The Battle of the Digital Giants

Do digital backs, combined with medium- or large-format cameras give you more than high-end (yet less expensive) DSLRs?

By Mark Dubovoy Back to

Dubovoy_JF_2008_1

Art is all about passion.Yet few forms of art engender as much passion as photography. Photographers are not only passionate about their work, they are almost obsessively passionate about their equipment and their methodology.This passion leads to heated discussions about many topics. One such topic is the use of DSLRs versus larger-format digital cameras.Therefore, I thought it would be very interesting to compare a top DSLR with a top medium-format digital outfit, as well as with a top digital view-camera system.

The purpose of this article is not to test every single DSLR, every single lens, every view camera, and every medium-format digital back on the market, but rather to take some representative examples to see if they lead to important or interesting findings. Also, rather than shooting test charts under laboratory conditions, I opted for the more practical approach of shooting mostly under real-world conditions.

I selected several state-of-the-art outfits for this comparison:

• A Canon 1Ds MK II body with Canon “L” lenses (see A at the top of this page). I will refer to this outfit as the “Canon.”
• A Hasselblad H3 with a 39MP Hasselblad digital back and Hasselblad lenses (B). I will refer to this outfit as the “H3.”
• A Hasselblad H2 (D) with PhaseOne P45+ digital back (C) and Hasselblad lenses. I will refer to this outfit as the “H2.”
• A Linhof M679 CS view camera with PhaseOne P45+ digital back and Rodenstock HR lenses (E). I will refer to this outfit as the “Linhof.”

These outfits are the Digital Giants.

General Comments

It would take hundreds of pages to explore in detail the characteristics of each outfit tested. However, I think I can make some useful general comments.

There are certain applications where a specific format has great advantages over the others. For example, I would not try to shoot a football game with a view camera. I would most likely use a DSLR because of the long lenses and fast shooting bursts required to capture the action up close. By the same token, I would never try to shoot serious architecture with anything other than a view camera, as camera movements and maximum detail are critical in this application. Likewise, for certain fashion shots, where fine detail in the fabrics and color accuracy are paramount, a medium- format SLR would be the preferred choice.

In spite of the above, there are vast areas of overlap where any of the formats can be used. Examples include product photography, still life, landscape, portraiture, nature, fine art, macro photography, industrial, and travel. In my experience, the overlap is big enough that in the vast majority of cases, it is up to each photographer to choose the type of equipment used.

Pro DSLRs are quite heavy. A typical professional DSLR outfit with lenses is about as heavy as a typical medium-format outfit, and similar in bulk once the equipment is packed in camera cases. The digital view-camera outfits have actually gotten smaller and lighter since the days of film, when a typical outfit was a 4×5 camera (sometimes a 5×7 or even an 8×10) with lenses and bulky film holders. This is because one can now carry a much smaller 6×9 cm view camera with a digital back, much smaller and lighter lenses, a smaller tripod, and no film holders, while still obtaining results that are in most cases better than using larger traditional view cameras. I estimate the weight and bulk savings to be in the range of 40–50%.

Bottom line: The DSLR and the medium-format outfits have about the same weight and bulk. The view-camera outfit is somewhat bigger and heavier, but not much. The DSLR is clearly the most convenient for handheld shooting. For tripod shooting, I find that the setup times for a DSLR and a medium- format camera are about the same, but I vastly prefer the larger viewfinder image and the better digital files that medium format offers. The view camera takes longer to set up, but for me this is a pleasure rather than a hassle. The precision, forethought, and accuracy required almost always results in better images for those who have the skills to properly use a view camera.

It is a common misconception that medium-format cameras use scanning backs that require very long exposure times. This is totally wrong. Medium-format backs have sensor arrays just like those in DSLRs, only bigger. The only other difference with

DSLRs is that DSLRs typically have CMOS sensors, and as far as I know, all medium-format digital backs have CCD sensors. Therefore, when shooting with either a medium-format SLR or a view camera equipped with a medium-format back, you can use exactly the same shutter speed, lens aperture, and ISO settings (up to a point) that you would use with a DSLR. The P45+ can shoot hundreds of images continuously at 2 frames per second, until the CF memory card is full. By comparison, the Canon 1Ds MKII can shoot faster bursts at 4 frames per second, but it needs to pause to write the files after 10–15 exposures. The newer Canon EOS-1D Mark III can capture 10 fps up to 30 frames.

Costs

Remember the good old golden days of film? In 1970, a Nikon F with the Photomic light meter and motor drive used to sell for $1,150 mail order. In inflation-adjusted dollars, this is equivalent to $8,230 today. This camera had a crude metering system and a motor drive, but no autofocus, no auto-exposure, nor any of dozens of functions we expect in our current cameras. I think it is safe to say that current pro-caliber DSLR’s are a bar- gain compared to the pro cameras of 30 to 40 years ago. And a medium-format outfit that looks very crude by today’s standards used to cost more than most luxury cars. Top-of-the-line view cameras like Sinars and Linhofs also were extremely expensive back then; much more so (in actual dollars) than the vastly improved current models. So as high as the prices for current medium- and large-format equipment seem at first blush, in real dollars they are a bargain compared to the equipment of a few decades ago.

