding a measurement tolerance. Those criteria determine which specifications are mandatory and which are preferences.

Read pixel size, sensor size and shutter together

Resolution states how many samples the image contains. Pixel size describes the physical spacing of those samples on the sensor. Neither tells you object-space detail without the lens magnification or field of view.

For a hypothetical sensor with 4,000 pixels across and a 3.45 μm pixel pitch, the active width is approximately:

That physical width affects lens coverage and the focal length needed for a chosen inspection width. Nominal optical-format labels are not literal sensor diagonals in inches. The lens sensor format explanation explains why coverage needs a separate check.

A larger pixel can collect more photons under otherwise comparable conditions, but pixel size alone does not establish sensitivity. Quantum efficiency, exposure, aperture, wavelength, read noise and saturation capacity also matter. Where available, compare measurements made under the EMVA 1288 framework rather than relying on pixel dimensions alone.

Shutter architecture answers a different question. A global shutter exposes the image pixels over a common interval. A rolling shutter exposes successive rows at different times, so motion can produce skew even when each row has little blur.

For a global-shutter candidate, examine the MV-CU020-90GC. Its catalog identifies color acquisition and a C-mount. That makes it a different starting point from the monochrome rolling-shutter examples later in this guide, not an interchangeable replacement.

Calculate sampling, exposure and payload together

Consider an illustrative inspection covering 160 mm horizontally. The smallest relevant feature is 0.20 mm, and the initial sampling target is five pixels across that feature. This is a starting assumption to validate with images, not a universal detection rule.

The camera resolution selection guide develops the distinction between sampling and measurement capability. Use it when an inspection tolerance, rather than simple presence detection, drives the decision.

Now suppose the part moves at 800 mm/s. Limiting motion blur to one object-space pixel gives the following exposure estimate:

This estimate assumes lateral motion across the image at the stated speed. Rotation and perspective can change local image velocity. The exposure limit is therefore a lighting requirement as well as a camera requirement.

Finally, suppose the illustrative image contains 4,000 × 3,000 pixels and the application needs 20 frames/s. Uncompressed image payload depends on the transmitted format:

These are image-payload estimates using decimal megabytes. They exclude protocol overhead, metadata and host-side copies. A nominal 1 Gbit/s link has a theoretical ceiling of 125 MB/s before overhead, so it cannot carry this example continuously without changing the acquisition conditions.

Separate bit depth from usable image quality

Output bit depth describes the number of available digital codes. An 8-bit image has 256 possible values; a 12-bit image has 4,096. More codes can preserve small brightness differences, but they do not create additional photons or eliminate noise.

Ask three separate questions:

• What conversion depth does the sensor or camera support?
• Which pixel formats can the camera transmit?
• Which formats does the processing application retain without truncation?

An image delivered in a 16-bit container does not necessarily contain 16 bits of meaningful signal. Equally, a packed format changes storage and transport requirements without changing the underlying scene information.

For faint defects, examine noise and clipping before choosing a higher output depth. If a feature changes intensity by fewer counts than the image fluctuates between exposures, extra digital codes alone will not make the decision stable.

Higher depth can still help when processing uses shading correction, subtraction or subtle intensity thresholds. Verify the benefit with representative images and the intended processing chain. For color acquisition, the RGB frame-rate explanation also shows why transmitted format must be included in throughput checks.

Check acquisition controls and delivered speed

Maximum frame rate is a conditional specification. Resolution, pixel format, exposure, readout mode and transport configuration can all affect the delivered rate. A smaller region of interest may improve performance, but not every camera accelerates equally for every crop direction.

External triggering also needs more detail than a yes-or-no entry. Confirm the trigger input's electrical requirements, supported polarity, exposure mode, minimum spacing and behavior when another trigger arrives too soon.

Use the hardware-trigger timing guide when the machine defines acquisition timing. A free-running demonstration does not establish that the same camera will accept every production trigger.

For lighting synchronization, determine what the camera output actually represents. An exposure-active output, trigger acknowledgement and general-purpose output are not interchangeable. Confirm their timing against the light controller's documented input requirements.

During evaluation, count requested triggers, accepted acquisitions and complete received images separately where the equipment exposes those counters. This distinguishes missed triggers from transport loss or application delays.

Save the intended settings, restart acquisition and read them back. The startup settings verification guide provides a useful acceptance procedure for configuration persistence.

Match spectral response, lens and illumination

Spectral response is a curve, not simply a wavelength range. A sensor may respond to a wavelength while the complete camera transmits little of it because of an installed filter or protective window.

Check the response of the complete optical path: illumination, lens, filters, camera window and sensor. For color cameras, also confirm how the color-filter array and any infrared-rejection filter affect the intended measurement.

Do not assume a filter can be removed or exchanged. Select a documented configuration suitable for the wavelength, or ask for confirmation through applications support. Internal modifications can introduce contamination, focus changes and unsupported operating conditions.

