After we ran headline articles last month about the first two digital SLRs to have full size CMOS chips we received a surprising number of emails from folks asking just this question. We also received emails from a number of folks indicating that full size chips were a bad idea because they would no longer turn a 400mm zoom into a 600mm zoom as current 1.5x chips do!

With this much misunderstanding and misinformation out there we thought it was time for an article explaining some basics about digital capture - in plain English.

Plain English however means that we won't go into minute technical detail on some of these points. So we ask forbearance among our more technical readers. For the sake of this article a chip is a chip is a chip - whether it be a CMOS, CCD, or Foveon chip we'll refer to it as an 'image chip'. While these various chips have advantages and disadvantages for the engineers designing the circuit boards we really don't care which style is used to capture the image (Foveon's claims notwithstanding).

Size Matters 
Size matters in two ways. It matters in the size of the chip and it matters in the size of the pixels on that chip. Up to a point smaller pixels are bad and larger pixels are good. Why is it that a 3 megapixel digital SLR can provide better images than a 4 megapixel point and shoot? Part of the equation is optics. But just as important is the fact that the digital SLR has an image chip that is quite a bit larger than the digital point and shoot in this example. That allows it to capture an image with more detail and tonal quality because the larger chip has larger pixels. That image will in-turn enlarge into better-looking 8x10s and 11x14s.

Size also matters when it comes to a full size vs. a smaller image capture chip in a digital SLR. In both general use and in print I've seen people refer to the 'advantage' of smaller digital chips in that they give you a free 'extension' at the telephoto end. The tradeoff being that your wide-angle lenses aren't as wide. While this generally describes what you see it's not really accurate and it's led to a misunderstanding of what's going on with full-frame chips and their real advantage.

The misunderstanding really comes from the marketing departments at Canon, Nikon and Kodak. Longtime readers of Vivid Light know that I have a special contempt for copywriters in marketing departments. I'm not sure if marketing departments are the training grounds for the guys who write congressional press releases or if it's where they go to retire but the result is the same - the truth can be a slippery thing.

The image on the left was shot using a 300mm lens on a standard SLR. The image on the right shows the effect of mounting the same 300mm lens on a digital camera with a 1.5x chip. The result is the equivalent of a 450mm lens. This is a good thing right?

Early on marketing literature from the camera companies started describing the chips in digital SLRs as 1.5x and 1.6x chips. The formula, said the marketing folks, was that you multiply the focal length of your lens by 1.5x or 1.6x (depending on the brand) to get the length of your lens when mounted onto one of their digital SLRs. The implication being that these chips actually provided some kind of magnification. This was touted as a great advantage for telephoto shooters and a minor disadvantage for wide-angle shooters. After all wide-angle shooters could always pick-up a wider-angle lens! And every lens manufacturer has indeed made "digital" lenses available. Sales have been brisk.

Image chips in current digital SLRs just crop the image provided by the lens. They don't offer any magnification. Is this really an advantage?

The image on the left was shot using a 300mm lens on a standard SLR. The image on the right shows the effect of mounting the same 300mm lens on a digital camera with a 1.5x chip. The result is the equivalent of a 450mm lens. This is a good thing right? This was a good thing said the marketing folks. What they didn't say was that it would be more accurate to describe that 1.5x chip as a 2/3rds chip.

The reality is that these chips are smaller than the area covered by a frame of 35mm film. There was no extra magnification, as was sometimes implied, and erroneously reported by journalists in a number of articles.

All these smaller chips do is to crop the image. That 'gain' is in fact no gain at all. It is a loss. And that loss occurs for both telephoto and wide-angle lenses.

Image chips in current digital SLRs just crop the image provided by the lens. They don't offer any magnification. Is this really an advantage? In the example shown here it would be no different than shooting with a 300mm lens on your film SLR and cropping the final image for a tighter shot.

By going to a full frame image chip you are able to capture the entire image provided by your lenses.

