Frame WARS!




See also our Full Frame Wars page.

How big is yours?

Digital photography has a not so secret number:
Mega. As in, "you got deca mega?"

Sure, mega stands for "million" but in most digitographers' minds, mega simply means More. More pixels, more speed, more Bigger-ness. The new vernacular of photography includes "Chimping" -- reviewing on screen with punctuations of "Ooo, ooo!" and "How much Mega does your camera have?"

The under-$2,000 cameras have grown. More mega has found its way into the image chips. The recent Nikon D200, D80 and D40x demonstrate the trend.

So does the Canon EOS 400D (Rebel XTi).

Within the D200, D80, D40x and 400D beats the heart of all digital cameras, the imaging device. In this case a DX format CCD made by Sony on Nikon chip manufacturing machinery or a home-grown CMOS chip fabricated in Canon's back room. It produces an image with around 10,050,000 pixels, but each chip has more around its edges for reading black levels and computational purposes. Call it 10.2 MP for Mega Pixels.

One of the first questions on everybody's mind, if not their lips, is, "What is that in real-world values?" People want to know not just the raw mega, but the meaning behind the numbers.

That's our job. Putting meaning in between the digits. So here goes:

The outer frame is the new Nikon Deca-Mega (ten million) pixels compared to other recent and competitor frames. The Canon 10MP CMOS is almost identical in size (8 whopping pixels wider, or at this scale, the width of that little black border).

Canon's 20D (and 350D), 8.2 MP frame is seen in blue. The D70/50/40's 6 MP image is shown in darker gold, but we have superimposed the Medium frame from the new D200, D80 and D40x on top to show a cool idea: Medium is the New Big.

Worth noting: the Canon 400D's Medium frame is 2816 x 1880 pixels, just a thread smaller.

In the competitive battle for hearts, minds and eyeballs, the ever-growing pixel count is a dominant force, causing people to regard this camera or that as being "better" due to the sheer number. After all, isn't having a billion dollars substantially less fun than having two billion dollars?

For the first part of the digital photography revolution, mega meant the difference between mediocre and mega-nificent. By the time images passed the 3 MP zone, they made real nice-looking 8 x 10 prints with 200+ pixels in every inch of paper. Things began to drift away from being quite so critical.

At 5 MP, that 11 x 14 in your hand now contained 185+ pixels in every running inch of print and you held it slightly farther from your eyes as you viewed it. Then the Nikon D70 and Canon 10D cameras showed us 6 MP images, and the print size grew to at least 16 inches in the long dimension.

Wouldn't you know it? Canon showed the 20D and 350D (Rebel XT) with 8 MP imagers and either the image acquired more detail or the print grew to 19 inches or more. How would Nikon compete? Now we know.

N.B: The world carries an ancient (in photography, ancient is anything over 20 years old) standard on its back: The presumption that it takes "300 dpi" to equal "photographic". Professors point to the pixel and chime in that this is what the "dot" in dpi is all about.

That's an obsolete standard, in my view. Your human eye only resolves 50 line pairs per degree of view, and typical viewing of a letter page or 8 x 10 print is reading distance, about 22 inches (0.53 m), making the page/print about 26 degrees in your field of view (long dimension). 26 x 50 pairs = 2600 pixels in 10 inches of visual material to satisfy the absolute ultra maximum of human perception.

The old standard sez: 3000 pixels are required to satisfy "photographic -ness," per image long dimension, so apparently photography is required to transcend the outer bounds of human perception by providing 115% of what is humanly possible to visually detect.

OR... That 3800+ pixel image from a 10MP camera will place around 150% of eyeball-detectable detail into the print, roughly. In other words, digicams have your personal best beat by a comfortable margin. Unless you are a freak of nature with radically higher vision.

Some die-hard publications actually use 300 dpi dot screens to display their images. Arizona Highways comes to mind. And they'll tell you that nothing short of a 4x5 camera can ever take advantage of such wonderful reproduction. Nat Geo, on the other hand, has rotogravure printing at 175 dot screen frequency for its covers and 180 dots/inch for interior images and sends its photographers out with 35mm and digital cameras.

If the 50 line pairs (100 pixels) equal to one degree of view were applied here, presumably Arizona Highways will be able to fill three degrees of your eyeballs (per inch of page). That's just great. Now you are holding a large magazine around six inches from your eyes so you can soak in all the wonder. But get out your magnifying glass and try to find detail that is one printing dot wide or tall. I dare you.

