Frequently Asked Questions
Machine Vision Questions, Answered
These are the questions engineers and buyers ask us most often when scoping a machine vision project, answered plainly and without spin. If you are still working out what you need, that is exactly what this page is for.
Some topics deserve more than a short answer. For pricing, see our cost guide, to compare suppliers across Europe see our suppliers guide, and to gauge how ready your application is, try our readiness assessment.
Last reviewed June 2026
Specifying a system
Building a vision system
What components do I need for a machine vision system?
A machine vision system needs five core components: a camera, a lens, structured lighting, a frame grabber or direct interface, and image processing software. Each has to be matched to the application, since a high-speed packaging line and a precision measurement station need very different parts.
Clearview supplies all five categories from manufacturers including Teledyne, Zebra Technologies, LUCID Vision Labs, Computar, Kowa and Advanced Illumination. You can browse them on our components pages.
How do I choose a machine vision system for quality inspection?
Start with the inspection task itself: what defect or feature you need to catch, how small it is, how fast the line runs, and to what tolerance. Those four answers drive the camera resolution, the lens, the lighting and the interface.
The defect type usually dictates the lighting first. Backlighting suits dimensional gauging and silhouettes, low-angle or darkfield lighting reveals surface scratches and embossing, and dome lighting tames glare on shiny or curved parts. As a guide, the smallest feature you must detect should span three to four pixels, so a 0.2mm flaw across a 200mm field of view needs a multi-megapixel camera. Clearview sizes all of this against your own samples in the Insights Test Lab before you commit.
How do I choose the right lens for my camera?
Lens choice comes down to three numbers: the sensor size, the working distance, and the field of view you need. Together they fix the focal length, and the lens must also cover the full sensor without vignetting and resolve enough detail for the camera's pixels.
It is worth getting right, because a high-resolution sensor behind a poor lens still gives a poor image. As a rough guide the focal length is about the sensor size times the working distance divided by the field of view, while telecentric lenses are used where accurate measurement matters. Our lens calculator works this out for you and recommends a matching lens from our range, and our engineers will confirm it in the Insights Test Lab. Lenses come from Computar, Kowa and Theia.
How do I set up a machine vision inspection system?
Setting up an inspection system follows a clear sequence: define the pass and fail criteria, choose and mount the camera, set up the lighting, connect the right interface, configure the software, then validate against known good and bad samples.
Get the lighting right first, because no software can recover detail the lighting never captured. In practice, lighting and optics cause more failed deployments than cameras do. Clearview supplies the illumination, lenses and software to do this, offers hands-on setup support through its engineering services, and runs KnowHow training for teams who want to run and maintain the system themselves.
What are the main machine vision camera interfaces?
The main machine vision camera interfaces in use in 2026 are GigE Vision, USB3 Vision, CoaXPress, Camera Link and Camera Link HS. The right one depends on the bandwidth you need, the cable length, and whether a dedicated frame grabber is acceptable.
GigE Vision is the most widely used for its flexibility and 100 metre reach, and now spans several speeds, from standard 1GigE up to 2.5, 5, 10 and 25 Gigabit, which closes much of the bandwidth gap to CoaXPress while keeping standard Ethernet cabling. CoaXPress and Camera Link HS remain the choices when bandwidth is most critical. Clearview supplies cameras across every interface, plus frame grabbers from Zebra Technologies (formerly Matrox Imaging) and Neousys.
| Interface | Typical bandwidth | Max cable | Frame grabber | Best for |
|---|---|---|---|---|
| GigE Vision (1 to 25GigE) | 1, 2.5, 5, 10 or 25 Gbps | Up to 100 m, less at higher speeds | No | Flexible, multi-camera and longer runs |
| USB3 Vision | ~5 Gbps | ~3 to 5 m | No | Single camera, short distance, low cost |
| CoaXPress (CXP-12) | Up to ~12.5 Gbps per cable | Tens of metres | Yes | High speed and high resolution |
| Camera Link | Up to ~6.8 Gbps | ~10 m | Yes | Established high-speed lines |
| Camera Link HS | Up to ~16 Gbps per cable, scalable | ~15 m copper, further over fibre | Yes | Very high speed and long fibre runs |
Applications
What vision can do
What is OCR in machine vision?
OCR (optical character recognition) in machine vision is the automated reading and verification of printed or engraved text on a production line, such as lot numbers, expiry dates, serial numbers and regulatory markings.
Industrial OCR is harder than document OCR because it has to cope with low contrast, variable fonts, curved surfaces and high line speeds. Clearview supplies Zebra Aurora software for industrial character recognition and verification, including deep-learning OCR for awkward surfaces.
What machine vision solutions are used in food and beverage manufacturing?
Machine vision in food and beverage covers label verification, barcode validation, date code reading, fill level inspection, packaging integrity checks and product sorting.
