Sensor quality: SLRs erode medium-format lead

It looks like Canon and Nikon weren't blowing smoke when they said their high-end SLRs cameras will compete with medium-format digital cameras used almost exclusively by professionals.

Given the image quality advantages that SLRs with larger "full-frame" sensors have over mainstream and much less expensive models with smaller processors, one might have expected another quantum leap from costly high-end medium-format digital cameras with sensors twice the area of top-end SLRs. Not so, according to new DxOMark Sensor test results set for release Tuesday by French test and measurement firm DxO Labs.

In the digital era, Canon's top-end SLRs give medium-format cameras a run for their money. (Click to enlarge.)

(Credit: DxO Labs)

The company tested image sensors from several medium-format cameras--the Mamiya ZD Back, Leaf Aptus 75S, Hasselblad H3DII 39, and Phase One P45+. These are the sorts of cameras used by fashion photographers and others who need lush tones, fine detail, and lots of megapixels to handle big photos such as magazine spreads.

But none outperformed the Nikon D3X SLR, whose score of 88 gives it the current top rank on DxO's sensor tests.

These medium-format cameras' overall scores were lowered by their poor low-light performance, but these cameras are generally used by studio photographers with carefully controlled lighting, so that in and of itself isn't a big factor. However, even overlooking that, high-end SLRs were competitive on the other attributes.

"We had expected that given their large-size pixel they would perform much better on dynamic range and color depth than digital SLRs," said spokesman Nicolas Touchard.

DxO measures the sensor performance by examining raw images, those that haven't been processed by the camera or software such as Adobe Photoshop into more convenient but limited formats such as JPEG.

Don't judge a camera solely by its DxO Mark score; many other factors ranging from price, autofocus, and lenses also are highly pertinent.

And bear in mind that DxO only checked a handful of medium-format models, and not the latest ones at that. For example, Hasselblad has begun selling a 50-megapixel model and plans to release a 60-megapixel one in 2009, and Phase One offers a a sensor housing that fits to medium-format camera body. Leaf's Aptus II backs employ even larger 56x36mm sensors and reach 56 megapixels.

Even in the film era, medium-format cameras were rarer and more expensive than 35mm film rivals. But in the digital era, the competitive dynamics have tilted more in favor of conventional SLRs, because the cost of producing image sensors gets dramatically higher as the sensors get larger.

These medium-format cameras' overall scores are lowered by their poor low-light performance, but these cameras are generally used by studio photographers with carefully controlled lighting. (Click to enlarge.)

(Credit: DxO Labs)

[Fly_Dog]

Sloppy caption -- where did Nikon go? "In the digital era, Canon's top-end SLRs give medium-format cameras a run for their money" (feel free to delete this if the caption is corrected).

[Shankland]

You're right--we fixed it. Thanks.

[Hep Cat]

And both digital formats are steadily eroding the last bastion of easy, cheap film superiority that remains - large format film 4x5 and larger.

It's hard to beat 20 square inches of film, but it looks like that day is coming.

[sondrasneed]

and with good riddance - never liked the darkroom personally, it's dark in there and chemistry has a shelf life - and often when I was in the "process". It's fine for the technician, but for the sloppy experimentalist, it was really difficult to repeat a cool accident.

[richardkaufmann]

Bigger sensors are better for a reason not measured by DxOMark.

With high pixel count (even full frame) DSLRs, using the highest f-stops cause a reduction in the image resolution regardless of the sensor. On my 1ds iii, this diffraction limit is around f9, and it's based on the size of the pixel. Larger pixels (e.g. in a medium format back) mean higher f-stops. Before the physics majors get you with this inherent limitation, higher f-stops also mean a greatly increased sensitivity to flaws in the lens. Bigger lenses, bigger sensors, more glass involved in the picture, ...

(Great discussion of this: http://www.luminous-landscape.com/essays/Equivalent-Lenses.shtml, see Nathan M's comments.)

So, yes, marvel at the incredible depth and low-noise characteristics of the eensy-teensy pixels of the 1ds iii and d3x. But when you want to slam the iris down to f16 to maximize the depth of field, all those extra pixels are just sitting there laughing at you.

p.s. There's some "unobtanium" Flash Gordon stuff that we may get any decade now. It's a "negative index material," that could help fix this diffraction limitation. http://physicsworld.com/cws/article/print/19415

[molotov]

Not too long ago I wrote about the fact that the current DSLR's with less-than full frame sensors. I said that they did not come close to the Full frame 35mm film superiority - I got slammed by people who counter-argued that their 10 megapixel DSLR's are just as good as film. How could I argue back with those people when such rediculous statements were made?
Now this new argument, given the research done by DxO, is quite plausible. Given that such big companies are behind the push, Nikon and Canon; the DSLR superiority over Full frame digital cameras is not so far fetched. Imagine; $8,000 DSLR delivering the same performance as a $25,000 Hasselblad?

[zmjman08]

"DxO measures the sensor performance by examining raw images, those that haven't been processed by the camera or software such as Adobe Photoshop into more convenient but limited formats such as JPEG."

By the way, the medium-format cameras all have on-chip noise reduction, which processes the raw data before writing the file that DXOMark evaluates. This makes the medium-format cameras appear to have less noise, simply because they have already run a noise-reduction algorithm.

Before putting too much stock in the new DXOMark tests, read this article from The Luminous Landscape:

http://www.luminous-landscape.com/essays/eyes-vs-numbers.shtml

The main point he makes is that the DXOMark test doesn't factor in camera resolution. This means that while the Canon and Nikon cameras may score as well on DXOMark test, the large-format cameras have nearly twice the resolution, making them better for professional-level photography or making large prints.

[xcopy]

DXO's "findings", based on extremely flawed tests, need to be taken with a grain of salt. About a 50lb. bag should do it, but get an extra bag just in case...

I'm sure non-photographers, the "snap shooters" and the wannabes that discovered photography in the last couple of years will will believe this. Ignorance is bliss as they say. In reality the digital game is no different from the film game; bigger is better for image quality.

As good as Nikon's new D3x is (and it may be superior to Canon's best), it is not a match for medium format.