Menu
Book a demo
Line Scan vs Area Scan Camera: How to Choose, Size and Cost Each One

Line Scan vs Area Scan Camera: How to Choose, Size and Cost Each One

When a line scan camera beats an area scan camera, how to size line rate, pixels, light and cables, and what a blind spot on a film web costs a year.
Line Scan vs Area Scan Camera: How to Choose, Size and Cost Each One

In short: An area scan camera takes a whole 2D picture in one exposure, so it suits parts that arrive one at a time. A line scan camera takes one row of pixels per trigger and builds the picture as the product moves past, so it suits continuous webs, long parts and round parts turned in front of it. On a moving web, the line rate is the web speed divided by the pixel size, and that one number sets the exposure, the light and the cable.

  • Pick area scan for packs, pouches, bottles and parts a photo-eye can trigger. Pick line scan with an encoder for film, foil, paper, nonwovens and print at speed.
  • Example plant: a 1.6 m film web at 300 m/min with 0.2 mm pixels needs 25,000 lines a second, a 40 microsecond exposure and about 205 million pixels a second.
  • Against a strobed web viewer that images a quarter of the web at best, full coverage keeps $58,228 to $72,305 of film a year out of the scrap in the example, priced with UN trade data.

Guide, updated 16 November 2026. Sources are listed at the end.

Choosing between the two is a geometry problem before it is a shopping problem. Below, each camera type is matched to the kind of station it serves, then sized with four numbers (pixel, line rate, exposure and data rate), and the choice is costed on one example plant with a film web and a pouch line. All of it is about inspection cameras that look at product. If the camera you have in mind is meant to count output for OEE instead, this sizing does not apply, and the guide to the best computer vision tools for manufacturing OEE covers that job.

How each camera builds an image

The standard that camera makers use to name camera features, the GenICam Standard Features Naming Convention (SFNC) from the European Machine Vision Association (EMVA), defines the two scan types in a line each. An area scan device is a 2D sensor whose image comes from a single acquisition. A line scan device is a 1D sensor whose image is acquired line by line (SFNC version 2.7.1, February 2022).

Area scan. The sensor is a grid of rows and columns. One trigger, one exposure, one frame. The part can stand still under the camera, or a short exposure or a strobe can freeze it while it moves. A photo-eye or the machine's PLC says when the part is in place.

Line scan. The sensor is one row of pixels, sometimes a few rows. Each trigger records a thin strip across the product. The motion supplies the second dimension: the web moves on by one pixel length, the camera takes the next line, and software stacks the lines into an image. There are no frame edges along the direction of travel, so a whole roll of film can be one continuous picture.

On a datasheet, both are measured the same way. EMVA 1288 Release 4.0 Linear, the EMVA standard for characterising image sensors and cameras with a linear photo response (16 June 2021), is valid for area scan and line scan cameras. It treats a line scan sensor as if it were an area sensor: the lab acquires at least 100 lines into one image and runs the same evaluations, except the vertical spectrogram. A supplier who quotes EMVA 1288 figures for a line scan camera is quoting numbers you can set beside an area scan datasheet.

Area scan or line scan?How each one builds an imageArea scanone 2D frameper partencoderLine scanone row per pulse,stacked by motionExample plant: 1.6 m web, 300 m/min0.2 mmpixel: 3 on a0.6 mm fault8,192pixels across1.6 m web25 kHzline rate5,000 / 0.240 µslongestexposure1.64Gbit/s raw8 bit data4encoder countsof 0.05 mmSize pixel and line rate first.Light and cables follow.
Example plant sizing for one film web station. Scan type definitions from the EMVA's GenICam SFNC 2.7.1 (2022). See the source.

Web or parts: the first question at the station

The choice starts with how the product reaches the camera. This guide sorts the stations into five patterns.

