Camera sensor size formats explained: what 1 inch means

Every camera datasheet leads with a sensor format such as 1/1.8" or 2/3", and this is the one specification that routinely misleads, because the label is not a dimension. It is a 1950s television-tube convention in which a "1 inch" sensor carries a 16 mm diagonal. This explainer converts every common format into real millimetres, then follows the number where it leads: into your field of view, your lens's image circle and your pixel pitch, with the arithmetic shown for each. By the end you can pair a camera and lens with confidence, whether you are browsing area scan cameras, building with board-level modules, or matching glass from the machine vision lenses range.

Why is a 1 inch sensor only 16 mm across?

Because the inch label measures the outer diameter of a 1950s television camera tube, not the sensor: that tube's usable image area was only about two-thirds of its diameter on the diagonal, and digital sensors inherited the number, so the labelled inch became the name and the millimetre diagonal became the reality. Every fractional format since divides that same nominal inch. Read "1/1.8 inch" as one inch divided by 1.8, and read every such label as a diagonal class, nothing more.

Two consequences follow. First, a format tells you nothing about resolution: the same 2/3" outline ships as low-resolution sensors and as multi-megapixel ones, because pixel pitch is chosen independently of the silicon rectangle. Second, the label is a class, not a drawing. Sensors sharing a format differ by a few percent in active area and sometimes in aspect ratio, so the width, height and diagonal printed in the datasheet are the numbers your optics actually see.

How big is each sensor format in millimetres?

The largest tube format's active area is about 12.7 x 9.6 mm, 2/3 inch is 8.8 x 6.6 mm, and 1/2 inch is 6.4 x 4.8 mm, all as 4:3 rectangles with the diagonals in the table below. The table gives the classic outlines each format is built around. For anything not listed, a shortcut lands close: divide the nominal 16 mm diagonal by the format's denominator.

So 1/1.8" gives 16 ÷ 1.8 = 8.9 mm, and 1/1.7" gives 16 ÷ 1.7 = 9.4 mm. The shortcut holds to within about a millimetre from 1" down to 1/2". Below that it reads low: current 1/2.8" sensors measure about 6.4 mm on the diagonal where the division predicts 5.7 mm, so for the smallest formats go straight to the active-area drawing.

Above the full inch the industry drops tube fractions for film-era names such as APS-C and APS-H, or plain millimetres, with diagonals running from roughly 22 mm upward.

Why does sensor format set your field of view?

Because a lens draws its image at a scale fixed by focal length and working distance, so your field of view is the sensor's own width and height divided by that scale: twice the silicon, twice the view. The lens paints at one magnification; the sensor decides how much of the painting you keep.

Now keep the 35 mm lens and swap in a 2/3" camera. At 310 mm the view drops to 8.8 ÷ 0.127 = 69 mm wide, and the part no longer fits. Reframing it costs distance: the required scale is 8.8 ÷ 100 = 0.088x, so the working distance becomes 35 x 1.088 ÷ 0.088 = 433 mm. Same lens, same scene, 40 percent more floor space, purely from a smaller sensor.

The blister inspection field of view formula works the same sums from the part side, and the pixel rule for camera resolution picks up where the 0.025 mm figure leaves off.

What lens image circle does my sensor need?

One at least as large as your sensor's diagonal, taken from the table above: the lens's image circle must cover it, or the corners of every image go dark. A lens's format label uses the same tube convention, so each lens is engineered around the circle of its own class. Fall short and the image fades to black in the corners, and you cannot fix vignetting in software without cropping away the edge field you paid for. Running a larger lens on a smaller sensor is free: it simply uses the well-corrected centre of a bigger image.

The lens finder has an image-circle filter for exactly this, in bands from "up to 8 mm" through "16-22 mm" and beyond; match the band to your sensor's diagonal from the table above.

Three C-mount lenses show the pattern: as the rated format steps from 2/3" through 1" to 1.1", the image circle grows with it:

All 3 share: Mount C-mount. Values as printed in each product's data sheet; open a part for its drawing and full specifications.

The LS-FFL-0679 is a 35 mm, 2/3", 6 MP design rated for an 8-11 mm image circle. The largest format in that compare needs a bigger circle, which leaves this glass 5 mm short in every corner; the 1" and 1.1" lenses beside it are the ones that cover it. Around 1.1" the C-mount thread runs out of glass, and larger formats move to F-mount, M58 and M72 barrels.

The lens half of this convention, which glass to buy for 1/1.8 inch, 2/3 inch and 1 inch sensors, is covered in lens sensor format explained.

Do smaller pixels give you more detail?

They give more detail only while the lens resolves it and the light supports it: halving pixel pitch quadruples the pixel count available on a format, but cuts each pixel's light-collecting area to a quarter and halves best-case signal-to-noise. Format fixes the rectangle; resolution decides how you cut it. The pitch that falls out is what lens and lighting must support:

The LS-BCM-0056 is a 1/2" GigE board camera with 12 MP on 1.6 um x 1.6 um pixels. The arithmetic closes exactly: 12,000,000 pixels x 2.56 um² = 30.7 mm², and the classic 1/2" rectangle of 6.4 x 4.8 mm is 30.7 mm² around an 8 mm diagonal.

