© 2002- 2007 KenRockwell.com
About these reviews
I first wrote this page in the dawn of digital photography in 1999. We dreamed of 6 megapixel cameras as the ultimate replacement for film. I had just dropped $5,000 on a 2.7 megapixel D1H.
Back in those days, and for resolutions up to about 6MP, a smaller DX or 1.6x sensor was, and still is, ideal for the reasons I explore below. See also crop factor to understand what happens to focal lengths and fields of view as sensor sizes change.
As of 2007, 10 MP DX and 1.6x cameras are owned by every junior high schooler and Nikon is pushing 12 MP DX format cameras.
Unfortunately, these sensors have so much resolution that too many pixels are getting jammed on the sensor in the space they have.
They have such high resolution that the lenses, and not the camera, are the limiting factors. They are so close together that they have to gasp for light, so noise and ISOs suffer.
I bought a Canon 5D in 2006, and love how the larger sensor cut loose and gave much better technical quality than anything I'd ever gotten from my DX Nikons. Unfortunately Canon lacks Nikon's ergonomic élan.
The rest of this is my article from 2002. Unfortunately the expense of fat sensors still exists in 2007, but for 10MP and higher resolutions, the larger sensors are better.
Digital SLRs take advantage of using sensors about 16 x 24 mm, close to the original standard 35mm (movie) film standard of 18 x 24 mm introduced over 100 years ago and still in daily use for multi-million dollar Hollywood movies and TV.
When miniature still cameras were designed to use 35mm movie film in the 1920s they were called "double frame" and shot a double wide frame of 36 x 24 mm. This remains the size of today's 35mm film (still) cameras which are fast becoming obsolete.
Some old timers think there is a need for a double-sized CCD sensor to mimic the old 35mm film (24 x 36 mm) for digital cameras. I have no idea why they would think this, except that before short zoom lenses like Nikon's 12 - 24 mm and Canon's 10 - 22 mm were introduced it was hard to get ultra wide shots on digital SLRs. There would be a minor potential improvement in noise with the bigger sensor, but not particularly important when comparing the two versions of these big sensors.
The huge visible difference in image quality due to sensor sizes comes when comparing SLR-sized sensors to the teeny-tiny sensors used in fixed-lens cameras. Compact digital cameras have sensors only a few mm on a side. Canon has a good explanation of why a bigger sensor leads to quieter images here. The difference between a compact camera and an SLR is obvious. It's not that big a deal when comparing double-frame (24 x 36 mm) and single-frame (16 x 24 mm) sizes in SLR sensors.
Today the current 16 x 24 mm size sensors are ideal and have many advantages over the dinosaur size of 35mm still film. Among these are deeper depth of field leading to sharper pictures since the focal lengths are shorter, and smaller, lighter and less expensive lenses since everything can just be smaller.
Magnification or Crop Factor
This is so ridiculously simple that some people are confused by it.
The CCD is smaller than film, so your image from any lens is the same as cropping out from the middle of a film frame, duh.
A more complex way to imagine this change in field of view is to play a game and imagine that your lens changed focal length. Typically this factor is 1.5, meaning a 50mm lens used on a typical digital SLR will give the same angle of view as a 75mm lens on a film camera.
This is all pretend; the focal length doesn't really change. It's just a way for people familiar with lenses on 35mm cameras to imagine the fields of view on DSLRs. Different brands of camera will vary. Common factors range from 1.0 (Canon 1DS) to 1.8 (Sigma).
See also Crop Factor.
DISADVANTAGES OF CCDs IN THE OLD 35mm FILM SIZE
Worse, slower, and more expensive.
Expense is the main reason. We are all used to huge drops in price and increases in computer and electronic performance every few months. These improvements are all directly related to reductions in process geometry (size) of all the chips that go into everything. Guess what: things still cost about the same per square millimeter! That's right; we are all so used to better and faster and cheaper only those of us in IC design and manufacture realize that this is due to us shrinking everything all the time. Thus none of this applies to a chip that has to stay the same size; all the benefits we take for granted are from making things smaller. Oh well, there simply is not going to be a much cheaper way to make a mammoth 24 x 36mm CCD.
Simply doubling sensor size much more than doubles the cost of the chip. Die size has a very high exponential effect on the cost of manufacture. Someone correct me here, but as I recall it's something like 20 times more expensive to double the die size. There are many factors that contribute to this, and this incredible expense is why there is no reason to use an outdated size just because that's the way it used to be done. Since the wide angle problem has been solved, why bother with the expense of a huge sensor and all the big heavy lenses we used to have to lug around?
Most all of the progress in computers and technology has come from reducing dies size and geometries for computer chips through better manufacturing techniques. The reason everything today is better, faster and cheaper is because the die sizes of electronic chips have been reduced continuously since their invention in 1958. Big chips like CCDs are dinosaurs; everything in the chip business has been focused on reducing die size and geometries for over 40 years.
Smaller die run faster and use less power than larger die. Smaller geometries allow more parts on a chip, increasing what it can do.
Speed is limited by signals having to stay in synchronization across the chip. Differential propagation delays mess us up. Yes, the tiny distances on a chip are very important when dealing with the speeds we do. All the signals on a chip have to stay in synchronization, so if one path gets too long or some temperature variation causes one path to be delayed relative to another you have signals getting crossed. Anything that can be done to reduce size increases speed. Chip size reduction is what has increased speed continuously, since smaller chips have less propagation delays.
The cost of making a chip varies as a very high exponent of the die area. This is because:
You get fewer chips per wafer if they are bigger.
Is Nikon going to go backwards and make an oversized 24 x 36 mm CCD camera?
I doubt it (2002). Why would they, now that you know a little bit more about this? I wouldn't wait around. I personally just bought my own D70 and 12 - 24mm lens at full price with my own money, so I'm voting with my wallet on this one. Of course only a few people at Nikon Japan actually know this, and I'm betting Nikon never bothers with an oversized CCD.
I see no reason for CCDs to bother with the old 35mm size, nor does Nikon as far as we've seen. Since Nikon makes short lenses for the smaller sensors of DSLRs unlike anyone else I see no reason to bother with the old size. You could pay $8,000 for the old-CCD-size Canon 1Ds, or one-eighth as much for a shorter lens for a Nikon to get the same wide angle views.
There are numerous advantages to the standard DX 16 x 24 mm sized CCD, and none to the old 35 mm film size. Unlike film, bigger CCDs cost an hellacious amount more to manufacture. Chips that huge are usually laughed out of wafer fab.
Now that Nikon got me and everyone to buy all their DX lenses, and hopes we all threw away our film lenses, they'll introduce a full-frame camera and make us all buy new lenses.
I hope they do: I want to buy a 13mm f/2.8 and/or 15-35mm f/4.