www.dpreview.com
www.dpreview.com

Leica S2 against megapixel arms race

Irakly Shanidze | Product Reviews & Previews | Published Feb 27, 2012

 

LEICA S2 vs PHASE ONE/LEAF APTUS II 12 COMPARATIVE REVIEW

PART ONE. LEICA S2

INTRODUCTION

As a working advertising photographer, I have to update my equipment periodically to stay current with technology. Being rather conservative, I tend to do it only when the upgrade cannot be avoided. This time the reason is that my favorite Contax 645AF bought in 2001 is coming to the point when reliability becomes a real issue. Considering that Contax is no more, and the system is no longer supported, continuing its use for commercial work amounts to testing my own luck every time I pack for a shoot. Another issue is an aging PhaseOne P25 digital back with a low-resolution LCD screen and battery life too short by modern standards. Changing from one camera system to another is never cheap. Replacing a medium format setup can easily put financial health of your business at risk. Moreover, make a wrong choice, and you will get stuck for many years with something that does not suit your nature and work style. Hence, it is never simple. With Rollei gone and Hasselblad now using Fuji lenses that never appealed to me in the way they render an image, only two possible choices remained in my price range. I had to decide between ultra-portable Leica S2 with a lineup of f/2.5 lenses and PhaseOne with modular hence flexible design and excellent Schneider optics. The former option would be a no-brainer if it was not for one little thing… It just so happens that a price of the S2 body and three lenses could buy me the PhaseOne camera with three lenses and a digital back that doubles sensor size and megapixel count of Leica S2. That is, if I choose Leaf Aptus II 12 back, because PhaseOne IQ180 is simply too expensive. Knowing that most digital imaging manufacturers are caught up in a “megapixel arm race”, I decided to test whether it is the megapixel count that solely determines an overall image quality of a resulting photograph, or other factors can compensate for a two-fold difference in resolution.
 

THIS IS WHAT I HAD TO CONSIDER:
 

 Leica S2 is not really a medium format camera by traditional standards. Its 37,5 megapixel (7512 x 4992 pixel with 6µm pitch) sensor is 45x30mm, which is a half of the common 645 frame. However, its fast (up to f/2.5) high-performance lenses and proprietary digital signal processing used instead of a low-pass filter are claimed to compensate for the greater DOF and possibly lesser detail forced by the sensor.
 Phase One, on the other hand, is a traditional modular 645 camera (actually, a slightly updated and rebranded Mamiya 645D, which is very popular among commercial photographers in the United States) with a form factor and a control layout very similar to Contax 645AF. The one I had on loan for the test came with Leaf Aptus II 12 (a 53.7x40.3mm 80 megapixel digital back, 5.2µm p i tch) . Since  the  back is interchangeable, it can be later replaced without a need to replace a body. 80mm and 110mm Schneider Kreuznach LS lenses used in this test are considered an industry standard in terms of optical performance at apertures as wide as f/2.8

For a studio test, as an outside color standard, I used a Sinar M / eMotion75 combo famous for its color fidelity and low-noise capture. Sinar eMotion75 digital back features a 48 x 33 mm 33 megapixel sensor with a 7.2µm pixel size. Even though Sinar M is not anymore produced, it is commonly associated with the gold standard of image quality in digital photography.

The camera came complete with Carl Zeiss 80mm and 120mm lenses that are unanimously considered the best of the best in Medium Format photography.

As Sinar eMotion 75 files are not directly compatible with Lightroom 3.6, they were converted to DNG first with proprietary software and then imported into Lightroom for further developing.

DISCLAIMER

Some of the tests performed may seem like hairsplitting, as any fault in color reproduction caused by a digital camera, can be repaired. However, it takes time, and time is valuable regardless of a purpose for which a picture was taken. On the other hand, when it comes to post-processing, it is always better to start with the best image possible. So, my main interest in this test was to see what is the best possible picture that these cameras can take under all possible conditions that I find plausible to encounter in my photographic endeavors.

STUDIO

Lighting: Helsel Integra Pro monolights, Hensel C-light 1000W incandescent lights.
Light modifiers: Hensel 22” (56cm) Beauty Dish, 7” (17.5cm) standard reflector,
3x6’ (90x180cm) strip, 1x5’ (30x150cm) strip.

