Line-Scan vs. Area-Scan Lighting: Why Getting It Wrong Costs You More Than a Bad Image

Your line-scan system may have the right camera, encoder, and software. Defects can still slip through. When that happens, teams often look first at processing power, smarter algorithms, or higher-resolution optics. In many cases, the issue starts earlier in the imaging chain.

The root cause is usually the light.

Lighting can feel like a solved problem: mount a bar light, verify the image during commissioning, and move on. That approach often works for area-scan cameras inspecting discrete parts. In line-scan applications, however, lighting problems can surface quietly, repeatedly, and expensively.

The reason is simple: area-scan and line-scan cameras collect light in fundamentally different ways.
 

How does area-scan lighting work?

Area-scan cameras capture an entire scene in one snapshot. The shutter opens, every pixel collects light at the same time, and the image is complete. That is why area scan remains the default choice for many machine vision installations: it is familiar, forgiving, and flexible.

Lighting is usually straightforward. For slow-moving or stationary parts, modest diffuse illumination can produce a usable image. If light levels are low, the LED illuminator can be strobed in short, intense pulses at up to ten times its rated power. That gives engineers a practical compensation lever when light is scarce.
 

What changes when you switch to line scan?

Line-scan cameras build an image one row of pixels at a time. As material moves beneath the camera, thousands of scan lines are stitched together to form a complete image. This approach is ideal for continuous web materials such as paper, film, foil, textiles, coated substrates, and metal sheet, as well as large or cylindrical objects that require high-resolution inspection. 

Line-scan systems can deliver inspection speeds 20 to 30 times faster than area-scan sensors for continuous objects. That speed creates a physical constraint that changes how engineers need to think about light.

Each scan line is exposed for an extremely short window, often around 30 microseconds. In that moment, every pixel in the camera's single row must collect enough light to produce a usable signal. Because the camera runs continuously, there is no practical strobe overdrive to rely on and no second full-frame capture behind it. The system has to get that line right as it passes.

Ethan Nash, Application Engineer at Cognex, offers an analogy that makes the physics click immediately:

I usually explain line scan by comparing it to an MRI. We capture one thin slice at a time, then build those slices into a complete image. For lighting, think of an overhead room light versus a reading lamp. Area scan spreads illumination across a broad field; line scan concentrates it into a very narrow line exactly where the lens is looking. Because we're not lighting a wide area, a focused bar light can deliver stronger, more controlled illumination at the inspection point.

Table compares area-scan and line-scan lighting needs, showing line scan requires stricter exposure, uniformity, and alignment.
 

Why does line scan need so much more irradiance?

The physics becomes clearer with a simple example.

Compare a line-scan camera with 1,000 pixels in its row to an area-scan camera capturing a 1,000 x 1,000 pixel image. If the area-scan camera uses a 30-millisecond exposure on a stationary object, the line-scan camera scanning the same scene in the same total time gives each line just 30 microseconds of exposure.

That 1,000:1 exposure ratio means the line-scan camera needs 1,000 times the irradiance, or light power per unit area, to produce the same pixel signal. Because area-scan lighting covers a full field of view while line-scan lighting covers only a narrow strip, total power can be roughly equal in the static case.

Everything changes when the material moves.

In motion, the area-scan camera must also cut its exposure to 30 microseconds to prevent blur, which means it needs far more light spread across the full field of view. The line-scan camera naturally limits blur to one-pixel width, so its irradiance requirement remains concentrated on the scan line.

In real-world moving applications, concentrated irradiance on a narrow strip is part of the inspection requirement. That is why machine vision lighting for line scan should be selected and positioned as a precision engineering decision, not treated as an afterthought. 

Inadequate lighting in a line-scan system doesn't always produce a dark image. More often, it produces data that looks acceptable but lacks the contrast needed for reliable detection. Fine surface features, such as a hairline void in a coating, a subtle texture variation, or a micron-scale scratch, can fall below the detection threshold. The inspection point may then pass product that should have been flagged.

Defects on a moving web are difficult to recover once missed. By the time a downstream quality check catches an issue, meters of material may have already passed inspection or reached a later stage in production. In regulated industries, like medical packaging or food-contact films, the consequences can extend beyond scrap and returns.