  • A continuous web. Film, foil, paper, board, nonwovens, labels on a roll, coated or printed material. There is no natural moment to take a picture, and frames would cut the product into tiles that have to be stitched back together. Line scan is the default. Several area cameras can cover a web too, at a cost shown in the worked example.
  • Parts that arrive one at a time. Pouches, cartons, bottles, caps, moulded parts, assemblies. Each one can trigger a frame as it passes a photo-eye, and each image holds one part. Area scan is the default.
  • Items on a steady, continuous line. Products in rows on an oven band, items on a sorting belt, birds on a shackle line. Line scan works here when the line moves at a measurable speed and the job needs a view without seams or a spectrum for each line. Scientists at the US Department of Agriculture's Agricultural Research Service ran a line scan machine vision system on an eviscerating line at 140 chickens per minute, built from an electron-multiplying CCD camera and an imaging spectrograph (Yang, Chao and Kim, 2009). Their spectral line scan system on a 140 bird per minute line picked out unwholesome birds with over 99 percent accuracy in online tests (Chao, Yang and Kim, 2010). The geometry is the lesson: a spectrograph spreads each line into wavelengths along the sensor's second axis, which only works if the product supplies the motion.
  • Round parts whose whole surface matters. Cans, tubes, rollers, bearings. Turn the part in front of a line scan camera and its surface unrolls into one flat image with no seams. A 66 mm can has a circumference of 207.3 mm, so at 0.1 mm per pixel one turn takes 2,073 lines, and turning 5 times a second needs about 10,400 lines a second (example data). Area scan can do the same job with several cameras around the part and software that joins the views.
  • Long or large parts. Boards, panels, profiles and sheets longer than one frame. Line scan images them in one pass, or the camera travels on a gantry over a part that stands still.

Size a line scan station in four numbers

Four numbers decide whether a station works: pixel size, line rate, exposure and data rate. Work them out before the first supplier meeting, from the line's rated top speed rather than today's speed. The figures below are the Example plant's web station: a printed polyethylene film 1,600 mm wide on a press rated at 300 m/min.

1. Pixel size, from the smallest fault you must find

Start from the smallest fault that matters to the customer and decide how many pixels must land on it. This guide assumes 3 pixels across the smallest fault; ask each supplier what their software needs. For a 0.6 mm ink spot or hole, 0.6 / 3 = 0.2 mm per pixel on the film.

2. Line rate, from speed and pixel size

Line rate (lines per second) = web speed (mm per second) / pixel size along travel (mm)

At 300 m/min the web moves 300,000 / 60 = 5,000 mm a second. At 0.2 mm per line, 5,000 / 0.2 = 25,000 lines a second, or 25 kHz. Across the web, 1,600 mm at 0.2 mm needs 1,600 / 0.2 = 8,000 pixels, so an 8,192 pixel sensor (8,192 x 0.2 = 1,638 mm of view) covers it with 19 mm to spare at each edge. Choose a camera whose maximum line rate sits above 25 kHz with margin, because the line rate rises in step with speed: at 400 m/min the same pixel needs 33,333 lines a second.

3. Exposure, capped by the line rate

A line scan camera cannot expose one line for longer than the time between lines: 1 / 25,000 of a second, or 40 microseconds here. An area camera looking at the same web hits the same ceiling for a different reason. In 40 microseconds the web moves 5,000 x 0.00004 = 0.2 mm, one pixel, and a longer exposure blurs the image by more than a pixel. That short exposure is why light becomes the hard part of a web station.

4. Data rate, from pixels and lines

8,192 pixels x 25,000 lines = 204.8 million pixels a second. At 8 bits a pixel that is about 1.64 gigabits a second before protocol overhead, more than a gigabit Ethernet link carries. This number chooses the interface, limits the cable length and sizes the PC that has to process the image.

Sizing, Example plant web stationOne line scan cameraFour area cameras (5 MP each)One area camera as a web viewer
Pixel on the film0.2 mm0.2 mm0.2 mm
Pixels across the web8,192 in one row4 x 2,448 = 9,792, with overlap2,448, one lane at a time
Trigger25,000 lines a second, from an encoder12.5 frames a second each, one every 400 mm from an encoder10 frames a second
Longest exposure40 microseconds40 microseconds, so strobes40 microseconds, so a strobe
Data204.8 million pixels a second250.7 million pixels a second50.1 million pixels a second
Share of the web imaged100%100%, with 3 seams25% at best

Example data. A 5 megapixel sensor here is 2,448 x 2,048 pixels; the area array's frames overlap so that nothing falls between them.

