Camera review rant
I was a bit annoyed reading the latest comparative review among five digital SLRs in Popular Photography, who are usually on the ball in their reviews. My camera, the Pentax *ist DS, ended up in last place on critical criteria such as image quality. Upon closer reading, it became clear that, first, the cameras compared were not in the same class and, second, the criteria for analysis of image quality were open to serious question.
The cameras ranked 1st and 2nd were the Canon Digital Rebel XT and Olympus EVOLT E-500, respectively, but both are 8MP cameras and cost $200-300 more than the remaining three 6MP cameras in the review, a substantial difference in the under-$1,000 market. It should come as no surprise that you get what you pay for, but it's not very useful to compare items of entirely different classes.
However, of the three 6MP cameras reviewed, the Pentax *ist DS trailed the Nikon D50 and the Konica-Minolta Maxxum 5D for image quality, and it is interesting to analyze why. One might fairly choose among these cameras on some basis other than image quality - for example, the Pentax has a stainless-steel chassis while the Nikon and Konica-Minolta are all polycarbonate plastic or fiberglass - but image quality is pretty important.
Where I primarily disagree with the review is that image quality is effectively assumed to be equivalent to resolution, which is not at all true. In fact, resolution is traded off in these cameras against light sensitivity, and Pentax applied a different, almost opposite, design philosophy than their competitors. The laws of optics dictate that trying to achieve high resolution in low light will produce "noise" in a digital camera, which is analogous to "grain" in a film camera: randomly colored flecks or spots scattered throughout the image, in extreme cases looking like "snow" on a television screen.
The Pentax imaging system is optimized for low noise, not high resolution. As a result, the Pentax has almost no noise at quite high light sensitivity (ISO800), very low noise at even higher light sensitivity (ISO1600), and surprisingly acceptable noise at extremely high light sensitivity (ISO3200). The Nikon cannot even be set for ISO3200, and the Konica-Minolta requires a special override to get to ISO3200. More importantly, in both the Nikon and Konica-Minolta, noise is unacceptable at ISO1600 and noticeable at ISO800.
For the review, the cameras were set to their minimum light sensitivity (ISO200), penalizing the Pentax on one of its strongest features. The Konica-Minolta actually earned points for having "image stabilization" firmware to fight camera shake, but this is just an elaborate way of achieving the same goal that the Pentax accomplishes by the more straightforward approach of making the camera more light-sensitive and therefore allowing for faster shutter speeds.
How much resolution does the Pentax sacrifice to get this light sensitivity? At high magnification, there is no question that there is some sacrifice, but no one looks at photos with a microscope unless they are writing a camera review. For images going on the web, there would be absolutely no difference. For prints, I doubt the difference could be detected up to 8x10, although it might be barely detectable on close examination at 11x14. Of course, if one is printing enormous enlargements to 16x20 or more, it would be possible to process the images digitally using a sharpening filter. High resolution itself may not be desirable, especially in portraits where "soft-focus" is often used deliberately to prevent every skin defect from becoming emphatically obvious.
In almost all practical situations, the Pentax would produce overall better images; the exception is where all of the cameras are forced to their minimum light sensitivity, as in the test. The Pentax is simply not designed to be used primarily on its minimum light sensitivity.
Pentax also just cut the list price on their entry-level *ist DL by $200, from $800 to $600 with a kit lens. Since the DL has the same imaging system as the DS, although with less autofocus sensor points and other reduced features, it may overcome these kinds of negative reviews.
Disclaimer: I used to receive a free subscription to Popular Photography as compensation for editorial contributions.
EDIT 2007-05-28 22:30 EDT: Over a year later, I just reread the original article on the web, and it seems quite different from my recollection of the printed version. In particular, my objections are explicitly acknowledged and the criticism of the Pentax greatly toned down, as its review now includes phrases such as "overall sharpness and detail are the lowest of the bunch, which may contribute to camera's extraordinarily low noise levels," but also asserts that "the sacrifice of maximum resolution for low noise will limit its appeal for true enthusiasts." As the article now notes, such "lowest of the bunch" performance is hardly open to criticism, as the 6MP Pentax is said to be "good for making prints up to 10x15 inches at 200 ppi" while the top-rated 8MP Canon is "good for making prints up to 11.5x17 inches at 200 ppi." Personally, I would have a lot more use for being able to shoot reasonable quality at ISO3200 than for making very slightly bigger enlargements.
