by Mark Segal and Nick Devlin
In this corner….
The upstart challenger, the Pentax 645D. A comeback kid who the skeptics couldn’t keep down. Weighing in at a feathery $10,000, this contender from the wrong side of the tracks has come to shake up the MFSLR world…
…and in this corner:
The Phase One P40+ on the Phase One DF camera body. This Danish welterweight aims to continue its reign as the king of quality in the 40MP weight class.
Let the fun begin….
Two friends, two cameras, one brick wall.
So after a couple of weeks with the 645D, I was going cross-eyed trying to figure out if what I saw was good, great or not quite right. I needed a comparison. So I called up my friend Mark Segal who lives not too far up the street and happens to own a P40+ mated to the new(ish) Phase One DF body. Being a social scientist at heart (and by profession), Mark was game to do a little head-to-head shoot-out. This was a friendly affair, meant to show what we could bring back from a morning out at the same location, pointing our cameras at the same things. Nothing too fancy or scientific. No claims to rival DxO, or any of the professional reviewers who make their living shooting the same damn bridge or plant over and over So take the results as you will, but not as gospel.
We decided that the best thing to do was to each write our own story, and present them one by one. By virtue of age and beauty, Mark’s comes first.
Between them Mark and Nick have about sixty years of serious / professional photographic experience. Both are regular contributors to this site, and I have a very high opinion of their photographic perspicacity and technical acumen. So, while not gospel by any means, you can regard what they write about photography and equipment with a considerable degree of trust – Michael
I received an email from Nick about a week beforehand saying that he remembered that I have a Phase One P40+ system, and would I like to help with a test shoot for comparison with his newly acquired Pentax 645D. Of course it was irresistible to learn first-hand what a system costing about sixty percent less than my P40+ system could do by comparison. So we arranged a shoot at the Toronto Brickworks, where there is an abundance of textures and lighting conditions suitable for testing sharpness, resolution, and dynamic range.
The simple philosophy behind these tests is that what matters most to owners of these systems is final outcomes – finished photographs. Sensor, lenses, firmware, and post-capture processing combine in the judgment and hands of the photographer to make photographs; the raw material is the raw image, the foundation for the finished product. Our approach is shamelessly visual – there are no test charts and no measurements in this report. It’s about what the photographs look like, and what they tell us about the quality of the equipment and software used to produce them.
Most raw images emerge in conversion software needing sharpening, luminosity and colour adjustments, especially as we recommend to begin post-capture processing with software default parameters all set to zero and White Balance “As Shot”. This way we see what the image contains without the “default” appearance suggested by the software engineers. As David duChemin so aptly puts it in his excellent book “Vision and Voice”, “Most of the time I want to begin with a fresh canvas. I don’t want my camera or Lightroom interpreting my image for me.” So after zeroing the file, as David puts it: “Your image should now look underexposed, bland and lifeless (ed. and we would add – unsharp). It’s a depressing place to begin, but it’s the cleanest canvas possible.” As well, it allows us to better visualize the total scope for image adjustments.
Our objective is to produce these clean canvases from our medium format cameras and subject them to just enough post-capture processing, allowing us to judge, visually, which has superior photographic properties – principally sharpness, resolution and dynamic range. By sharpness we mean edge definition or crispness, and by resolution we mean the ability to resolve fine texture detail. We are less concerned about contrast and colour, because no two camera models produce the same look “out of the box” when opened in raw converters – especially different ones – and those properties can be edited to produce the image appearance we wish – provided the raw material is good enough to begin with.
Turning to the tested equipment, the Phase One P40+ system consists of the said digital back, the Phase One 645-DF camera, and Phase One lenses; 75mm~150mm zoom, 45mm wide-angle, and the 120mm Mamiya macro lens. The Pentax 645D system consists of the camera, with the following Pentax lenses: 45-85mm zoom, 45mm and 120mm macro. For fun, we threw in the Canon 1Ds MkIII with the Canon 24mm~105mm zoom lens and the Leica M9 with the 50mm f/1.4 Summilux lens. We wanted to see how the cameras performed with each of these lenses. Most of the shots were intended for the sharpness and resolution comparison, but one batch is devoted to dynamic range.
The Brick Wall and Blue Door Set
We shot this set with the medium format cameras equipped with their zoom lenses, and with the Canon 1Ds MkIII and Leica M9. Figure 1 shows the full scene, captured by the Pentax (frame 787, f/18/ 1/8th sec., ISO 100).
