Sensor Size Matters - Part 1

Posted by

We get a lot of questions about sensor sizes and crop factors.  Most people know the difference between a standard (APS-C) crop sensor and a full-frame sensor. Not many, though, know how much smaller a 2/3” sensor is than a 4/3 sensor, and fewer still the difference between those and a 1/1.8” sensor. Plus Canon and Nikon have thrown new sensor sizes into the mix in the last year and a lot of people aren't sure exactly where those sensors fit in among the better known ones. Hardly a day goes by that someone doesn't ask if the Fuji X-10 sensor is bigger than the Nikon J1 sensor. Is the Canon GX-1 sensor as big as those or more like a point and shoot?

The problem is even more complicated now that SLR lenses are being used on video cameras and video lenses on SLRs cameras. People want to know things like "is Super 35mm format equivalent to a crop sensor or full-frame?" Other people, trying to sell their 16mm film lenses with adapters for 4/3 cameras fail to mention how much smaller 16mm film was than 4/3 sensors. Not to mention the marketers, always ready to make things as confusing as possible, are doing things like calling a 1/1.7” sensor “large” (it is compared to a cell phone camera chip, I guess).

Anyway, since I haven’t been able to find a single source to answer all these sensor format questions, I thought I’d put it all together here. The table below shows the dimensions, in millimeters, of the various sensor (or film) sizes. Please note that the dimensions may vary slightly from camera to camera. For example, Canon's APS-C sensor is slightly smaller than Nikon's, but slightly larger than Sigma's. The aspect ratio of the sensor (4:3, 3:2, 16:9) will cause some variation, too. For example, the 35mm Cinema, Super 35mm, and APS-C crop sensor formats are nearly the same size (look at the sensor area) but of slightly different rectangular proportions.

Format height width Diag Area Crop Examples
mm mm mm (mm2) factor
Medium Format 44.0 33.0 55.0 1452 0.7 Pentax 645
Full-frame 24.0 36.0 43.4 864 1.0 FF SLRs
Red Epic 14.6 27.7 31.3 404 1.3 Red Epic/Scarlet
35 Cine 13.7 24.4 28 334 1.4 Red One
Super 35mm 13.8 24.6 28.0 339 1.4 Canon C300
APS-C crop** 15.0 22.0 27.3 329 1.5 crop SLRs
1.5" 14.0 18.7 23.4 262 1.9 Canon G1X
4/3 13.5 18.0 22.4 243 2.0 Four-thirds
Nikon CX 8.8 13.2 15.8 116 2.7 Nikon J1/V1
Super 16 7.4 12.5 14.5 93 3.0 film only
2/3" 6.6 8.8 11.0 58 4.0 Fuji X-10; camcorders
1/1.7" 5.6 7.4 9.5 42 4.6 Best P&S
1/1.8" 5.3 7.2 8.9 38 4.8 Best P&S
1/2" 4.8 6.4 8.0 31 5.4 camcorders
1/2.5" 4.3 5.8 7.2 25 6.0 P&S
1/2.7" 4.0 5.4 6.7 21 6.4 P&S
1/3" 3.6 4.8 6.0 17 7.2 camcorders

First, About those inch (") sensors

Commonly used sensor abbreviations make absolutely no sense. (Get it, "sense", "sensor" - I have to have at least one pun per article. It's in my contract). Larger sensors are measured in millimeters: full-frame, Super 35mm, APS-C, etc. The 4/3 marketing people probably thought "half as big as full frame" wasn't a good way to present things, so 4/3 it was. But it's easy to find how big a 4/3 sensor is in mm.

But then we get into all of these fractional-inch-type-measurements for the smaller sensors. That measurement system originated in ancient times (the 1950s to 1980s) when vacuum tubes were used instead of CCD or CMOS sensors in video and television cameras. The image sensor was, in those days, referred to in terms of the outside diameter of the vacuum tube that contained it.


A video camera tube (courtesy Wikepedia Commons)

Why do manufacturers keep using such an archaic measurement? Because it helps them lie to you, of course. If you do the math 1/2.7 equals 0.37 inches, which equals 9.39 mm. But if you look at the chart above you'll see that a 1/2.7" sensor actually has a diagonal of 6.7 mm. Why? Because, of course, a thick glass tube used to surround the sensors. So they calculate the sensor size as if the glass tube was still included. Makes perfect sense to a marketing person who wants to make their sensor seem larger than it is. What sounds better: 1/2.7" or 'less than 10% the size of a full frame sensor'?

Calculating the Crop Factor from the Sensor Diagonal

It surprises me how many people do not really understand what the "crop factor" is, and technical explanations seem to make it worse for newcomers. But in simplest terms if I set up several cameras with different size sensors at point A, put the same 100mm lens on each of them and took a picture the picture taken with the smaller sensors would seem more magnified than the pictures taken with the larger sensors. The picture taken with the APS-C size sensor would appear magnified 1.5X compared to the full-frame picture. Or put another way, a picture taken with a 150mm lens on the full-frame camera would frame exactly the same area as one taken with a 100mm lens on the APS-C sensor camera. Hence the term "1.5 crop factor".

