Camera resolution for a 1 meter field of view, 2 m standoff

You need to image a 1 m square area at about 0.5 mm per pixel, from a camera standing roughly 2 m off the part — and the real job is metrology, because a 5 mm wide, meter-long feature has to be measured, not just photographed. Those four numbers fix everything downstream: the pixel count, the sensor format, the focal length and the accuracy you can honestly claim. This guide runs the arithmetic end to end, names camera and lens pairings that come out of it, and lists the checks that keep a 0.05 mm measurement from drifting into a 0.3 mm one. The part is static and the lighting is controllable, so the optics are the easy half; coverage, calibration and perspective decide whether this gauge works.

How many megapixels does a meter-wide field at 0.5 mm per pixel need?

Four megapixels is the exact floor: 1000 mm ÷ 0.5 mm per pixel = 2000 pixels per axis, and 2000 × 2000 = 4 MP. Order above the floor, because a mounted camera loses border pixels to lens coverage and alignment, and measurement eats margin faster than detection does.

The number that sizes this system is not the 0.5 mm figure — it is the tolerance you want on the 5 mm width. Detection needs three or more pixels across a feature; at 0.49 mm/px the width already spans ten, so seeing it is easy. Measuring is a separate budget: edge repeatability of 1/5 to 1/10 of a pixel converts pixel scale into accuracy, so ±0.05 mm asks for 0.25 mm/px or finer, while ±0.25 mm sits comfortably at 0.49 mm/px. Work the pixel count in the camera resolution calculator, then filter the area scan cameras finder by resolution class and sensor format.

Can a 5 MP camera measure a 5 mm wide feature?

Yes — about ten pixels span the feature, and sub-pixel edge finding locates each edge to 0.05-0.10 mm, so the width repeats to about ±0.1 mm. That serves a ±0.25 mm tolerance with margin and comes up short at ±0.05 mm; only pixel count and edge contrast move that line.

The common 5 MP layout is 2448 × 2048 pixels, and the short axis carries the meter: 1000 mm ÷ 2048 px = 0.49 mm/px, with the long axis at 0.41 mm/px framing 1.19 m. Ten pixels across the 5 mm feature is enough for a fitted edge, provided the edge spread covers two to three pixels and the contrast is high — a clean, symmetric grey-level transition locates to 1/10 of a pixel, a noisy one to 1/5 or worse. Two habits buy that margin indoors: shut out ambient light with a shroud or bandpass filtering, and average frames, because the part is static — 16 frames cut random noise by a factor of four.

What focal length covers a meter-wide field from about 2 m?

Between 11 and 17 mm for sensors up to 1", and about 43 mm for a 25 MP APS-H square sensor, from f = WD × m ÷ (1 + m) with m = sensor short side ÷ field. The short side, not the pixel count, sets the lens.

Put the meter on the sensor's short axis and the table below falls out. Standard focal lengths come in steps, so you land a little off 2 m rather than exactly on it, and standoff tolerance trades one-for-one into field size: ±100 mm at 2 m is ±5% of field. The geometry is the same as in lens field of view: why sensor size and focal length set it, and machine vision lens selection: focal length calculation works the same formula from the lens side.

Run your own sensor-and-standoff pair through the lens focal length calculator before committing to a lens.

Which camera and lens should I order?

A 12 MP camera with a 12 mm lens at 2.2 m covers the meter at 0.33 mm/px; a 20 MP camera with a 16 mm lens narrows the edge error to about 0.03 mm; a 25 MP square sensor gives a true square field at 0.20 mm/px.

Worked example: 12 MP, 12 mm, 2.2 m. The LS-CMR-0166 is a 12 MP USB3 camera with 4000 × 3000 pixels on a 1/1.7" sensor, rolling shutter — harmless on a static part. A 1/1.7" array of that size measures about 7.4 × 5.5 mm with 1.85 µm pixels, and the arithmetic runs:

• Field: the meter across the 3000-pixel axis → 1000 mm ÷ 3000 px = 0.33 mm/px; the 4000-pixel axis frames 1330 mm.
• Feature: 5 mm ÷ 0.33 mm/px = 15 pixels across the width.
• Magnification: m = 5.5 mm ÷ 1000 mm = 0.0055.
• Focal length: f = 2000 mm × 0.0055 ÷ 1.0055 = 10.9 mm → buy the 12 mm focal length.
• Standoff with 12 mm: WD = 12 mm × 1.0055 ÷ 0.0055 = 2194 mm ≈ 2.2 m.
• Edge error: 0.33 mm/px ÷ 10 = 0.033 mm best case, ÷ 5 = 0.067 mm typical.

For tighter tolerances go mono and bigger: the MV-CS200-10UM is a 20 MP mono camera, 1" sensor, C-mount, USB3. That sensor class runs 5472 × 3648 pixels at 2.4 µm — 13.1 × 8.8 mm — so the meter sits on the 3648 axis at 0.27 mm/px while the long axis frames 1.5 m; a 16 mm lens frames 1.09 m vertically at 2.0 m. The matching glass is the LS-FFL-0567: 16 mm focal length, C-mount, rated 20 MP on a 1.1" format, F2.8-22, distortion under 0.3%. A 1.1" rating covers the 15.9 mm diagonal of a 1" sensor with margin.