The cameras ranked 1st and 2nd were the Canon Digital Rebel XT and Olympus EVOLT E-500, respectively, but both are 8MP cameras and cost $200-300 more than the remaining three 6MP cameras in the review, a substantial difference in the under-$1,000 market. It should come as no surprise that you get what you pay for, but it's not very useful to compare items of entirely different classes.
However, of the three 6MP cameras reviewed, the Pentax *ist DS trailed the Nikon D50 and the Konica-Minolta Maxxum 5D for image quality, and it is interesting to analyze why. One might fairly choose among these cameras on some basis other than image quality - for example, the Pentax has a stainless-steel chassis while the Nikon and Konica-Minolta are all polycarbonate plastic or fiberglass - but image quality is pretty important.
Where I primarily disagree with the review is that image quality is effectively assumed to be equivalent to resolution, which is not at all true. In fact, resolution is traded off in these cameras against light sensitivity, and Pentax applied a different, almost opposite, design philosophy than their competitors. The laws of optics dictate that trying to achieve high resolution in low light will produce "noise" in a digital camera, which is analogous to "grain" in a film camera: randomly colored flecks or spots scattered throughout the image, in extreme cases looking like "snow" on a television screen.
The Pentax imaging system is optimized for low noise, not high resolution. As a result, the Pentax has almost no noise at quite high light sensitivity (ISO800), very low noise at even higher light sensitivity (ISO1600), and surprisingly acceptable noise at extremely high light sensitivity (ISO3200). The Nikon cannot even be set for ISO3200, and the Konica-Minolta requires a special override to get to ISO3200. More importantly, in both the Nikon and Konica-Minolta, noise is unacceptable at ISO1600 and noticeable at ISO800.
For the review, the cameras were set to their minimum light sensitivity (ISO200), penalizing the Pentax on one of its strongest features. The Konica-Minolta actually earned points for having "image stabilization" firmware to fight camera shake, but this is just an elaborate way of achieving the same goal that the Pentax accomplishes by the more straightforward approach of making the camera more light-sensitive and therefore allowing for faster shutter speeds.
How much resolution does the Pentax sacrifice to get this light sensitivity? At high magnification, there is no question that there is some sacrifice, but no one looks at photos with a microscope unless they are writing a camera review. For images going on the web, there would be absolutely no difference. For prints, I doubt the difference could be detected up to 8x10, although it might be barely detectable on close examination at 11x14. Of course, if one is printing enormous enlargements to 16x20 or more, it would be possible to process the images digitally using a sharpening filter. High resolution itself may not be desirable, especially in portraits where "soft-focus" is often used deliberately to prevent every skin defect from becoming emphatically obvious.
In almost all practical situations, the Pentax would produce overall better images; the exception is where all of the cameras are forced to their minimum light sensitivity, as in the test. The Pentax is simply not designed to be used primarily on its minimum light sensitivity.
Pentax also just cut the list price on their entry-level *ist DL by $200, from $800 to $600 with a kit lens. Since the DL has the same imaging system as the DS, although with less autofocus sensor points and other reduced features, it may overcome these kinds of negative reviews.
Disclaimer: I used to receive a free subscription to Popular Photography as compensation for editorial contributions.
EDIT 2007-05-28 22:30 EDT: Over a year later, I just reread the original article on the web, and it seems quite different from my recollection of the printed version. In particular, my objections are explicitly acknowledged and the criticism of the Pentax greatly toned down, as its review now includes phrases such as "overall sharpness and detail are the lowest of the bunch, which may contribute to camera's extraordinarily low noise levels," but also asserts that "the sacrifice of maximum resolution for low noise will limit its appeal for true enthusiasts." As the article now notes, such "lowest of the bunch" performance is hardly open to criticism, as the 6MP Pentax is said to be "good for making prints up to 10x15 inches at 200 ppi" while the top-rated 8MP Canon is "good for making prints up to 11.5x17 inches at 200 ppi." Personally, I would have a lot more use for being able to shoot reasonable quality at ISO3200 than for making very slightly bigger enlargements.