OK, that's pretty easy. But what if you are shooting video with a 50mm lens on an APS-C size camera, and want to frame the shot identically on a camcorder with a 2/3" sensor? Well you could probably convert back and forth from APS-C to full frame and then to 2/3" sensor using the handy table I made for you above. But you might have noticed in that table that the diagonal measurement of the sensor size is proportional to the crop factor. For example, 43.3mm (full frame sensor diagonal)  / 22.4mm (4/3 sensor diagonal) = 2, etc.

So, to make that conversion from 4/3 sensor to 2/3" sensor we can just divide the diagonal measurements of the sensors (27.3mm diagonal for the APS-C sensor, 11mm diagonal for the 2/3 sensor). The result is about 2.5, so we'd need a 20mm lens on our 2/3" sensor video camera to frame the shot the same way.

Really, really look at the sensor area

The numbers for crop factor and diagonal measurement of the sensors minimizes the actual differences in sensor sizes. If you want to really understand how much larger one sensor is than another, look at the column for the surface area of the sensor. The diagonal measurement and crop factor of a full-frame 35mm sensor is only 1.5 times longer than an APS-C camera, and twice the size of a 4/3 sensor. But the area of the full frame sensor is more than double that of a crop sensor, and almost 4 times that of a 4/3 sensor.

If the resolution of the cameras are the same, larger sensors mean larger pixels resulting in better ISO performance. (You can compare pixel sizes for a 12 Mp Nikon D700 and a 12 Mpix 4/3 camera by just pretending the above sensor size diagram is pixel size - no wonder there's a difference in high ISO perforamance). Or, instead of bigger pixels on the sensor, the manufacturer may put more pixels, which gives the camera higher resolution. Or some of each. Most full frame cameras have both more pixels and bigger pixels than most 4/3 cameras.

There are other factors involved, of course. Newer sensors have better microlenses and newer cameras better computer chips, both of which can make a big difference in high ISO performance. Underexpose the picture two stops and ISO performance doesn't matter either - you can't differentiate black from black. Put a crappy lens in front of the camera and the sensor's resolution doesn't matter - the camera can't photograph what the lens doesn't resolve. If you're just putting web-sized jpgs up none of it matters much. If you're making large prints every bit of it matters a lot.

But the area of the sensors explains why so many video people are abandoning their old camcorders and picking up AG-AF100s, Sony F3s, and video capable SLR cameras in droves. As an example, consider that a few years ago a very good $15,000 camcorder came with 2/3" sensors (58 square mm in area). Today, about the same amount of money will get you a Sony F3 with a Super 35 sized sensor (339 square mm in area, nearly 6 times larger), or 5 or 6 crop sensor SLRs with roughly the same size sensor.

It's also interesting to look at some new mirrorless and fixed-lens camera systems in terms of sensor size. Nikon chose to create the CX sensor size for it's new J1/V1 cameras. The reason is obvious: there was a large gap between the smallest SLR sensor (4/3) and the largest video (2/3") and point and shoot (1/1.7") sensors. The CX sensor fills that gap nicely. The CX sensor should be better than any point and shoot, but not so good that it takes business away from their SLR cameras. (There's an old saying that if you don't eat your own lunch, somebody else will. But apparently Nikon doesn't believe in that.) The Fuji X10 is using the very largest non-SLR sensor, the 2/3", which until now has only been used in video cameras. Canon, on the other hand, is releasing their G1X with a sensor slightly larger than the 4/3 sensors, although still smaller than their APS-C cameras.

There's a lot more to a camera than its sensor size, of course. Lenses come to mind. (Yes, Sony, I'm talking to you. Other companies are getting rich selling decent lenses and adapters to shoot on your mirrorless cameras.) It will be interesting, though, to see how the choice of sensor size affects the image quality of these new cameras. Software algorithms, electronics, and better microlenses all make a difference, but small pixels are still small pixels.

But we'll discuss the effects of sensor and pixel size more in the next article.


Patrick's comment made me consider that while I unconsciously group sensors into categories, I didn't really present that in the article. So maybe it will help if I do it here:

Sensor Area mm2 Sensor Type examples
1200+ Medium Format* Leica S2, Hasselblad, etc.
800-900 Full Frame Canon 5DII, Nikon D700, etc.
300-400 Crop Frame APS-C SLRs and mirrorless, Red, Super 35, 35 Cine
200-300 4/3 type Canon G1X, 4/3 cameras
about 100 CX Nikon J1/V1, Super 16 film
40-60 2/3", 1/1.7" best camcorders and P&S, Fuji X10
under 40 P&S sensors camcorders, P&S
  • Amol Kolhe

    Traditionally (when most cameras used mostly identical technologies), sensor size was all that mattered.

    But with new innovations from various companies happening in sensors & the electronics involved, it is becoming more apparent that size is not all that matters. What really matters is the quality of the image that comes out of it.

