Learn / Sensors
Why a 1/2.8″ sensor is not 1/2.8 inch
Where optical-format names come from, how to estimate the real sensor size, and what to use instead.
Sensor datasheets describe size with names like 1/2.8″, 1/1.8″ or 2/3″. It is natural to read these as inches and expect a 1/2.8″ sensor to be about 9.1 mm across. It is not: a typical 1/2.8″ sensor has an active diagonal of about 6.4 mm, roughly 70% of the name. Here is where the names come from, and what to use instead when you choose a lens.
The names come from camera tubes
Before solid-state sensors, video cameras used vidicon tubes. A tube was named by the outside diameter of its glass envelope: a "1-inch" tube was 25.4 mm across. The light-sensitive area inside was much smaller, with a diagonal of about 16 mm, roughly two thirds of the tube diameter.
When CCD and CMOS sensors replaced tubes, manufacturers kept the old names so a sensor would match the lenses designed for the tube it replaced. A sensor called "2/3-inch" has about the same image area as the old 2/3-inch tube, even though nothing on the chip measures 2/3 of an inch.
A rule of thumb that works
Because the names are historical, there is no exact formula, but two rules of thumb predict the active diagonal well:
- For formats of 1/2″ and larger: about 16 mm of diagonal per inch of the name. A 2/3″ sensor has about 16 × 2/3 ≈ 11 mm.
- For formats smaller than 1/2″: about 18 mm per inch. A 1/2.8″ sensor has about 18 / 2.8 ≈ 6.4 mm.
Checked against real sensors, these rules land within a few percent of the datasheet. Some examples, calculated from resolution and pixel pitch:
| Sensor | Pixels | Pitch | Active diagonal | Sold as |
|---|---|---|---|---|
| Sony IMX307 | 1920 × 1080 | 2.9 µm | 6.39 mm | 1/2.8″ |
| Sony IMX335 | 2592 × 1944 | 2.0 µm | 6.48 mm | 1/2.8″ |
| Sony IMX477 | 4056 × 3040 | 1.55 µm | 7.86 mm | 1/2.3″ |
| Sony IMX334 | 3840 × 2160 | 2.0 µm | 8.81 mm | 1/1.8″ |
| Sony IMX264 | 2448 × 2048 | 3.45 µm | 11.01 mm | 2/3″ |
Same name, different shape
The format name fixes the diagonal, not the shape. Older reference tables list 4:3 dimensions, but many security and automotive sensors are 16:9. The IMX307 is a 1/2.8″ sensor at 5.57 × 3.13 mm, while a 4:3 sensor of the same type would be about 5.14 × 3.86 mm. Both have nearly the same diagonal, but the 16:9 sensor is 8% wider and 19% shorter, so its horizontal field of view is wider and its vertical field narrower with the same lens.
What to use instead
For any calculation that matters, compute the active area directly:
width = horizontal pixels × pixel pitch, height = vertical pixels × pixel pitch, diagonal = √(width² + height²)
Then check that the lens's image circle covers that diagonal. A lens specified "for 1/2.8″" usually has an image circle of about 6.4 mm or a little more; put it on a slightly larger sensor and the corners go dark. Field of view, depth of field and pixel density should all be calculated from the real width and height, never from the format name.
Convert any format or sensor to its real size, crop factor and Nyquist frequency.
Sensor format calculator →Key points
- Optical format names come from vidicon tube diameters and are not real dimensions.
- Expect about 16 mm of diagonal per inch from 1/2″ up, and about 18 mm below that.
- The name fixes the diagonal; 4:3 and 16:9 sensors of the same type have different widths.
- Use pixels × pitch for exact size, and match the lens image circle to the real diagonal.