For machine vision lenses, match the mount, image circle, working distance and required detail across the field. A mechanically compatible thread does not establish adequate sensor coverage or edge sharpness.

The LS-FFL-0567 is a lens candidate whose catalog lists sensor-format coverage and C-mount construction. It illustrates the optical checks needed alongside camera selection; compatibility with a particular assembly still requires verification.

Use the lens finder to narrow the optical choices, then apply the machine vision lens selection guide to the working distance and inspection field.

Lighting must deliver useful contrast within the exposure budget. Raising gain amplifies signal and noise; it does not replace missing photons. Before accepting a camera, capture the darkest acceptable part and the most reflective acceptable part under the proposed illumination.

Decide whether area scan or line scan fits the motion

Area scan captures a rectangular image and suits discrete parts, indexed stations and scenes needing simultaneous spatial context. Line scan cameras build an image from successive lines and suit continuous webs or controlled relative motion.

For an illustrative transport speed of 800 mm/s and an along-motion sampling target of 0.040 mm per line:

Line rate does not replace the across-web pixel-count calculation. You must satisfy both directions, then provide enough illumination for the line exposure and enough transport capacity for the resulting data.

The LS-LCM-0016 is a catalog example with GigE acquisition and an M42 lens mount. Treat it as a line-scan candidate to investigate, not evidence that any required width, line rate and output format can operate together.

Use the line-rate setup guide to connect acquisition timing to transport motion. Encoder synchronization is particularly useful when speed varies, provided the camera and motion interface support the required arrangement.

Build a shortlist by eliminating mismatches

The LS-CMR-0053 and LS-CMR-0166 illustrate why similar resolution does not imply equivalent acquisition performance. Both catalog entries identify rolling shutters, while their interfaces and listed frame rates differ.

All 2 share: Sensor size 1/1.7". Values as printed in each product's data sheet; open a part for its drawing and full specifications.

Apply the worked example as an elimination exercise. Its 240 MB/s monochrome payload exceeds the theoretical ceiling of a nominal 1 Gbit/s link. The GigE option therefore cannot satisfy those unchanged full-frame transport conditions.

The USB3 option remains a candidate for transport evaluation, not an approved solution. Its listed frame rate alone does not establish operation at the required exposure, pixel format, trigger timing and host configuration. Its rolling shutter also remains a separate motion constraint.

The earlier global-shutter candidate addresses shutter architecture, but that does not establish adequate sampling or optical compatibility. Do not trade away required pixels simply to obtain a preferred shutter.

Use the camera finder to filter mandatory requirements first. Then rank the remaining candidates by demonstrated image quality, integration effort and system margin. If none meets every mandatory condition, change the system architecture or request a different candidate rather than silently relaxing acceptance limits.

Recognize when the usual answer is wrong

A global shutter is not automatically necessary for every inspection. A stationary part under stable illumination may work well with a rolling shutter. Conversely, a short exposure alone does not remove rolling-shutter skew from a moving scene.

Monochrome is often appropriate when brightness contrast carries the decision. Color becomes necessary when otherwise similar parts must be separated by hue. Test the actual discrimination task rather than choosing color for general appearance.

More pixels are not always helpful. With unchanged field of view, they reduce object-space sampling distance but increase payload and can demand better optics. They do not correct defocus, glare, vibration or insufficient contrast.

Likewise, a wider spectral response is not automatically an advantage. Unwanted wavelengths can reduce material contrast or introduce focus differences. The useful response is the one matched to the inspection and illumination.

A larger sensor can also require a larger image circle and different lens geometry. Compare the complete camera, lens, light and acquisition path rather than treating the camera body as an isolated purchase.

Check the complete configuration before ordering

Use this checklist to turn the shortlist into a testable specification:

• Sampling: Record inspection width, required feature size and pixel coverage. Include placement variation and edge-of-field performance.
• Motion: Calculate the exposure budget and verify shutter distortion at maximum production motion.
• Image quality: Test dark surfaces, reflective surfaces and low-contrast defects without clipping important regions.
• Data format: Confirm the transmitted pixel format, packing and processing application's handling of higher-depth images.
• Throughput: Measure sustained delivery at the actual image dimensions, exposure and acquisition cadence.
• Controls: Verify trigger levels, accepted-trigger behavior, lighting synchronization and recovery after interrupted acquisition.
• Optics: Confirm sensor coverage, mount compatibility, working distance, focus retention and wavelength suitability.
• Configuration: Restart the system and check that the intended settings are restored and read back correctly.
• Acceptance evidence: Retain representative images, acquisition settings and pass/fail limits so the selection can be repeated.

Start the camera configuration process with these requirements rather than a preferred megapixel count. If shutter timing, spectral response or transport performance remains undocumented, send the requirement list through the contact page before committing to a camera and lens combination.