Leica S2 / Summarit 2.5/70 I decided to try first because it seemed that anyone who has ever shot a 35mm DSLR was capable of operating S2 in manual mode without even looking in the manual. The menu, however, was not as simple as I thought it would be, and short-long-push multifunction buttons was too new of a concept. So, in a retrospect, reading the manual first would have been a good idea.

SENSOR. COLOR FIDELITY

Three-flash set-up, 1:2 main-to-fill ratio
Main light: beauty dish (upper left side of the frame)
Fill light: styrofoam board on the floor, lower right, tilted towards the model
Hair light: 3x6’ strip, above the head, contra-jour.
Background lit with a spill from the hair light.
ISO160, F/11, 1/60 sec, WB set to Flash

Fig. 1     Leica S2, Leica Summarit 2.4/70

Straight Lightroom 3.6 (LR) conversion from DNG, no color adjustments. Subjectively pleasant, however, color temperature is higher than expected.

 Fig. 2     The same shot, auto white balance correction in LR.

The difference in color temperature between Fig.1 and Fig.2 is 1250K. A control exposure with Sinar M /eMotion 75 had the same color shift, which means that it can be attributed to lower color temperature of aging flash bulbs. Auto WB correction was a one click affair, as the color cast has been uniform across the frame. An additional color correction attempt using a white-gray standard (a stuffed cat with a red nose) did not improve the picture compared to the LR auto setting.

SENSOR. SHADOW DETAIL

Fig. 3     100% crop 

As before, noise reduction settings were left at their default values (Luminance 0, Color 25 / Detail 50). The difference in processing was that the Fill slider was moved up to 10 in order to  open up shadows and provoke shadow noise.

Adding fill in LR is a reliable way to reveal latent color noise. However, it did not work in this instance: a texture of the model’s turtleneck is clearly visible with no moire, or digital noise. Also note smoothness of the skin tone and absence of grain and color noise and color shifting in dark spots (under the jawline, between fingers). Despite the lens high contrast, subtle transitions in shadows are clearly visible and also do not show any color artifacts.

LENS PROPERTIES. RESISTANCE TO BACKLIGHTING

A 3x6’ strip used as a hair light was purposefully set less than two feet above the model’s head and turned towards the camera. It caused no visible loss of contrast, ghosting or flare. As always, I did use a lens hood during the shoot.

LENS PROPERTIES. PLASTICITY

 Fig 4.   100% crop 

Analysis of another crop of the same image (Fig. 4) has shown that Leica Summarit 2.5/70 is suitable for portraiture, even though it is a general purpose lens. The way it renders small details on the skin, even at f/11, is rather flattering (Fig. 4a):

Fig. 4a

while strands of hair and eyelashes are sharp and iris patterns are legible, pores and age lines are muted, while general skin texture is still there. This gentle behavior, in addition to its unquestionable aesthetic quality, translates into improved efficiency: when the lighting is chosen correctly, the picture requires a minimal amount of retouching (if any). 

Note that the model’s face on the picture sustained no touch-up whatsoever, and there is almost no make-up. Should I have decided to turn the photo into a beauty shot, literally in less than ten minutes the skin could look pristine. What is really important, retouching of this kind can be completely performed in LR and would not require opening the DNG file in Adobe Photoshop. I try to avoid using Photoshop because high-resolution files, when converted can fill a hard drive very quickly. For example, a 37.4-megapixel Leica DNG file, which weighs 75.8MB converts into a 225MB TIFF. In addition to that, when files this big are imported back into Lightroom, it noticeably slows down. This is especially unpleasant when working with a laptop on location, or on a trip. 

LENS/SENSOR PROPERTIES. OVEREXPOSURE COLOR SHIFTING

One of the most unpleasant situations in digital portrait photography is an unwanted color change in areas occupied by specular highlights (i.e. patches of skin reflecting a light source). Even when the patch is not fully overexposed and still retains its surface pattern, it is unevenness of the texture that creates the problem: tiny areas of skin surface angled in a certain direction act as mirrors and reflect light into the lens. These reflections are often overexposed, and if any of three color channels has a value of 255 (which in most cases means clipping), color shifts around the burned out area occurs. This phenomenon also depends on quality of light: the harder it is, the higher the possibility of the color shift.