The encoder keeps every pixel the same length

The line rate above assumes the press always runs at 300 m/min. It does not. Presses ramp up after a splice, slow for a roll change and stop. If the camera fired at a fixed 25 kHz, a web running at half speed would be imaged twice as densely, and every feature would look twice as long. Fault sizes, positions on the roll and print register would all be wrong.

The fix is to trigger lines by distance, not by time. A quadrature encoder on a roller that the web wraps sends pulses as the web moves, and the camera takes one line per set number of pulses. The SFNC's Encoder Control chapter describes how cameras handle this:

  • The divider sets the distance per line. The EncoderDivider feature sets how many encoder increments are needed for one output pulse, which in practice is the distance between lines. Example data: a roller with a 500 mm circumference and an encoder giving 10,000 counts a revolution after decoding moves 500 / 10,000 = 0.05 mm per count. A divider of 4 gives one line every 0.2 mm. At 300 m/min the encoder runs at 5,000 / 0.05 = 100,000 counts a second, so check its rated frequency as well.
  • Reverse motion does not repeat the image. In the standard's Position output mode, pulses come only from forward increments, and the image looks as if the motion had only gone forward, even if the web briefly runs back. A web can move backwards briefly, for example when the press is jogged during threading.
  • A stopped encoder can raise an alarm. The encoder status can trigger an event when no pulses arrive for a set timeout. A slipping wheel or a broken cable then shows up as an alert rather than as a quietly stretched image.

Area cameras on a web use the encoder too. In the Example plant's area option, each camera fires one frame every 400 mm of travel, which keeps the overlap between frames at 9.6 mm whatever the speed. Two rules follow for both options. Mount the encoder on a roller the web wraps firmly, because a roller that slips reports a false distance. And put the encoder on the acceptance test list further down: image a printed scale at low and full speed, then compare the lengths.

Light: what a 40 microsecond exposure demands

EMVA 1288 gives the relationship that sets the lighting budget. The mean number of photons that hit a pixel depends on the pixel area, the irradiance on the sensor and the exposure time. In the standard's working units: photons = 50.34 x pixel area (µm²) x exposure (ms) x wavelength (µm) x irradiance (µW/cm²).

So for the same signal, a shorter exposure has to be paid for with brighter light or larger pixels. The table compares four cameras that each need 20,000 photons per pixel at 530 nm, green light. The pixel sizes and exposures are example data.

Camera and stationPixelExposurePhotons per µW/cm²Irradiance needed at the sensor
Line scan on the film web7 µm0.04 ms50.34 x 49 x 0.04 x 0.53 = 52.320,000 / 52.3 = 382 µW/cm²
Area camera on the same web3.45 µm0.04 ms50.34 x 11.9 x 0.04 x 0.53 = 12.71,575 µW/cm²
Area camera on the pouch conveyor3.45 µm0.2 ms63.5315 µW/cm²
Area camera, part held still3.45 µm2 ms635.131.5 µW/cm²

Three things stand out. The web forces a 40 microsecond exposure on any camera, line or area. At that exposure, the line scan camera's larger pixels (7 µm against 3.45 µm, about four times the area) need about a quarter of the irradiance the area camera needs. And a part held still for 2 ms needs about one fiftieth of the light of the area camera on the moving web.

In hardware terms, a line scan station uses a line light: a narrow, very bright strip focused on the one line the camera sees, which runs whenever the press runs. An area camera on a moving web needs a strobe that fires a short, intense flash over the whole frame, timed to the exposure. A part held still can use an ordinary ring, dome or back light. For mounting heights, angles and shielding from daylight, see camera placement and lighting on a production line.

When you compare two cameras for a web, ask for their EMVA 1288 data at your exposure. The standard's summary sheet reports quantum efficiency (the share of photons turned into signal), the absolute sensitivity threshold (the photons needed for a signal equal to the noise) and the saturation capacity. A camera with better figures at 40 microseconds can save the cost of a brighter light.