On the other hand, perhaps this is another example of people insisting on things that don't make as much sense as they thought. Remember, "people" told all their friends and neighbors that flight was impossible, the world was flat, a vote for Nixon was a good thing and so on, at various times in the past. People, it appears, will say anything. WMD, anyone?

Do you sense a note of shrill irony in my rant? I hope so. Unless we refuse to run with the pack and question things due to their exalted levels of nonsense, progress will stop.

It's not the person who treasures the status quo who moves the world forward.
--Peter iNova, 2006.
The human-erogonomic scale of "photographic -ness" is less demanding. Let's say that we like the photographic qualities of a fine print that is reproduced in a top-quality magazine. Those are coming into your eyes from dot screens that have around 180 dpi.

They're close to "photographic" but the dots get in the way. Photos have even higher quality. But printing dots are not as accute as pixels on a page. By the time you deliver 180+ well-formed image pixels into every running inch of glossy paper, you're looking at photographic quality's anchor point.

By about 220 pixels per running inch of ink-jet, the race is over. The print has hit the wall. Unless you're viewing the image through a magnifying glass, more pixels won't help you one bit.

So where does 10.2 MP put us? At 180 ppi (pixels, not dots), the giant 3872 pixel field of the Nikon D200 and D80 delivers potentially 21.5 inches (0.54 m) of picture. The printer that makes this for you will cost around $2,000 because it will be substantially larger than tabloid size at 14.3 inches wide. That $1699 camera body (or $999 D80, or $799 EOS 400D) just cost you an extra two grand!

Or, perhaps your A3+ (13 x 19 inch paper) prints with affordable printers will simply be a little sharper and more "photographic." Not to put too much spin on this, but I have a bunch of incredibly good looking, sharp, sizzling A3+ prints from the D70's 6 MP image that tend to suggest that the D200 / D80 / 400D photographer is going to be shopping for some really big printer hardware. (Epson 4800 = $2K to $2.5K.)

Shooting the D200 / D80 / D40x / 400D with Medium frames is almost exactly like shooting the D70's Full frame. Yet the 10MP Medium frame is a down-conversion resulting in more perfect pixels. Those D70 / D50 / D40 shots produce crowd-pleasing, eminently photographic 16 inch prints, by the way, as noted.

Inevitably, the idea of comparing these digital images to 35mm film is always lurking in the background. Early tests (2003-ish) showed 35mm film to be less than stunning when compared with the 6 MP imagers of the day!

My standard mantra on this is that you can't beat a direct scan of the world, no matter how many chemicals and dyes you try to throw at it. Digital cameras take an instant color separation on the spot, and film is on the wane.

So are tin-types, B&W print paper and "Say 'Cheese'!" Unless, of course, you are Wallace and/or Gromit.

Compared to the Canon 20D's 3504 pixel long dimension, the 3872 pixels of the Nikon D200 / D80 / D40x sport only 10.5% more tiny tiles of color across the shot. That's not a lot of difference--5% on each side. Not enough to make you sell the Canon for the Nikon, but if you have been shooting with the D70 / 50 / 40's 3008 pixels, the new cameras are elevating the bar by a substantial 29%. That's commensurate with the difference between a 5 MP image and a 3 MP image -- in other words, a full step forward. If you start putting five dollars a day (two Starbucks) into a Camera Fund, you'll have just about enough in a year. Start saving.

What's coming around the bend? More pixels, no doubt. The Canon Full Frame 5D and Nikon D2Xs both have over 12 MP. Doubling the megapixel count means increasing the density or size of the image chip. If Nikon were to acquire an imager with 20MP, all they'd have to do is bring out a full frame camera, or reduce each photosite for DX lens cameras to 71% of their current size (reciprocal of the square root of 2).

Since photosites are now perched at 6 micrometers across, that would take them down to about 4.3 micrometers square--half the surface area of today's 10MP photosites. Advances in CCD and CMOS tech might make that size photosite into a relatively noise free idea, but it's going to take some work.

Maybe not. Consider this, the 8MP small format image chips have working photosites in the sub-2 micrometer range with about 10% of current 10MP photosite surface area. Why am I harping on surface area? That's the direct indicator of how photosites statistically count photons. And statistically larger samples tend to agree with each other from photosite to photosite. Agreement = lower noise.


Check out our first impressions of the D200 here. The D80 here. Our reaction to the EOS 400D here. And for grins check the Full Frame Wars editorial here.

(c) 2007 Peter iNova. All rights reserved. Do not reprint. Simply add a link to this page.