The approximate actual street price for the Canon body is $8,000. Hasselblad has many periodic promotions; the H3 camera body with the Hasselblad 39MP back can currently be purchased for $32,000. The P45+ back sells for $30,000, including an unusually good three-year warranty and an upgrade program for future backs. A Hasselblad H2 body costs $3,100. You can use practically any view camera with the P45+ and all other medium-format backs. Some sell for as little as a few hundred dollars; others sell for thousands. Some are field cameras made out of wood, while others are monorails made out of various metals, and everything in between. The Linhof I used for these tests is Linhof’s top-of-the-line small view camera. It is a marvel of machining and precision, and as such, it commands a premium price of approximately $8,000. The bottom of the M679 line sells for about $3,000, the main difference being less complicated movements and fewer micrometer controls for the movements.

A final consideration is the cost of film. Although the price of a medium-format back seems high, folks that used to shoot thousands of sheets of 4×5 film per year (at about $5 per sheet, including processing), or those that used to shoot medium-format roll film can amortize the cost of the back very quickly based on the cost of film alone, not to mention adding the cost of high- quality scans and/or proofs.

Test methodology

All images were taken with the cameras mounted on very sturdy tripods and used in “mirror up” mode to avoid vibrations in the reflex cameras. Focusing was done using autofocus in the reflex cameras and, obviously, manually with the Linhof.

The exposures were made using both the built-in meters and handheld meters. I paid very close attention to the histograms for optimum exposures. I also matched the focal lengths (or equivalent focal lengths) of lenses as closely as possible. Most test images were shot at ƒ/8 in an effort to be as close as possible to an optimum f-stop for the DSLR and the medium-format outfits (note that ƒ/8 is not necessarily the optimum f- stop for the view-camera lenses, so in principle, the view camera should have been at a slight disadvantage). All the images shown in this article were shot at ƒ/8, but I did make other tests at varying apertures. All the tests I performed led to identical conclusions, regardless of subject, lighting conditions, or apertures.

All the photographs were shot as Raw files. Raw conversion was performed in CaptureOne Pro for the Canon and the P45+ images and in FlexColor for the Hasselblad images. Each photograph was individually color balanced with the use of a calibrated digital grey card in the scene. All sharpening, color, and contrast controls were set to zero in the Raw converters. The only time I made any adjustments was if a minor adjustment was needed to match the exposures precisely in Photoshop.

My monitors were profiled, calibrated, and warmed up for 30 minutes before performing any comparisons. All objects were illuminated using calibrated GTI GraphicLites, and all prints were dried for at least 24 hours and then viewed in a GTI calibrated Print Viewing Station.

Color

I took a number of photographs of test charts such as the GretagMacbeth Color Checker and the GretagMacbeth color targets for camera profiling. I also took a number of photographs of everyday objects such as flowers, books, fruits, vegetables, fabrics, and color candles.

As described in the above methodology, I placed these charts and objects under lights that were calibrated for both color temperature and brightness. This allowed me to compare the actual test charts and the real objects to their photographic renditions. The photographic renditions were viewed on care- fully calibrated and profiled monitors. They were also viewed as prints under the controlled and calibrated conditions of the GTI Print Viewing Station. I always matched the brightness of the calibrated lights first, so the monitors, the prints, and the objects matched in terms of brightness.

After looking at dozens of images taken with each outfit, the inevitable conclusion is that the medium-format digital backs produce colors that are noticeably more accurate and cleaner than the DSLR. They also produce better saturation, higher dynamic range, and a visibly wider color gamut. These advantages were particularly apparent in deep, highly saturated colors such as the primaries and their complementary colors, green foliage, yellow flowers, and blue skies. They also were particularly apparent in subtle colors such as pastels and skin tones. The Linhof View Camera outfit with the P45+ back was the best of the lot by a surprising margin (see further comments on view-camera lenses below), followed by the P45+ mounted on the Hasselblad H2, the Hasselblad H3D 39, and the Canon, in that order.

Sharpness and general quality

Figure 1a shows the original photograph of a small circuit board, with a hand holding the digital grey card next to it. The Canon did an excellent job of rendering this scene. Figure 1b shows a comparison of the Canon versus the H3. Note that except for some blurring and pixilation, the Canon comes close to the H3. Figure 1c compares the H2 and the H3. The H2 is clearly better. It is noticeably sharper, with wider dynamic range and much more accurate color.

The real shocker for me was when I took the P45+ back off of the H2, installed it in the Linhof, and took the same photo- graph using the same digital back, but a different camera. Figure 1d compares the H2 to the Linhof.

Notice how the Linhof simply demolishes the competition. Frankly, although I had expected to see some increase in quality using the Rodenstock HR lenses, never in my wildest dreams did I imagine that the difference would be of this magnitude. Everything from sharpness to color fidelity to dynamic range to color gamut to plain “snap” and a glowing three-dimensional look is dramatically better with the Linhof. Note that I used the same back in the Hasselblad and the Linhof, so this difference has to be attributed largely to the Rodenstock lens.