Fig. 5     Leica S2, Leica Summarit 2.4/70

DNG, WB set to flash, no color correction

In order to test Leica S2 behavior under conditions described above, I set up an open face beauty dish as a main light and provided no fill. The picture (Fig. 5) shows how Leica S2 / Summarit 2.5/70 combo handles a resulting five-stop lighting ratio on the face. If there is any color distortion in or near highlights, it is undetectable. A situation like this is not unrealistic, even though this kind of lighting setup is not something that an experienced photographer would purposefully  employ on a woman’s face. In instances when lighting cannot be completely controlled (like in available light portraiture, or studio schemes with dramatic lighting) high contrast may be unavoidable. That is when color shifts often become an issue, which in this instance was not the case under conditions designed to simulate this very situation.

I suppose that this ability to withstand high-contrast hard lighting without color distortions is due to a high dynamic range of both the lens and and the sensor.

LENS/SENSOR PROPERTIES. LOW-KEY SUBTLETY

Results of the overexposure test encouraged me to try Leica S2 in low-key lighting arrangement. Low-key is one of my favorite lighting techniques for portraiture. It is visually strong and not that hard to work with, provided a camera/lens combo can handle high contrast and delicate shadow detail well enough.

For the test I built a three-flash setup with a 3:1 main-to-fill ratio.

Main - 3x4’ softbox, fill - 5’ octabox
Background - 7” reflector with a 20° honeycomb grid 
Accent: 1x4’ strip

Camera settings were standard for the studio:

ISO160, f/11, 1/125 sec., WB set to flash.

Fig. 6     Leica S2, Leica APO-Elmar-S 3.5/180

This image (Fig. 6) had to be corrected in LR by boosting fill to 25 and increasing contrast to +70 in order to reveal enough detail in the black turtleneck while maintaining a sufficient tonal ratio. At ISO160 it did not result in any noise. 

Fig. 6a

Plasticity and dynamic range of the lens appeared to be on par with Summarit-S 2.5/70. It handled details equally well on black fabric (Fig. 6a) and the brightly lit right side of the face. Separation from the background is gentle, yet defined. It effectively calmed down a somewhat uneven skin texture despite high-contrast lighting. I had to take out two blackheads caving in to an ultimatum from a model, but did not do anything else to “improve” the skin. (Fig. 6b).  

Fig. 6b

White balance had to be corrected using the dark-gray seamless background as a standard. The difference in color temperature was 1350K.

AVAILABLE LIGHT

 As Leica S2 has been known as an “action camera”, it seemed logical to do an available light test in real life conditions: Manhattan in December. I shot two series of pictures, one in broad daylight (noon, Central Park) and another in mixed light (2-3am, Times Square). In addition to testing overall performance, I paid extra attention whether the camera could maintain believable skin tones in different lighting situations. 

For the entire test my ISO of choice was 640. I did it three reasons: 

  1. To see whether ISO640 is high enough to adversely effect the dynamic range and color saturation.
  2. ISO 640 is extremely convenient for outdoor shooting during the day, because it makes handheld exposures possible up to f/11. 
  3. Because it is higher than usable ISO of nearly any medium format back on the market.

AGAINST THE SUN

A photographic technique used in a picture below is quite common for outdoor available light portraiture in order to turn the sun into a soft light source. That is why I felt compelled to reproduce similar conditions in my test (Fig. 7). 

Fig. 7     Leica S2, Summarit 2.5/70, 1/1000 sec, f/2.5, ISO 640, DNG, AWB, direct conversion to JPEG

In this arrangement the sun acts as a hair light.  Something light-colored in front of the model (in this case, pavement) reflects the sunlight in her face thereby filling shadows. A frequent undesirable consequence of this technique is its tendency to result in burned out patches of hair caused by not enough fill needed to compensate for a substantial difference in illumination of the face and the hair. In my case the difference was at least 5EV, yet the result has been perfectly acceptable. While the highlight is much warmer and brighter than the rest of the hair, it looks believable color wise, and the detail is preserved across the bright patch. At the same time, the skin tone is perfect throughout the whole tonal gradient. Slight reddishness on the nose and chicks is not a distortion, but a result of 43°F temperature.

DIRECT SUN 

Direct sunlight on the skin poses even more challenge than backlighting. Combination of high contrast and hard light quality is the best if you want to turn a female sitter into a victim. Here I decided to make the job for the S2 even more difficult by placing the model on a dark rock serving as a “black reflector” in order to deepen shadows (Fig. 8). Due to very strong contrast, direct sunlight indeed caused a color shift above the left eyebrow and on a patch of the skin adjacent to the nose where the skin looks yellowish and over-saturated (Fig. 9). Yellow cast on the left forearm and elbow is not a result of the described phenomenon. Rather, it is simply a reflex from a dress. Expecting perfect color reproduction in such a difficult situation would be unreasonable, to say the least. Leica S2 did not do a miracle, but got out of this situation with a minimal damage. 