Label verification is the most critical, driven by retailer codes of practice and the cost of withdrawing mislabelled product. Clearview provides dedicated label verification solutions powered by its own Kinetic Velocity software.
Can machine vision inspect reflective or shiny surfaces?
Yes. Reflective and shiny surfaces are one of the most common machine vision challenges, and they are usually solved with lighting and optics rather than a different camera. The aim is to control how light reflects into the lens so the feature you care about stands out.
Common techniques include diffuse dome lighting to wrap even light around curved metal, polarising filters to cut hot spots and glare, and darkfield or low-angle lighting to reveal engraving and surface defects. This is exactly the kind of problem Clearview proves out on your own parts in the Insights Test Lab before you buy anything.
Costs & results
Cost, return and fit
How much does a machine vision system cost?
It depends on the application. A single smart camera for a simple check can cost a few thousand pounds, while a multi-camera, custom-engineered inspection system can run to tens of thousands once software and integration are included.
The main cost drivers are resolution and speed, the number of inspection points, and how much custom software and engineering the job needs. Our machine vision cost guide breaks this down with honest price ranges for components, systems and services.
What ROI or results can machine vision deliver?
Machine vision delivers results manual inspection cannot match: objective, repeatable pass and fail decisions made the same way every time, at line speed, with a traceable record of what was checked. It does not tire, drift or vary between shifts.
The returns usually come from fewer escapes and recalls, less scrap and rework, reduced labour on repetitive checks, and a data trail that supports audits and continuous improvement. The strongest business cases combine quality protection with a measurable cut in the cost of getting it wrong, and Clearview's feasibility studies help you estimate the gain before you invest.
Is machine vision only for large manufacturers?
No. Machine vision is no longer only for large manufacturers. Compact smart cameras and off-the-shelf components have brought the entry cost down, so a single well-chosen camera can solve a specific problem on one line for a modest spend.
Smaller manufacturers often start with one high-value inspection, such as label or date code verification, and expand from there. Clearview works with businesses of every size and will tell you honestly if vision is not the right answer for your problem yet.
Working with Clearview
How we work with you
Do you offer machine vision feasibility testing?
Yes. Clearview runs feasibility studies and proof-of-concept demonstrations in its Insights Test Lab in Thame, Oxfordshire, where you submit sample parts and our engineers evaluate them under different camera, lens and lighting setups.
This tells you whether vision can reliably meet your requirement, on your actual parts, before you commit to hardware. A loan equipment programme is also available so you can evaluate components in your own production environment. See the Insights Test Lab for more.
What is the difference between buying from a distributor and buying direct from a manufacturer?
Buying from an independent distributor gives you components from several manufacturers plus vendor-neutral advice on how to specify the system. Buying direct from a manufacturer can mean deeper support on that one brand, but limits you to a single product range.
This matters because most vision systems combine parts from different makers, so the best result rarely comes from a single catalogue. Clearview is built around that independent model. Our suppliers guide compares the wider market.
| Independent distributor | Manufacturer-direct | |
|---|---|---|
| Product range | Multiple manufacturers | One manufacturer |
| Advice | Vendor-neutral, application-led | Product-specific |
| Best when | You need to combine the best parts | You already know the exact product |
Which camera brands does Clearview supply?
As of 2026 Clearview supplies cameras from more than a dozen manufacturers, led by Teledyne and Zebra Technologies, alongside LUCID Vision Labs, CIS, Photonfocus and Vieworks.
Zebra also supplies smart cameras and fixed industrial scanners, Neousys supplies embedded vision computers and frame grabbers, and FLIR supplies thermal cameras. This multi-vendor range lets our engineers pick the best camera for each job on resolution, speed, interface and environment. Lenses come from Computar, Kowa and Theia, with illumination from Advanced Illumination and ProPhotonix. The full range is on our components pages.
Do you provide machine vision training?
Yes. Clearview KnowHow runs machine vision training through courses, webinars and seminars, for engineers at every level from newcomers to experienced integrators.
Topics range from machine vision fundamentals and choosing cameras and lenses, through lighting and illumination technique, to advanced areas like 3D, deep learning and application-specific inspection. Training runs in person at our facility or yours, and online as live webinars. For current dates and formats, see the KnowHow training pages or ask our team.
Does Clearview build custom vision solutions?
Yes. Clearview designs custom vision solutions through its ClearviewFormula process, a four-stage method covering discovery, a technical deep dive, analysis in the engineering labs, and an in-person solution presentation.
Feasibility testing on your real production samples runs in the Insights Test Lab, and loan equipment is available for on-site evaluation before you commit to a full system. See our custom vision solutions page.
Let's Talk
Still have a question?
Tell us what you are trying to inspect, measure or guide, and one of our certified machine vision experts will help you work out the right approach. No pressure and no obligation.