Those photographers and journalists who complain that they lose the "magnification" if they switch to a camera with a full frame image chip don't understand what's going on. If they switch to a full frame chip, shoot with the same lens, and crop the final image they'll still have the same quality because they're doing the same thing!

The difference is that their wide-angle lenses will still provide their full angle of view with a full frame image chip - meaning they can forego the purchase of expensive super-wide lenses.

No Free Lunch 
So if full frame chips are so great why weren't they used from the start?

The answer is that the technology wasn't ready. When you capture larger images you need more memory in the camera, larger image buffers to hold images while they're waiting to be written to memory cards, and faster and larger memory cards to process and hold those larger images. While these technologies existed three years ago they were larger, more power hungry, and more expensive.

Advances in component design have allowed camera designers to package all these features into a camera that is a reasonable size, works reasonably fast (2 or more frames per second), will take a reasonable number of images before it kills the batteries, and can be sold at a (ah hem) reasonable price.

But there is another serious problem that faces the designers of cameras with full size chips.

Canon's solution to controlling the angle of light striking their CMOS  chip is to place a grid of micro lenses in front of the chip to bend the  incoming light to a 90 degree angle

Image chips are made from silicon and silicon is much more fussy than film about the angle at which light strikes it. When the angle of the incoming light is relatively close to 90 degrees the chip is able to accurately capture the color and tone of the incoming light and render a sharp image. The edges of the smaller image chips are still close enough to the axis of the lens that the angle of the incoming light is a non-issue.

Canon's solution to the issue of the angle of incoming light is to place a grid of micro lenses over the image capture chip. Each pixel has it's own dedicated micro lens. But get out near the edge of the frame, as you do with a full frame chip, and the angle of the incoming light can become a problem with some lenses (depending on the position of the rear lens element and the design of the lens itself).

The result can be odd artifacts, color shifts, or lack of sharpness at the edge of the frame. Canon designed a grid of micro lenses that are positioned over the image capture chip. These micro lenses change the angle of incoming light so that it strikes the chip at an angle closer to 90 degrees. The rep I spoke with from Kodak said this wasn't really an issue with the DCS 14n. Experience will tell us if either of these cameras have problems with specific lenses as a result of the angle of incoming light. Canon indicated that controlling the light angle is a design criterion in at least one lens that is now in development.

Keeping Up 
Digital cameras are going to continue to evolve at a fever pace for at least the next five years.

The challenge for manufacturers is to maintain some kind of consistency among new models. Then there is the problem of maintaining customer satisfaction. If a customer spends $5,000 or even $1,900 on a new camera they expect that camera to have a reasonable lifespan. If customers perceive that their new baby is obsolete in 12 or 18 months (or less) they won't be happy customers for long.

In part this is a perception problem. The digital camera that you buy today will still be providing the same image quality a year or even five years from now. But as newer technologies are introduced the concept of what is acceptable is constantly revised upward.

One way to address this problem is to build on a standardized chassis. When changes are incorporated into newer designs there is not a significant cosmetic difference from older designs. This helps existing customers from an ego standpoint.

A standard chassis may also allow manufacturers the option to offer upgrades when new technologies are introduced. These upgrades could be offered at a fraction of the cost of a new camera.

Nikon has upgraded the size and write speed of the image buffers on the D1X - a current model. So where does that leave existing D1X owners? Nikon is offering to upgrade the cameras of current owners for $246.50 including shipping. They are also offering firmware upgrades for the D1H/X and D100 and a software upgrade to Nikon Capture that will double the image size for images captured in raw mode with D1H/X cameras (see the news page for details).

Eventually the market will stabilize as improvements to new models reach a point of diminishing returns. A 200 megapixel camera won't offer a significant improvement over a 100 megapixel camera no matter what your application. At that point digital cameras will be judged on innovations that further their image making capabilities - just as film SLRs are judged today. Until then things will be both interesting and frustrating for consumers and manufacturers alike.

  Subscribe to Vivid Light 
Photography by email 

Tell Us What You Think