Fig. 8 Leica S2, Summarit 2.5/70, 1/1500 sec, f/2.5, ISO 640, DNG, AWB, direct conversion to JPEG

An unintended benefit of this test became apparent when I looked at the picture at 100% magnification. Note that the model’s head occupies roughly 5% of the frame. Yet, on the 100% crop (Fig. 9) irises and individual strands hair are clearly visible. That is, by the way, on a picture shot from 30 feet, at f/2.5 without a tripod

Fig. 9

AVAILABLE LIGHT. JPEG OPTION

Even though JPEG is rarely my   format of choice, I do rely on it mostly in “fire-and-forget” situations. Like when I need to give files to a client right after a shoot. Giving away RAW files is never a good idea, and in most cases the client does not need them. Uncompressed full-resolution JPEGs are the best option, especially if they are created automatically. 

When I saw that Leica S2 has a dual CF/SD card slot, shooting some JPEGs became a part of the test plan. Unfortunately, if there is a way to make the camera shoot DNG and JPEG files simultaneously to different cards, I could not find it. That meant, in a real life scenario I would have to choose between DNG to JPEG conversion, or JPEG only. Oh, well…

Fig. 10     Leica S2, Summarit 2.5/70, 1/30 sec, f/22, ISO160, JPG, AWB

The first test was for color at ISO160, AWB. It was shot from a tripod at 1/30 of a second simultaneously in JPEG and DNG. The JPEG file was shot with standard camera settings, in sRGB color space. Both files looked identical in LR and did not require additional WB adjustment (Fig. 10).

Fig. 10a

Level of detail can be seen on Fig 10a, which is a 100% crop. As I was not sure how the shells looked in reality at this magnification, I made a 16x20” Epson 9600 print from the sRGB file, which looked realistic.   

Then I tried an action shot. Actually, all I wanted was to see how the camera handles AWB and renders colors on ISO640, and cars just happened to be passing by. The situation was also perfect for testing exposure metering, as the scene was so close to neutral gray. 

Fig. 11     Leica S2, Summarit 2.5/70, 1/23 sec, f/22, ISO640, JPG, AWB

At ISO640 and f/22 the camera correctly suggested 1/25 of a second. Colors did not need further adjustment (Fig. 11). The photo was cropped to about 85% of the full frame and sharpened (amount 28, radius 3, masking 0).

Panning on a relatively slow shutter speed resulted in too much motion blur to make any statements about detail treatment, so I had to shoot one more picture (Fig. 12).

Fig. 12     Leica S2, Summarit 2.5/70, 1/4000 sec, f/2.5, ISO640, JPG, AWB

Similarly, AWB worked perfectly, and the photo was exposed correctly in the aperture priority mode.

Magnified to 100% the picture did not look as pristine at shells that were shot at ISO160 (Fig. 12a): the surface did not look even, and edges of some highlights had slightly visible purple fringing. However, no color noise.

Fig. 12a

In an attempt to get rid of fringing, I found that Leica S2 custom lens profiles were not available for JPEGs.  Therefore I had to switch to a manual lens correction mode. Fringing went away with “Defringe all edges” selected. Setting luminance noise/detail to 28/20 and sharpening to 56/3.0/25/0. I was able to smoothen out an unwanted unevenness on the car surface without losing sharpness. 

ARTIFICIAL LIGHT (THE FUN PART)

Shooting in artificial light, which is not specifically designed for photography, is always a problem. I mean, ALWAYS. Even if all light sources are identical tungsten fixtures,  which is the best case scenario, possibilities for infrared contamination are plenty. If you think this cannot happen to you because you do not own an M8, do not be so sure. Have you ever noticed that fingers and knee caps look suspiciously purplish? That is it, right there, your infrared light in action. And how about when the main light is tungsten, and fill comes from an environment friendly (and very photographer unfriendly) energy saving fluorescent bulb? 