    For example, Fuji with its aps-c mirrorless cameras is able to achieve image quality and iso performance that rivals full frame. Similarly with their 2/3 sensor, they are able to get low light performance that is comparable to results from larger sensors.

    Olympus with their OM-D has shown us that micro four third sensor can produce images almost as good as aps-c.

    While the fundamental difference of physics remains the same that larger sensor will gather more light, but we're seeing lot of innovation that is bending the rules a little. But in the end its more choices for consumers, which is awesome.


  • Walt French

    Perhaps smartphone users are less critical about their photos, but clear articles like this one would be even more helpful in including examples of some representative models.

    Actually, when I look at the 21:1 ratio between the areas of my APS-C and the iPhone sensors, it's quite amazing that the iPhone is as good as it is. The newer technologies you mention must be contributing a LOT. As this also shows up in RAW files, I suspect it's not the limitations of the older in-camera chips.

  • Nicholas Bodley

    The camera tube you show is a vidicon, much smaller and much less costly than image orthicons, which were the best. Pretty sure that the first vidicons were one inch in diameter (glass, or photosensor?). When smaller ones were developed (such as ⅔ or ½ inch, iirc), that started the designations by fractions of an inch. (Of course, vidicon lenses were made for specific sizes of tubes.)
    The current fractional-inch designations for digicam sensors must have developed from the vidicon scheme, although vidicon fractional sizes used integers only for numerator and denominator.

  • Tom

    In terms of swabbing out lenses between formats I think an important issue is: How big it is the area of illumination. In other words if one were to put a super 16 on a 4/3 camera how much light fall off would once see in the corners. I have been looking for and have not been able to find a table listing the area of illumination at the focal plain for various designed for various film and sensory sizes. Does anyone on this for him have any suggestions?

  • Brian

    The diagonal measurement for the 4/3 is listed above as "22.4" but the pythagorean theorem (for 13.5 x 18) yields 22.5. Is this just a typo? Or is this a real measurement, implying some sort of non-planar surface?

  • Richie

    Just to be clear, the above comment was made in jest, but it's interesting how the pixel size between the two sensors appears to be identical.

  • Richie

    36 x 24 = 864 (Nikon FX)
    24 x 16 = 384 (Nikon DX)

    864 / 384 = 2.25

    16MP (D7000) x 2.25 = 36MP (D800)

    So the D800 sensor might just be two D7000 sensors glued together!?

  • DaveB

    Above, it is written, "....Put a crappy lens in front of the camera and the sensor’s resolution doesn’t matter."

    Has anyone taken the same body and tested with 3 different lens, but same focal length, to show the differences between crappy, average and great with images?

    I'd like to see a site that says, "Brand X is a crappy lens, and here's why."

  • http://www.digitphotography.com.au David

    Another great article. I think it's worth emphasising a couple of points. First, the Canon APS-C crop factor is 1.6, while Nikon and Sony are 1.5. I know you mention Canon's are a bit smaller, but it's not spelled out clearly in the table. (I think the double asterisk was meant to point to a note somewhere but it's been left out)

    Secondly, it's worth explaining why diagonals are used to calculate crop factors and make other comparisons. I'll explain it like this. The image projected from the back of a lens is generally circular. If you measure the diameter of this circle, then that is the diagonal of the biggest sensor you could put behind the lens to capture the image without vignetting. So by comparing diagonals, you are kind of comparing the diameters of the image circles that match those sensors. As Michael points out above, you can have two sensors with the same vertical dimension but different aspect ratios and hence different diagonals (e.g. G1X and 7D/60D/600D).

  • Esa Tuunanen

    Roger Cicala: "Plus it works. I just saw on one forum where someone was claiming the G1X sensor was larger than APS-C because 1.5″ equals 38mm. So maybe Canon knew exactly what they were doing."
    Yep, average consumer is just like dung fly, except there are less dung flies in cow's pile than people in shop selling latest "Second coming of Jesus" hype product. (and dung flies accept only fresh stuff while consumers are happy with museum age crap)

    Actually 2/3" sensor was used in many digicams. You just have to look ten years into past to time when digicams were tools of enthusiasts and before they become average consumer's toy.
    5MP Sony made sensor was used in four Minolta Dimages (7, 7i, 7Hi, A1) and in two Sony's "grenade launchers" (F707/F717) and 8MP sensor of 2004 was used by every major brand's camera, depicted in nice family portrait in this page:
    KonicaMinolta A2 would actually give about perfect basic body design for DSLR challenging EVF based high end mirrorless body instead of current P&S/retro fashion toys. Even its ~920k dot EVF was almost half decade ahead of time. (at time when ~230k dot LCD was the greatest thing since sliced bread according marketroids)

  • Carl

    Walter, what "glass" do you want?

  • Walter Freeman

    I really hope Pentax can get their stuff in order. The K5, RAW noise reduction or not, looks like a brilliant little camera. I'd shoot one in a heartbeat if they made the glass I wanted (and if their AF is up to snuff, have no idea about that.)