Data and cables: 205 million pixels a second

The data rate from the sizing step has to reach the processing PC without dropping lines. CoaXPress 2.1, the camera interface standard from the Japan Industrial Imaging Association (JIIA CXP-001-2021, revised 13 January 2021), defines connection speeds from CXP-1 at 1.25 Gbps to CXP-12 at 12.5 Gbps per coax cable, and a camera may use more than one cable.

  • The line scan camera at 1.64 Gbit/s fits one CXP-3 connection (3.125 Gbps), or one CXP-6 (6.25 Gbps) with room for a faster press later.
  • The area array sends 4 x 5.01 million pixels x 12.5 frames, or 250.7 million pixels a second: 22% more than the line scan camera, because the frames overlap. Split over four cameras that is about half a gigabit a second each, which fits a gigabit Ethernet link per camera.
  • Cable length depends on speed. The standard's example gives around 25 metres at 6.25 Gbps for thin or highly flexible cable, around 15 metres at 12.5 Gbps, and over 200 metres with thicker cable at 1.25 Gbps. Measure the cable route from the press to the cabinet before choosing the speed grade.
  • Line scan timing is tighter. The same standard treats a time stamp accuracy of 10 µs as the minimum for area scan applications and targets about 10 ns for line scan. Line triggers arrive tens of thousands of times a second, so timing errors an area camera would never notice show up as uneven lines.

Decision table: which camera for which station

Use one row per station. The cost column lists the hardware each option takes and its electricity for a year of 4,000 run hours, priced with Eurostat's electricity prices for medium-sized non-household consumers in the first half of 2026, excluding VAT and recoverable taxes: from €0.0928 per kWh in Finland to €0.2666 in Ireland, against an EU average of €0.1800 (Eurostat electricity price statistics). Power figures are example data. This guide found no independent price survey for camera hardware, so purchase prices are left to supplier quotes.

OptionGood atWhat it takes, and power a yearWho should pick itAsk the supplier
Area scan, part held still or indexedAssembly checks, labels, gauging on parts that stop1 camera, ring, dome or back light. Example 30 W: €11 to €32 a yearStations where the part stops or the machine indexesIs the camera data measured to EMVA 1288? How is the job switched at changeover?
Area scan with a strobe, parts moving one at a timePacks, bottles, pouches and cartons on a conveyor, one frame per part1 camera, strobe, photo-eye. Example 40 W: €15 to €43 a yearLines where each part can trigger its own frame and fits one field of viewWhat exposure at our conveyor speed, and how many pixels of blur?
Several area cameras on a webNarrow webs, or plants that want to stay on cameras and links they already run4 cameras, 4 strobes, encoder frame trigger, stitching software. Example 120 W: €45 to €128 a yearWebs up to one or two frames wideHow are frames joined at the seams, and what happens to a fault that sits on one?
Line scan with an encoder on a webFilm, foil, paper, nonwovens and print at full speed, with no seams1 camera per 8,192 pixels of width, line light, encoder, a fast link such as CoaXPress. Example 320 W: €119 to €341 a yearContinuous product, or product longer than one frameWhat maximum line rate at our pixel size? Where does the encoder go, and what alarm fires if it stops?
Line scan with rotationThe full surface of cans, tubes, rollers and bearings in one flat image1 camera, line light, a drive that turns the part a known amount. Example 100 W: €37 to €107 a yearStations where the whole surface matters and the part can be turnedHow is the rotation tied to the line trigger, and how many turns per part?
Spectral line scan (imaging spectrograph)Food and materials where the difference is in colour or chemistry beyond red, green and blue1 camera with a spectrograph, strong broadband light, heavy data. Example 400 W: €148 to €427 a yearProducts that pass at a steady pitch, as on the USDA poultry lineWhich wavelengths, at what line speed? How is the model tested on our product?