Put the same 12 MP on a 1" rectangle, 12.7 x 9.6 mm = 121.9 mm², and pitch becomes √(121.9 ÷ 12) = 3.2 um. Each pixel then holds 3.2² ÷ 1.6² = 4 times the collecting area, and because shot-noise-limited signal-to-noise is the square root of the electrons collected, best-case SNR doubles. That is the case for large formats in one line of arithmetic.

The MV-CS200-10UM packs 20 MP onto a 1" sensor. Twenty million pixels over 121.9 mm² is 6.1 um² each, or 2.5 um pitch, and a 2.5 um pixel asks the lens for 1000 ÷ (2 x 2.5) = 200 lp/mm at the Nyquist limit. Glass that resolves 100 lp/mm serves a 5 um-pixel camera happily and visibly softens on this one.

The decision is about light, not detail. With a strobe and microseconds to spare, take the small pixels; the resolution is real. Under dim continuous light, choose fewer megapixels on the same format, or the same megapixels on a bigger one, and keep signal over speckle. The trade-offs in small pixel camera selection extend this to dynamic range and lens cost.

Which sensor format should you choose?

Take 1 inch or larger for fine detail and dim light at short range, 2/3 or 1/1.8 inch for everyday inspection where cameras and lenses stay small and plentiful, and 1/2 inch and below for embedded boards where the optics must shrink. Each case has a reason:

Fine detail, dim light, room on the bench: 1" and up. Large silicon at moderate pixel counts gives 3-4 um pixels, wide views at short range, and full use of the format's image circle. Around 20 MP on a 1" sensor is the workhorse combination for electronics and print inspection.

Everyday factory inspection: 2/3" or 1/1.8". Lenses are plentiful and compact, the camera finder is deepest here, and at 5-12 MP the pitches land where standard 6-10 MP-rated glass still resolves cleanly.

Embedded and space-starved builds: 1/2" and below. Board-level cameras in these formats take M12 and C-mount optics a few millimetres across; the M12 lens selection guide covers the threads, back-focal length and IR-cut decisions those builds raise.

Working distance fixed, view must widen: step the format up, not the focal length down. A wider sensor widens the frame with the same lens, keeps magnification low so depth of field stays usable, and avoids the corner distortion that very short focal lengths bring. One counterweight: image circle is the expensive property in a lens, so step up for an optical reason, not by habit.

When is the format label the wrong number to use?

When two sensors share a label but not the same outline: a 16:9 "1 inch" sensor measures 13.9 x 7.8 mm where a 4:3 one measures 12.7 x 9.6 mm, a 9 percent width difference that directly changes your field of view. Four more traps sit beside it:

• Aspect ratio hides inside the label. Two cameras of the same class frame the same scene 9 percent differently across the width; software that assumes a fixed field of view per model will not get one.
• Small formats break the shortcut. Below 1/2", dividing 16 mm by the denominator reads low; the 1/2.8" case above is 0.7 mm out on a 6 mm part, so use the active-area drawing there.
• Resolution is not in the label. A format alone cannot separate a 2 MP sensor from a 12 MP one; always carry the pixel count and pitch alongside it.
• Lens megapixel ratings assume a format. A lens sold as "5 MP" was rated on one sensor size; the honest currency is line pairs per millimetre against your pitch, 1000 ÷ (2 x pitch in um), so 3.2 um pixels ask for about 156 lp/mm.
• Past C-mount the labels change character. A 25 MP APS-H camera such as the MV-CH250-25TC-F-NF ships with an F-mount, and its lenses are specified by image circle in millimetres, not by tube fractions.

What should you check before ordering?

Five numbers, and the format label is not among them: the sensor's width and height in millimetres, its diagonal, your lens's image circle, and the lens's resolving power stated against your pixel pitch. Work down this list before anything goes in a cart:

• The sensor's active area, width x height in mm, and its aspect ratio, taken from the sensor's own drawing rather than the format class.
• The diagonal in mm, and a lens image circle at or above it, picked with the image-circle filter before the focal length.
• Pixel pitch from √(area ÷ pixel count), and a lens resolving at least 1000 ÷ (2 x pitch) lp/mm at your working aperture.
• Focal length and working distance re-run with the lens's actual focal length; the nearest standard step, 33.8 mm to 35 mm in the example above, moved the working distance from 300 to 310 mm.
• Mount mechanics, C versus CS flange and adapter clearance, which the C mount versus CS mount guide untangles before you cut an adapter.

Then run the pairing through the configurator so camera, lens, light and controller are checked in one pass:

If the combination is unusual, large formats, long distances or odd filters, bring the five numbers above to the contact page and we will check the pairing with you.