Nash encounters this failure pattern regularly:

I see this often: customers assume the camera or software is the problem, when the real issue is lighting alignment. In line scan, the light must hit the exact point the camera is focused on. If the camera is slightly rotated or the light angle is off, the focal points no longer align, resulting in uneven illumination across the image. When working distance, angle, and alignment are corrected, the image becomes much more evenly lit, stable, and reliable.

Request a Demo → 
 

Three lighting mistakes line-scan engineers make

Most practitioners new to line scan carry over instincts from area-scan experience. Three assumptions cause the most production problems.

1. Using the wrong light type

Nash identifies the issue directly:

The detail I see people overlook is the need for line-scan-specific lighting, or at least a bar light with a true line-scan filter. Line scan systems use very short exposure times, so the light must be tightly focused. A standard bar light wastes intensity by spreading illumination too broadly. A dedicated line-scan light delivers usable intensity exactly where the camera needs it.

Standard ring lights or dome lights designed for area scanning won't achieve the spatial precision that line scanning demands. Dedicated machine vision lighting delivers the focused, uniform beam geometry the application requires.

2. Treating mounting as a setup step rather than an engineering decision

Line-scan cameras are sensitive to the exact geometric relationship between the light, the inspection line, and the camera. Even a small rotation in the camera or light mount can create brightness differences that mimic real defects or make subtle ones harder to see.

Nash is direct about the discipline this requires:

Where I see line scan lighting go wrong most often is treating it as if 'close enough' is good enough. With area scan, a rough working distance or angle may still work. With line scan, the margin is much smaller. Working distance, camera angle, light position, and illumination angle all need to match the lab setup precisely. I rely on photos, measurements, and notes because small changes can create very different production results.

Lighting placement should be specified during design and documented clearly, rather than adjusted during installation “by feel.”

3. Ignoring speed variation in the exposure budget

In most line-scan installations, exposure time is set based on maximum line speed. If conveyor speed varies due to tension fluctuations, mechanical jitter, or upstream process changes, the time available per scan line also varies. Some areas become underexposed. Others become overexposed. An inconsistent image defeats the purpose of automated inspection.

Lighting should be engineered for the most demanding case: fastest line speed, maximum working distance, and the tightest defect specification. Systems that pass commissioning at normal speed but struggle at peak production are often lighting-constrained.

Explore machine vision lighting options → 
 

When is line scan the right choice?

Line-scan technology fits four scenarios particularly well:

1. Continuous web materials, including paper, film, foil, textiles, nonwovens, coated substrates, and metal sheets
2. Cylindrical or curved surfaces, where the object rotates past the camera to "unwrap" the full surface for inspection
3. Large discrete parts, where a single area-scan frame can't deliver the required pixel resolution across the entire object
4. High-speed production lines, where area-scan strobe lighting becomes physically impractical at line rates exceeding two meters per second

In each scenario, lighting should be specified as part of the optical system design. Irradiance, wavelength, beam geometry, working distance, and mounting angle all interact. Changing one without accounting for the others can degrade performance in ways that may not appear during commissioning, then surface later at production speed.

Cognex line-scan vision systems, including the In-Sight 3800 Line Scan, handle continuous web inspection, cylindrical surface inspection, and large-format defect detection with AI-powered analysis at full line speed. The system pairs advanced processing with purpose-built machine vision lenses and dedicated lighting, giving manufacturers a co-engineered optical foundation. Even the most capable camera still depends on lighting that delivers adequate irradiance, uniformity, and geometric precision. 
 

Build the light into the system, not as an afterthought

According to SNS Insider, the machine vision lighting market reached USD 1.87 billion in 2024 and is projected to reach USD 3.65 billion by 2032. That growth reflects a broader shift: manufacturers are treating illumination as a precision component rather than a commodity accessory.

In line-scan applications, that shift matters. Lighting is a design variable that influences what defects your system can detect, at what speeds, across what material widths, and under what production conditions.

Get the camera right. Get the encoder right. Get the software right. Then engineer the light with the same discipline. Without that foundation, even a sophisticated inspection system can miss the defects it was designed to catch.