I was obviously intrigued and surprised by this first finding, so I took a different circuit board and performed a similar comparison. The results are shown in Figure 2. Again, the Linhof demolished the competition.

I also was intrigued by the fact that the Canon did relatively well in the first test, photographing the circuit board. In the past, I had noticed that the Canon is quite good at rendering objects with many high-contrast horizontal and vertical lines, but not as good with other types of objects. Figure 3a shows a photograph of a backyard with grass and trees. Figure 3b shows a section of that image taken with the Canon. Figure 3c shows the same section of the same image taken with the Linhof. While the Canon image is blurry, lacking in detail, and heavily compressed in dynamic range, the Linhof image is stunning. It is sharp and brilliant. Bottom line, the Canon is simply not in the same league as the Linhof (nor does it cost nearly as much as a Linhof plus a back).

As a final comparison, figure 4 is an image of a scene with a relatively high brightness range. Note that the H3 has a difficult time with the brightness range; the shadows start to get blocked and lose detail, the greens are somewhat muted compared to the real cactus, and the little purple cactus flower loses its brilliance. In general, the highlights seem somewhat compressed.

By contrast, the Linhof image is truly outstanding in capturing the real color and brightness range of this scene. Magazine reproduction may not show this, but the Linhof image is also much sharper and really brings to life all the fine details.

Conclusions

The first obvious conclusion is that the medium-format digital outfits I tested are significantly better than a top DSLR, both in tests and under real-world working conditions. This in spite of the fact that the DSLR tested is quite an accomplished performer. (In fact, before beginning this test, I expected the gap to be much narrower.)

The second conclusion is that, at least in my limited set of tests, the PhaseOne P45+ came out as the clear winner of the two digital backs. I am sure that there are other fine digital backs on the market, but again, the purpose of this article was not to test every single digital back available today.

The third, and to me much more important, conclusion is that the view camera is still at the pinnacle of photographic capture—and by a huge margin. You just about have to see real images “in the flesh” to appreciate how enormous this margin is. A view camera with a good digital back and modern digital lenses simply trounces any other digital capture device.

It is important to note that I did not use any view-camera movements in the test images. With this additional set of capabilities for image management and depth-of-field control, the view camera ends up on an even more exalted pedestal. In fact, with the additional accuracy that digital now offers (perhaps the correct term is “demands”), my personal view is that camera movements are more important now than they ever were in the days of film.

It is fascinating to me that, quietly and almost imperceptibly, there has been a true revolution in view-camera lens design. This revolution has been led by Rodenstock and Schneider. The pixel pitch of the P45+ back is approximately 7 microns. The Rodenstock HR lenses can resolve close to 5 microns. Recently, I was allowed to borrow an upcoming ultra-wide Schneider 28mm view-camera lens. I can vouch for the fact that this lens is incredibly sharp and practically distortion free. Although data on this lens are not published yet, the dealer tells me that final production lenses are expected to resolve 200 lines per millime- ter all the way to the edges. These kinds of numbers would have been unthinkable only a few years ago. But it is not only about sharpness and resolution; it is also about color fidelity, color saturation, contrast, lack of aberrations and flare, consistency from lens to lens, and so on. I am extremely impressed by what these companies have accomplished.

It appears to me that with the exception of view cameras with state-of-the-art digital lenses, all high-pixel-count digital systems are at present lens-limited. What we need most from DSLR and medium-format manufacturers is not sensors with more pixels, but better lenses.

As a final comment, I remember fondly the good old golden days of film. Many photographers used to shoot with 35mm- film cameras and make excellent 8×10-inch prints. They used to passionately preach that a larger-format camera was not necessary unless you made much larger prints, and yet, 8×10- inch prints from medium-format negatives always looked bet- ter. Furthermore, if you have ever seen a contact print from an 8×10-inch negative, you know that prints from smaller-for- mat cameras can’t look like that.

A similar thing happens with digital. You do not need to go to big print sizes to see the difference. Yes, you can make excellent 8×10-inch or 13×19-inch prints from a DSLR, but the same size print from a Linhof/P45+ looks massively better. The DSLR is simply not in the same league.

The most important lesson I learned by performing these tests is that unless I need a very long telephoto, extremely fast capture speed, or extreme mobility, I will always opt for a view camera.


About the Author

Mark Dubovoy
MDubovoy
Dr. Dubovoy is highly regarded as a technical expert in many aspects of photography and printing technology. He is a regular writer of technical articles for The Luminous Landscape and photo technique magazine and is a lecturer at various workshops. His photographs are included in a number of private collections, as well as the permanent collections of the Museum of Contemporary Art in Mexico City, the San Francisco Museum of Modern Art, the Monterey Art Museum, the Berkeley Art Museum and the Museum of Modern Art in Nanao, Japan. He is a partner and Board Member of The Luminous Landscape, Inc., and holds MS and Ph.D. degrees in Physics from the University of California, Berkeley.