COLOR FIDELITY

Having all of the above in mind, I decided to put Leica S2 to the test in the living hell of   artificial light, called Times Square. In case you don’t know, this place is brightly lit at night by gargantuan LED displays, neon, tungsten, sodium vapor and think-of-a-name-and-it-is-there lights of all imaginable colors. Now, throw into the mix halogen and xenon headlights of cars passing by, and you’ve got the picture. I mean, I have got the picture (Fig. 13). 

Fig. 13     Leica S2, Summarit 2.5/70, 1/180 sec, f/2.5, ISO 640, DNG, AWB, direct conversion to JPEG

I looked at it stricken by a mixture of amazement and disbelief. By all accounts, the skin tone was supposed to be wrong. All I got was a slight red cast on the dark side of the face, which was consistent with the environment and did not look wrong. Before that night, I firmly knew that no camera in the world would automatically set white balance off the eye whites on a picture like this. Now I am not that sure. Considering that only light-reflecting (not light-emitting) surfaces can be reliably used to set white balance, nothing on this photo could be used as the gray standard, except the eye whites. This was the most amazing discovery of the day. At the same time, no trace of red in the model’s black coat, and the eyes are blue. Trying not to think about a possible secret face-recognition module in the Leica S2 digital processing algorithm, I proceeded carry on with the test.

LOW-LIGHT PERFORMANCE 

Next to me a photographer was shooting someone with a professional 35mm DSLR and two Speedlight flashes (one on-camera and a slave on a side). Anyone who has ever shot a medium format camera will understand what an amazing feeling it was to realize that I did not feel stupid by not having a flash. ISO 640 gave me a 1/180 shutter speed at f/2.5, plenty for a handheld shot with a 70mm lens. From the day shoot at the Central Park I already knew that shooting wide open would not influence sharpness, but autofocus was still a concern. The concern was no more when after the first picture I realized that AF speed and precision was just as good as in the daylight. Even for someone like me who is not fond of AF, its implementation in Leica S2 was not annoying at all. In addition to the full manual AF overrige, the camera has a capability of assigning a shutter trigger and the AF function to separate buttons. Just to be fair, I have to mention that people in Solms did not invent it. This feature was first introduced in 1999 in Contax 645AF and then largely ignored by most MF manufacturers. Luckily for me, Leica kept the German tradition alive.

Fig. 14     Leica S2, Summarit 2.5/70, 1/180 sec, f/2.5, ISO 640, DNG, AWB

You might have noticed that on the Fig. 13 the model was sitting at the red table. Next challenge for the S2 was to see how a reflex from the table would affect the skin tone when the face gets closer to red surface. Well, it did not (Fig. 14).

In an attempt to reveal some more detail in blacks, I lifted shadows by setting Fill to 16 in LR. The texture became slightly grainier, but no color noise.  At this point I was almost frustrated that none of my actions could throw this stubborn camera off track.

Running out of options, I decided to change the lens. Summarit 2.5/35 came to mind because wide-angle lenses are notorious in their ability to catch all kinds of unnecessary lights causing internal reflections galore. Also, one of the huge LED screens nearby suddenly went off, which effectively cut amount of available light in two. Knowing that switching to a shorter focal length will make it possible to use a longer shutter speed, 35mm lens seemed like a great idea. The result again was far from what I had expected (Fig.15): excellent color reproduction, great contrast and not a trace of flare.

Fig. 15     Leica S2, Summarit 2.5/35, 1/90 sec, f/2.5, ISO 640, DNG, AWB, direct conversion to JPEG

The latter was especially hard to believe considering the number of light sources of various sizes blasting right into the lens at different angles. 

The last shock I experienced upon blowing up the picture to 100% (Fig.12). Like I said, ISO640, wide-open, no tripod, no flash. I swear!

Fig. 15a

Oh, one more thing… In case you were wondering what is perfect bokeh, take a look again at Fig. 15.

PART TWO. PHASEONE/LEAF APTUS II 12 

My main interest in comparing Leica S2 to Leaf Aptus II 12 was to see if it can in any way compete with a digital back, which, with its 80-megapixel full-frame sensor, offers the highest resolution currently available to a photographer not involved in some super-secret government project. 

I hypothesized that S lenses that reportedly surpass Schneider Kreuznach LS optics in sharpness and resolving power combined with filterless anti-aliasing solution may give Leica some advantage in this otherwise uneven and unfair contest.  