The capture settings are not necessarily identical between the two cameras for any of the images. That was intentional. We were not aiming to see what each camera does at the same settings, but rather to see how they perform at their best using the settings we believe are optimal for each camera and lens. The settings, however, are not much different. For example, in the comparison discussed here, both exposures are at f/8, but each camera selected its own shutter speed. The base ISO of the Phase One P40+ is ISO 50, while for the Pentax it is 100. Logically therefore, the Pentax shutter speed is higher than that of the Phase One. Both cameras were mounted on tripods.
I fined-tuned the white-balance and exposure of the Pentax image in Lightroom 3.2, exported it to Photoshop and applied High Resolution Capture Sharpen and Inkjet Output Sharpen at 360 PPI Glossy.
I selected frame 1680 of the Phase One series (f/8, 1/5th sec., ISO 50) white-balanced and exported it from Capture One to Photoshop, where I tuned luminosity and vibrancy to more or less match the Pentax shot and applied the same sharpening. At this point, we observe that there are differences of raw capture interpretation between the Phase One shots seen in Capture One and the Pentax images seen in Lightroom 3.2, but, with one exception discussed below, I used Capture One for rendering the P40+ images, as there is no profile yet for this back in Lightroom.
Tuning white-balance and luminosity is not strictly necessary for comparing resolution and sharpness between well-exposed images. I did this anyhow for the simple reason that tonality can influence the viewer’s perception of sharpness and resolution; hence I put the tonality of the comparison candidates into comparable range. I opted for low-ish contrast renditions to insure that micro-detail would not fall victim to saturation or luminosity clipping.
We can now compare the resolution and sharpness of the two images, but as mentioned, for fun, we also shot the same scene with our Canon and Leica cameras.
Figures 2 and 3 show center detail from the Pentax and Phase One cameras respectively.
Examining these images close-up on a NEC PA271W high-resolution wide-gamut display, we are of the view that the detail rendition seen in Figures 2 and 3 is close, with a slight edge in favoring Pentax.
Figures 4 and 5 repeat these comparisons, but this time in the upper left corner, for the Phase One and Pentax cameras respectively.
This comparison tells the opposite story: the Phase One image shows sharper rendition of edges and texture than does the Pentax. Nick and I both observed the same differences, the only difference between us being that he thinks they are more substantial than do I. We cannot be conclusive about the cause without further testing. The two possibilities are (i) real differences in the performance of the lenses (we were at about the same focal length) or (ii) slight differences in camera alignment relative to the wall which show in these images as depth-of-field constraints, notwithstanding that the images were made at f/8. In Nick’s write-up below, Nick explains what he thinks may be happening and it sounds entirely plausible. Turning now to the Leica and Canon comparisons, Figures 6 and 7 show the center and corner details for the Leica M9, while Figures 8 and 9 do likewise for the Canon 1DsMkIII.
While the size of the Leica file is smaller than those of the medium format cameras, and therefore display smaller at 100% magnification on screen, we observe that the Leica files have the best sharpness and resolution of all compared here.
The Canon 1Ds MkIII images are not competitive with any of the others presented above, but I say more about this at the end of the article.
High ISO Performance
We ran a comparison between two medium format cameras for high ISO performance. The apples-to-apples comparison is limited to ISO 800, because that is the limit of the Phase One system using the sensor in 40MP mode. The Pentax is capable of ISO 1600. The place to look for noise in this scene is in the lower right corner under the bench. As a general comment on noise, both cameras are remarkably good performers at this speed. Noise reduction is applied at Capture One’s default setting for the Phase One image. It is applied to the Pentax image in Lightroom for default colour noise reduction only.
This comparison indicates that the Pentax image is both cleaner and better defined in this lower corner at ISO 800 compared with the Phase One P40+ image. Even at ISO 1600 the noise performance of the Pentax sensor appears to be very, very good.
We focused on this potentially difficult area of the image because the rest of the image simply didn’t reveal any prospect of a noise issue.
Dynamic Range Comparison
Within the same complex where we made the above images, we sought a location with a suitable mixture of indoor and outdoor lighting to test performance in extremes of highlight and shadow in a single capture without performing HDR blends. This is an area where the post-capture processing capability of raw converters and image editors becomes critically important. One doesn’t begin to know the whole story simply by opening such an image in a raw converter, because – taking this photo for example – it would look something like this (f/8; 1/60th sec., compensation -2, ISO 50 (an ETTR just below sky clipping):
It so happened that the “High Dynamic Range” tool of Capture One (basically Highlight and Shadow tone recovery sliders) were completely inadequate for making a usable raw conversion, either on its own or complemented with other luminosity tools in that application. Therefore notwithstanding that Lightroom 3.2 does not yet have a bespoke profile for this image format, I instead converted the image in Lightroom, which has excellent tools for these situations, and uses a generic profile for its unprofiled formats. The result and the settings are shown in Figure 14 below.