Read the verdict for each station from the row that fits:

  • Choose line scan with an encoder if the product is a continuous web or longer than one frame, and its speed can be measured at a roller.
  • Choose area scan with a strobe if parts arrive one at a time and each fits one field of view. A photo-eye then gives a clean trigger and one image per part.
  • Choose several area cameras on a web only if the web is narrow enough for one or two frames across. Beyond that the option multiplies cameras, strobes and seams.
  • Choose line scan with rotation if the whole surface of a round part must be seen without seams.
  • Choose a spectral line scan if the fault shows in colour or chemistry rather than shape, and the product passes at a steady pitch.

One kind of station is missing from the table on purpose: a manual packing or assembly table where the question is how much was made and when work stopped, not whether each unit is good. That is a counting job, not an inspection job, and none of the sizing above applies to it. Fabrico's AI cameras (an add-on) count output at stations where a machine has no usable signal. They are not an inspection system: they do not check units for faults, read codes, recognise stops or propose reasons. How counts become availability and performance is covered in OEE with computer vision, and code checks at the labeller in label and code vision vs an OEE camera.

Worked example: a printed film web and a pouch line

Example plant. Every figure in this section is example data except the value of the film, which comes from UN trade statistics. The plant extrudes and prints polyethylene film on a press 1,600 mm wide, rated at 300 m/min, and makes pouches from it on a separate line at 120 a minute. The quality manager wants a camera at both stations. Today the press has a strobed web viewer: one area camera on a traverse that shows the operator a still image of the print. The smallest fault the customer cares about is a 0.6 mm ink spot or hole.

Station 1: the printed web

The sizing is the one worked above: 0.2 mm pixels, 8,192 pixels across, 25,000 lines a second, a 40 microsecond exposure and 204.8 million pixels a second. Three options are on the table.

A. Keep the web viewer. One 5 megapixel camera at 0.2 mm sees 489.6 x 409.6 mm, or 0.2 m², per frame. At 10 frames a second it images 10 x 0.2005, about 2 m² of web a second, while the press makes 1.6 x 5 = 8 m² a second. At best a quarter of the web is ever imaged, and a person has to be looking at the screen while the fault is in view.

B. Four area cameras with strobes. Four of the same cameras side by side cover 4 x 489.6 = 1,958 mm, enough for the 1,600 mm web with about 120 mm of overlap at each of the 3 seams. The encoder fires a frame every 400 mm, 5,000 / 400 = 12.5 frames a second per camera. Full coverage, at the price of 4 strobes, 3 seams and 22% more data than option C.

C. One line scan camera with an encoder. One 8,192 pixel camera, a line light across the web and an encoder on a roller. Full coverage with no seams, one cable and one set of optics to align.

How much web each option imagesExample plant: 1.6 m web at 300 m/minA. Web viewer: 1 camera, 10 frames/s25%6 of every 8 m² a second unseenB. 4 area cameras with strobes100%3 seams, 4 strobes, 22% more dataC. 1 line scan camera, encoder100%No seams, 1 cable, 1 line lightFilm past a fault: 22.08 kg a minute
Example data. Each strip is 12 m of web; green is the share that a camera images.

B and C both give full coverage, and the engineering tips this web to C: one camera instead of four, no seams to manage and less data. The money question is A against either of them.

What the viewer's blind spot costs

The plant's stop log shows how long a print fault runs before the crew stops the press to clean the plate or fix the cause. With the viewer, the example puts that at 9 minutes on average. With a full coverage station that raises an alarm, it is 2 minutes: the time to see the alarm, stop and clean. The 7 minute difference is film printed with a known fault on it.

  • Film made per minute: 1.6 m x 300 m = 480 m². At 46 g per m², a 50 µm film, that is 480 x 46 = 22,080 g, or 22.08 kg a minute.
  • Per fault: 22.08 x 7 = 154.56 kg.
  • Per year, at 3 such faults a week over 46 production weeks, 138 in all: 154.56 x 138 = 21,329 kg.

The value per kilogram comes from UN Comtrade, the United Nations trade database, for Germany's 2024 trade in polyethylene film and sheet (HS 3920.10). Exports were $2,200.0 million for 649.3 million kg and imports $1,163.7 million for 426.5 million kg (net weights estimated by UN Comtrade), or $3.39 and $2.73 a kilogram (our arithmetic). These are values of the film alone, before print, ink, press time and any customer claim, so they make a floor.