While quite a few factors and camera features may affect image quality one way, or another, I paid the most attention in the studio test to everything that directly concerns image quality. Ergonomics and other advantages that come with a conventional SLR design were not taken into consideration, as in professional photography convenience frequently takes a back seat when image quality is paramount. Also I decided not to discuss obvious differences inconsequential to image quality, such as focal lengths and frame aspect ratios. 

The PhaseOne/Leaf Aptus II 12 studio test ended up to be more thorough than the one for Leica S2. The reason is that tests are frequently driven by findings, and it was the case here. That is why I spent more time on issues concerning color fidelity and opted out of a mixed light test. Also I decided to forego Capture One (C1) software and used Lightroom 3.6 instead because my previous thorough comparison of C1 vs LR did not reveal any advantage in processing in ether application, yet LR offers a more convenient integrated file management and web publishing solution. 

SENSOR. COLOR FIDELITY

Three-flash set-up, 1:2 main-to-fill ratio (same as in the Leica S2 test)

ISO200, F/11, 1/100 sec, WB set to Flash (ISO200 choice was consistent with ISO160 for Leica S2, as it was also the second setting from the base).

Fig. 16     PhaseOne/Leaf Aptus II 12, Schneider 2.8/80 LS, 1/100 sec, f/11, ISO 200, DNG

The first shot clearly did not look right (fig. 16).  Note the area below the red curve. It has a noticeable purple cast, while the rest of the frame is fairly normally balanced. My first educated guess was that it could be caused by a lens hood being to shallow. Shielding the lens from a hair light, however, produced an identical result. 

Fig. 17 AWB in Lightroom Fig. 18 Manual WB off the right eye white

Adjusting white and overall color balance on this photo automatically was not possible. In an attempt to compensate for the purple cast, Lightroom could not balance skin tones correctly (Fig. 17). Setting white balance off a white belly of my stuffed feline color standard did not work either, as it turned out to be exactly what Lightroom did in auto  mode. The third attempt was to use model’s eye white as a standard (Fig. 18). It resulted in even a denser purple cast (Fig. 18a).

Fig. 18a

Further examination of the file was even more unexpected. At ISO200, which is only one stop above its base, level of noise was much too high (Fig. 18b). 

Fig. 18b

The picture was converted with default noise reduction settings (Luminanse 0, Color 25/Detail 50). Fill slider was kept at 0. In comparison, Leica S2 file was noise-free at ISO160, even after Fill slider was set to 10. Moreover, color noise was virtually undetectable at ISO640 either. (Actually, Leica S2 was better even at its highest ISO 1250, but this is off-the-record, as the Leica S2 test officially was performed at ISO640).

The purple gradient covering the lower part of the image and an unusually high color noise level rose prompted a long exposure test to investigate a possibility of an infrared (IR) contamination. The result is on the Fig. 19. 

Fig. 19     ISO800, F/2.8, 1/5 sec, AWB Fig. 20      AWB in Lightroom

This time the photo, did not seem to have a visible purple gradient. However, an attempt to set WB automatically made the skin green again. It was a proof of the purple cast still present (Fig. 20). 

Setting WB off the model’s eye (Fig. 21) revealed the problem clearly. Note colored upper right and both lower corners and a blue cast on the turtleneck. At the same time, the skin looks too yellow in the penumbra.

 Fig. 21      WB set off the left eye white

In an attempt to isolate the problem I tried overexposure (Fig. 22). The shutter speed was 1/10 sec, and the studio was darkened to prevent any stray light, which would make the daylight balanced image so purple. 

 

Fig. 22      ISO400, F/11, 1/10 sec, WB set to Flash

The purple gradient did not disappear. Rather, it changed its direction from up-down to left-right. Incidentally, this photo also reveals the lens vulnerability to effects of stray light and contra-jour. Indeed, loss of contrast and absence of the black point is evident on all photos where a light source directed towards the camera was present.

To make sure that the background did not have a cast of its own, I shot a picture with Sinar M / eMotion75 and the same lighting setup. A photo on Fig. 23 shows light fall-off, but no cast on the background, which is correctly rendered as a color of rust.

Fig. 23     Sinar M / eMotion75, ISO100, F/13, 1/125 sec, WB set to Flash

A picture from Leica S2 was virtually identical in color rendering to Fig. 23. Hence it was no need to show the same image yet again. 

With just one Leaf Aptus II 12 back at my disposal I could not determine with certainty  whether this behavior was typical of this model, or it is a quality control issue. What was clear, however, that color noise was present regardless of the ISO used and that it was increasing proportionate to lengths of exposure. 