The symmetric case wants a square die: the MV-CH250-25TC-F-NF is a 25 MP camera on an APS-H sensor, F-mount, 10GigE. A 25 MP APS-H array is 5120 × 5120 pixels at 4.25 µm — 21.8 mm square — so the field is a true square meter at 0.20 mm/px, m = 0.0218, and a 50 mm lens stands off at 2.34 m. This model is a color camera, and a Bayer pattern costs single-channel edge detail — read color camera spatial resolution: beyond the pixel count before you gauge with one.

Whatever sensor you pick, the lens must keep up in line pairs per millimetre: 1.85 µm pixels sample at 270 lp/mm, 2.4 µm at 208, 4.25 µm at 118. Lens selection based on resolution: feature size to MTF shows how to read that against a lens curve.

When does a square meter field waste pixels?

When the feature is a meter-long line 5 mm wide, aim the sensor's long axis along it: 4000 pixels on the meter give 0.25 mm/px and the 3000-pixel axis still covers 750 mm of width. Keep the feature inside that band and the extra resolution is free.

Splitting the field across two cameras is the other lever: each camera sees half the meter at a finer pixel scale, at the price of an overlap region and a calibrated transform between views. It pays when the budget forces small sensors — but a continuous feature that crosses the seam collects the stitching error directly. How to locate a large part when the camera field of view is too small covers the geometry of that trade.

How accurate will the measurement really be?

About ±0.1 mm with the 12 MP set-up on a flat part: 0.03-0.07 mm from sub-pixel edges, about 0.02 mm from perspective across ±10 mm of height, and 0.01-0.02 mm of local distortion residual after a full-field calibration.

The budget, item by item:

• Edge repeatability: 0.033 mm at 1/10 pixel, 0.067 mm at 1/5. Contrast and a 2-3 pixel edge spread decide which end you get.
• Perspective: width error = feature width × height error ÷ standoff = 5 mm × 10 mm ÷ 2194 mm = 0.023 mm at ±10 mm of part height, 0.11 mm at ±50 mm. Flatness is an accuracy input, not a mounting detail.
• Distortion: a 0.3% lens over the full field moves raw positions by up to 3 mm at the corners. A polynomial correction fitted on a grid that spans the field leaves roughly 0.1-0.3 mm of absolute position error and 0.01-0.02 mm of local width error. Geometric vs TV distortion: read lens datasheets for measurement explains which number a data sheet quotes.
• Motion and blur: one pixel of blur equals the whole pixel scale — 0.33 mm here. Nothing moves in this build; lock the focus and the mount, and if presentation ever moves, camera resolution for measurement accuracy in motion sets the exposure ceiling.

Do I need a telecentric lens for a meter-wide field?

No. An object-side telecentric lens needs a front aperture at least as large as its field, so a meter-wide field means more than a meter of front glass — at 2 m standoff, conventional optics plus calibration is the correct metrology choice.

Telecentricity buys freedom from the perspective term above, and that term is already 0.02-0.1 mm here, small against the edge term. Telecentrics earn their cost on fields of a few tens of millimetres, where the front element stays buildable; the telecentric lenses in the shop serve exactly that regime, and entocentric vs telecentric lenses: when to use each draws the line. For contrast, telecentric lens field of view calculation from magnification shows what magnification-based sizing would demand of a large field. If part height varies by tens of millimetres, calibrate the scale per height instead of chasing a meter-wide telecentric.

What f-number and exposure should I run at 2 m?

Stay between f/2.8 and f/5.6: the Airy disk reaches two pixels at N ≈ 1.5 × pixel pitch in µm — f/2.8 on 1.85 µm pixels, about f/3.6 on 2.4 µm, f/5.2 on 3.45 µm. Stopping down past that spends the resolution you paid megapixels for.

Airy disk size and the diffraction limit: match pixel to feature derives that limit, and aperture f-number and depth of field in machine vision cameras covers the aperture side. Depth of field is not a constraint at this magnification:

That result uses c set to two pixels (3.7 µm) on the 12 MP example at f/2.8 — the whole part height sits inside focus, so run the aperture near wide open for sharpness, not stopped down for depth. Exposure is equally free on a static target: 10 ms class integrations under diffused LED panels, gain low, histogram mid-scale. One discipline matters under mains-powered light.

What should I check before ordering?

Five checks stand between this arithmetic and a working gauge: lens image circle, lens resolution class, mount, the focal-length arithmetic at the standoff you can actually hold, and a calibration target that fills the field.

• Image circle against sensor diagonal: 15.9 mm for 1", 17.6 mm for 1.1", 30.8 mm for the 21.8 mm square APS-H die. A 2/3" lens with an 11 mm circle vignettes the corners black; C-mount optics reach about 1.1", and APS-H wants F-mount or M58.
• Lens resolution against sensor Nyquist: 270 lp/mm at 1.85 µm pixels, 208 at 2.4 µm, 145 at 3.45 µm, 118 at 4.25 µm. Match them with matching lens resolution to pixel size: the Nyquist budget method.
• Standoff arithmetic with the real parts: redo the focal-length sums with the lens you chose and add 5% sensor margin; the camera finder and lens finder list both sides by the filter values used here.
• Calibration target coverage: the grid must fill the meter, or the distortion fit extrapolates past its data — calibration target field of view and lens coverage margins sizes it.
• Averaging and strobe plan: settle the frame count and whether the light strobes before you order the controller.

Next step: feed your field, standoff and tolerance into the configurator, which walks camera, lens and light in order, or browse the machine vision lenses for the focal length class your sensor needs. More sizing guides sit under the cameras topic.