21,329 x 2.73 = $58,228 a year at the import value, and 21,329 x 3.39 = $72,305 a year at the export value. Per fault that is $422 to $524 of film, and per minute of blind running, $60 to $75.

The baseline is realistic on purpose: a press with a viewer and an attentive crew, not a press with no camera at all. The range also leaves out faults the viewer never shows and the customer finds, which this example does not price.

On an OEE chart, the 21,329 kg is quality loss and the minutes to stop and clean are availability loss. Once press stops are logged with a time and a reason, the 9 and 2 minute figures become measured numbers rather than estimates. For the losses to track on the extrusion side, see OEE for blown film extrusion, and for where inline checks sit against checks at the rewinder or in the lab, inline vs offline inspection.

Station 2: the pouches

The pouch line is a different kind of station. Pouches leave the machine one at a time at 120 a minute, 2 a second, on a conveyor running at 500 mm a second. Each pouch is 200 x 150 mm, and the faults on the list (a seal, a fold, the print position) need 0.1 mm pixels. One 5 megapixel area camera at 0.1 mm sees 244.8 x 204.8 mm, so one frame holds one pouch, and a photo-eye triggers it. The longest exposure without more than a pixel of blur is 0.1 / 500 = 0.0002 seconds, 200 microseconds: five times the web station's 40, so a compact strobe is enough.

A line scan camera could image the pouches, but it would need an encoder on the conveyor, a steady speed and software to cut a continuous image back into pouches. Area scan wins this station for the same reason line scan wins the web: it matches how the product arrives.

Book a demo and see losses per line and shift.

What each option costs to run

Running cost has three parts: electricity, upkeep hours and what the station misses. Electricity is priced as in the decision table. Upkeep is priced with Eurostat's hourly labour costs for 2025: €34.9 in the EU and €40.3 in euro area industry (Eurostat hourly labour costs). Power and hours are example data for the web station.

OptionPowerElectricity a yearUpkeep hours a yearUpkeep a yearRunning cost a yearWeb imaged
A. Web viewer40 W160 kWh: €15 to €436: lens cleaning, traverse€209 to €242€224 to €28425% at best
B. Four area cameras120 W480 kWh: €45 to €12824: seam checks, strobe timing, alignment€838 to €967€882 to €1,095100%
C. Line scan320 W1,280 kWh: €119 to €34112: encoder check, flat field, alignment€419 to €484€538 to €825100%

Example data; the low end uses Finland's electricity price and the EU labour cost, the high end Ireland's price and euro area industry. Electricity is not what decides. The line scan station draws the most power, because its line light runs whenever the press runs, and still costs under €350 a year at the highest price in the EU. Upkeep is a few hundred euros for any option. The film at stake between A and the full coverage options, $58,228 to $72,305 a year in this example, is dozens of times the running cost of B or C.

Two running costs are harder to price. A line light can lose output as it ages, so a station calibrated on a new light reads darker months later, and the flat field correction has to be repeated. EMVA 1288 reports photo response non-uniformity, so you can compare how evenly two cameras respond before you buy. And an encoder wheel wears; a worn wheel changes the distance per count, which the stretch test below catches.

Acceptance tests before you sign off the station

A station that passes a demonstration on the supplier's sample can still fail on the line. Run them on the running press, not on a sample. Write them into the purchase order.

TestHow to run itPassA fail points to
StretchImage a printed scale at a third, two thirds and all of top speedThe same length in pixels at each speed, within 1% (our pass mark)No encoder trigger, a slipping roller or the wrong divider
Reverse jogJog the web back 200 mm and forward again during a scanNo repeated band in the imageEncoder output counting both directions
Encoder lossUnplug the encoder during a runAn alarm within the set timeoutNo encoder status event configured
Top speed blurRun at rated speed with the planned exposureNo blur wider than one pixel on a sharp print edgeExposure too long for the speed, or strobe timing off
Edge to edgePass the same fault sample at each edge and in the middleThe same contrast across the webUneven light, no flat field correction, lens fall-off
Seams (area array only)Pass a fault sample across each seamFound once, not twice and not zero timesGaps or double counting in the stitching
Data at top speedRun at full speed for an hourNo dropped lines or frames in the logInterface, cable or PC at its limit
Your own faultsRun real faults collected from your productionEach size class found, false alarms countedPixel too coarse or the wrong light angle

Run the stretch test again after any work on the web path. If a roller is changed and the encoder moves with it, the distance per line changes too.