Another step in isolating the possibility of the IR leak  was shooting outdoors in cold weather. 29°F suited my needs, as it provided a perfect opportunity to tie purple cast to ambient  temperature. In accordance with my theory, the purple cast  should have vanished, or at least decrease significantly. The picture shows, however, that something was still wrong.

Fig. 24 ISO200, 80mm, F/3.6, 1/500 sec, AWB Fig. 25 WB set off a white patch on a billboard

Fig. 24 is an outdoor photo taken with an AWB camera setting. It has an obvious cyan cast. Custom WB in Lightroom using a white patch on a blue billboard as a standard turned the picture purple. Color noise, however, decresed substantially (although was still evident in the shadows).

LENS PROPERTIES. PLASTICITY AND OVEREXPOSURE 

Fig. 26 is utterly convincing with respect to Phase One / Leaf Aptus II 12 formidable ability to resolve feathers on wings of a bird flying 1500 feet away.

Fig. 26     100% Crop. Note flying birds and a tree with leaves visible on its branches.

This certainly makes this combo an ultimate choice for landscape photography, but how about portraits? Portraiture places special requirements on optics: a lens must be razor-sharp and, at the same time, soft and gentle. This could have sounded like nonsense if it was not for the fact that lenses satisfying such requirements do exist. 2.8/80 Schneider Kreuznach Xenotar, which used to come standard on Rolleiflex 2.8F comes to mind. Also I had a pleasure of working with two lenses of this kind in the first part of our review. 

For the plasticity test we properly converted and examined photos shot with a Schneider Kreuznach 2.8/110 LS lens wide open (Fig. 27, 28) and at f/11 (Fig.30). Lighting setup is similar to the one employed on Fig. 4. 

Fig. 27     PhaseOne/Leaf Aptus II 12, Schneider 2.8/110 LS, 1/8 sec, f/11, ISO 200, DNG

I assessed the manner in which the lens renders the skin at full aperture on the  50% crop  (Fig. 28). Skin rendering seems to be  very precise in highlights. It looks texturized is well lit areas, yet flat in shadows.

Fig. 28     PhaseOne/Leaf Aptus II 12, Schneider 2.8/110 LS, 1/8 sec, f/11, ISO800, DNG

In comparison with a close-up on Fig. 31, it looks too harsh. Note that dark details (eyelashes, eyebrows, hair) are rendered much more delicately, however. It is opposite to what I have expected to see in this test. 

Color correction, still imperfect, for the photo on Fig. 27 required rather extensive tweaking in Lightroom (Compare to the original file on Fig. 29). The final settings were as follows (only changed values shown):

 BASIC: HSL: DETAIL: LENS CORRECTION: 
ColorTemp 2850K
Tint +28
Blacks 9
Brightness +50
Purple saturation -100
Blue luminance -100
Sharpening Amount/Radius/Detail/Masking 58/3.0/25/0
Noise Luminance/Detail/Contrast 13/50/0
Noise Color/Detail 94/17

enabled/custom

Lens Profile:  Schneider Kreuznach 2.8/110 LS

Two brightness gradients were added symmetrically on both sides with value of -40 (left) and -32(right).

 

Fig. 29      Original file, AWB

Fig. 30     Schneider 2.8/80 LS, 1/10 sec, f/11, ISO 400, DNG, 100% crop

 Behavior of the 2.8/80 lens stopped down to f/11 is shown in detail on Fig. 30  (approximately 5% crop of a full frame), which was shot with strobes. Rendering appears to be rather indiscriminate with respect to light and dark details: pores and fuzz are well-defined across  the tonal gradient up to the point of 3EV overexposure where details vanish abruptly due to clipping. The clipping, however, did not cause any color shift. Stopping down seems to increase microcontrast thereby improving the lens capability to resolve small details in darker tones. 

Overall contrast in both lenses used during the test was strongly affected by stray light, which did not seem to improve with stopping down. In my opinion, hoods supplied by Schneider are too shallow for the corresponding focal lengths.  

Fig. 31      Leica S2, Summarit 2.5/70, 1/60 sec, f/11, ISO 160, DNG

For a general color/detail side-by-side comparison of both cameras please refer to an image on Fig. 31 (approximately 10% crop of a photo on Fig. 2). It was shot with Leica S2 in nearly identical conditions: 1/60 sec, f/11, ISO 160, DNG and the same lighting setup with a main-to-fill ratio of 2:1 (the ratio on Fig. 30 was 3:1). Overall contrast is high with no trace of flare. The same level of detail is maintained across the whole tonal gradient with an emphasis on darker tones. Rendering of the skin is smooth, yet darker elements are in sharp focus. Skin appears to look realistic, while other colors are properly balanced. This photo would take me a half of an effort to retouch than the one on Fig. 28, or 30, not even considering dealing with color correction. 

LENS PROPERTIES. RESOLVING POWER

For Leaf Aptus II 12 this test was an opportunity to leave Leica S2 in the dust. Indeed, an 80 megapixel sensor met its match in the face of 2.8/80 Schneider Kreuznach LS lens. If you noticed on Fig. 25 a tiny speck in the sky and thought that it was probably dust on the sensor, you made a big mistake. Just look at Fig. 26 and see for yourself. Such resolving power at less than a stop from its wide open setting is not something that you can see on every corner. 

It is important to remember that this result was hardly possible without a tripod. This, however, would not be a fault of the lens.  With sensor resolution this high, motion blur becomes a real issue, even at infinity and relatively high shutter speeds. The building was approximately 1500 feet from the camera. On a such distance, shooting without a tripod simply was not an option if we wanted to see lines between bricks on the building.

CONCLUSIONS

This exhaustive two-part test was designed to assess color fidelity, image quality and features of Leica S2 and PhaseOne /Leaf Aptus II 12 combo. The goal was to find our whether these two similarly priced systems are otherwise comparable in situations frequently encountered by professional photographers.  

Leica S2 features and design make it an obvious choice for location work. Yet its performance in a studio people photography was excellent. Perhaps Leica’s tradition of building cameras well-suited for low-light work was a reason why S2 viewfinder is so bright. It was very helpful in a dimly lit studio. As far as image quality, Leica S2 gets a gold star.  Even when white balance is not perfect it takes almost no effort to correct it. AWB reliability was beyond expectations. S2 managed to balance colors correctly is one of the most difficult unintended instances of mixed light that I personally have ever encountered. Regardless of conditions, time and again Leica S2 produced noiseless, color balanced DNG files, even at ISO640. The only area, which in my view needs  improvement is JPEG mode. Rendering of textures in JPEG mode at ISO above 160 could be better. It would be helpful to have custom Lightroom lens profiles for JPEG files. An option to write DNG and JPG files separately on different memory cards is a must. I was surprised not to find it among otherwise well-thought-out list of menu items.   

Leica S2 seems to be perfectly suited for demanding people photography such as fashion, commercial portraiture, etc for two reasons. First and foremost, its excellent optics, which is delicate on the skin, yet meticulous in representing small details elsewhere, regardless of an aperture setting. Second, Leica S2 workflow is simple and uncluttered, more reminiscent of a pro-level 35mm SLR than a medium format camera. Shooting in available light on streets of Manhattan made me fully accept Leica’s claim that you will never need a 35mm DSLR once you choose Leica S2. Being a die-hard user of Contax 645AF since 2001 due to its excellent line-up of fast Carl Zeiss lenses, clever ergonomics and full integration with digital capture, I must concede that, at any available sensor resolution, its the only advantage over Leica S2 is modular design, which makes it more convenient for studio product photography. Outside of this highly specialized field, Leica S2 is far ahead in every aspect.

PhaseOne /Leaf Aptus II 12, in contrast, immediately presented us with a host of color problems that would be disruptive, if not detrimental, to a regular commercial photo shoot, when tethered. We used an untethered mode, therefore surprises came later. Analysis of test results led us to a conclusion that the Leaf back suffers from internal IR leak, or some electronic malfunction, which manifests itself similarly to the IR contamination. A symptom of the problem, the purple tint, was not uniform, which made auto color correction impossible and required a substantial effort during editing. In addition, noise at ISO200 was high enough to raise a serious concern. 

Where PhaseOne /Leaf Aptus II 12 comes back with vengeance is resolution. Image detail in our outdoor infinity test was mind blowing. The standard Schneider Kreuznach 2.8/80 LS lens resolved lines between tiles on a building standing more than 1500 feet away. However, such formidable resolving power appeared to have a flip side: the cameras rendering of human skin is harsh and indiscriminate, ether wide open, or stopped down.

Irakly Shanidze, Copyright © 2012