Next steps for this week and this quarter

This week

  1. Plant or production manager: list each station where a camera is planned, and write next to it how the product arrives: a web, parts one at a time, items on a steady line, or round parts.
  2. Quality manager: take the smallest fault a customer has complained about, measure it and divide by 3. That is the pixel size to put in the request.
  3. CI manager: from the stop log, find how long the last five print or coating faults ran before the line stopped. That number prices the blind spot.

This quarter

  1. Size each web station with the four numbers at rated top speed, and send them with the request so that every quote answers the same question.
  2. Specify the encoder, its roller and its alarm in the request, not as an extra.
  3. Ask for EMVA 1288 data for each camera offered, area or line, and compare sensitivity and saturation at your exposure.
  4. Put the acceptance tests in the purchase order, with the press speeds they run at.
  5. Measure the blind spot. Log press stops with a reason and a time, so the minutes a fault runs are on record. A light MES ties those stops and the scrap to orders and products, line by line and shift by shift.

Where Fabrico fits

Fabrico is a cloud OEE and light MES platform: OEE and losses by machine, line and shift, from the PLC or a Fabrico IoT device, with reasons operators add on a tablet or phone.

On manual stations, AI cameras (an add-on) count output.

Book a demo and see OEE per line and shift.

What people ask next

Is a line scan camera better than an area scan camera?

Neither is better in general. A line scan camera suits product that moves continuously past it: webs of film, paper, foil or nonwovens, long parts, and round parts turned in front of it. An area scan camera suits parts that arrive one at a time and fit one field of view, such as packs, bottles and pouches. Start from how the product reaches the camera.

How do I calculate the line rate for a line scan camera?

Divide the web speed by the pixel size along travel. At 300 m/min the web moves 5,000 mm a second; at 0.2 mm per pixel that is 5,000 / 0.2 = 25,000 lines a second. Size it at the line's rated top speed. The longest possible exposure is then 1 / 25,000 of a second, or 40 microseconds.

Does a line scan camera need an encoder?

On a line whose speed changes, yes. An encoder on a roller triggers one line per fixed distance, so a pixel stays the same length when the line ramps up, slows or stops. In the GenICam naming standard this is the encoder divider: the number of encoder counts per line, for example 4 counts of 0.05 mm for a 0.2 mm line.

When are several area scan cameras better than one line scan camera on a web?

When the web is narrow enough for one or two frames across, or when the plant wants to stay on cameras and links it already runs. On a 1.6 m web at 300 m/min, the example in this guide needs 4 area cameras with strobes, 3 seams and 22% more data than one 8,192 pixel line scan camera.

Why does a line scan camera need so much light?

Each line can only be exposed for the time between lines, 40 microseconds at 25,000 lines a second. EMVA 1288 shows that the photons that hit a pixel scale with pixel area, exposure time and irradiance, so a short exposure must be paid for with brighter light or larger pixels. An area camera on the same moving web faces the same 40 microsecond limit.

What does a camera station on a web cost to run?

Less than the faults it misses. In the example, electricity is €15 to €341 a year depending on the option and the country, at Eurostat prices for 2026, and upkeep is a few hundred euros. A viewer that lets a print fault run 7 minutes longer, 3 times a week, puts $58,228 to $72,305 of film a year at stake.

Sources

Sources accessed on 16 November 2026 unless marked as accessed on another date. Dates are the publication dates shown on each page.

Track OEE and downtime in real time
See where lines lose time
Book a live demo.
See where lines lose time
See where lines lose time

Real Fabrico tiles, sample data.

By clicking the Accept button, you are giving your consent to the use of cookies when accessing this website and utilizing our services. To learn more about how cookies are used and managed, please refer to our Privacy Policy and Cookies Declaration