Figure 15 shows the processed result from a gamma shift in Luminosity Mode done in Photoshop to boost the contrast of the indoor area, the outdoor area being masked.
Nick’s camera positioning and exposure technique was not quite the same as mine for this shot – he had more outdoor content, so I cropped Nick’s image to more closely match the Phase One image in post-capture processing, also done in Lightroom using a similar approach to the settings. Nick’s Pentax image (f/11, 1/80th sec., ISO 100, 0 compensation) required considerably more highlight adjustment and approximately the same shadow area adjustment as my Phase One image, because none of Nick’s images in this series avoided some clipping of the sky. It also got a contrast boost to the indoor area (with the outdoor area masked) in Photoshop.
Essentially the Phase One image ended-up with a better achieved luminosity balance between the indoor and outdoor areas, including a more satisfactory rendition of the sky. It would have required more exotic techniques to replicate the Phase One result in the Pentax image, and even then, it’s not clear that it would have succeeded. The net difference between the two exposures, all factors included, is only +0.75 f/stops for the Pentax image. Hence while it would be tempting to conclude that the Phase One system handles extremes of luminosity better than the Pentax, we believe the jury is still out on this one till we’ve tested further with alternative exposure strategy – particularly for the Pentax.
Figure 17 above is the selected Pentax (0834) shot at f/11, ¼ sec., ISO 200. The Phase One comparator is Figure 18 below, shot at f/8, 1/5th sec., ISO 50 (There are many more bikes in Nick’s shot because he was ahead of me and in the interval a school of bikers flooded the space). Comparing these images, barrel distortion is quite similar. Center sharpness is compared in Figure 19: Seen at 100% screen magnification, sharpness and texture resolution are comparable. Figure 20 shows edge sharpness: Here again, seen at 100% screen magnification, the results are very similar, perhaps with a very slight edge favouring the Pentax image. It would require very large sized prints to see any real difference of sharpness or resolution between these images in print. Figure 21 is a 100% screen magnification of a small center segment of a piece of lumber photographed at close range with the Pentax and Mamiya 120mm macro lenses. The Pentax shot was f/16, 1/4th sec., ISO 100; the Phase One shot was f/8, 1/60th sec., ISO 100. The hues of the raw images differed enough that I thought it desirable to equilibrate them to a point that the differences would not distract from examining sharpness and resolution. The equilibration was performed primarily on the Phase One image, making moderate changes to hue and luminosity, without colour or luminosity clipping and ensuring not to lose any detail in the process. Seen at 100% screen magnification, the differences of sharpness and resolution between these images are very slight, with a very thin advantage to the Mamiya lens used on the Phase One camera; I have not printed these images, but I expect these differences would be largely unnoticeable in a large print of the full image. While this set of image quality comparisons is not exhaustive, the visual evidence suggests that on the whole these medium format camera systems perform more or less equally, with small margins of advantage attributable to the one or the other, depending mainly on the imaging conditions and the lens. Considering the price difference between them and the vast ergonomic and interface superiorities of the Pentax camera relative to the Phase One 645DF, the Pentax system looks very promising indeed. That said, two key factors potential purchasers may wish to ponder are (i) the newness of the Pentax camera means that factors such as durability and support are to be observed over time, and (ii) the range of lenses now available for this camera remains to be further developed. Perhaps the other most striking outcomes of these comparisons are the excellence of the Leica M9 images and the comparative weakness of the Canon 1DsMkIII images. Perhaps this latter observation invites some caution about generalizing a whole condition from an image of a brick wall, because in day-to-day use, I find that my 1Ds MkIII images, captured with the Canon 24~105mm f/4 L lens do emerge from post-capture processing into print with fine resolution and edge sharpness where it matters – as damn well it should considering the cost! I have so many images which seem to underscore this point, so I thought I should provide an example of what I mean: Every detail of the fine cotton wool, the skelton and the spider shows very sharply and well-defined in this image. Depth of field can be a huge issue, as we all know. Here’s a case shot at close range, f/4, where I accidentally focused on the clothing rather than the eyes: The texture of the clothing is tack-sharp, but the eyes…….pilot error, not Canon’s. Well, we had a hiccup on location with the Canon, whose lesson of experience is simple: “Be really, really careful”; but this story is mainly about medium format. Nick and I both think these comparisons are useful indicators that the Pentax camera has definitely broken new ground in the medium format “price for quality” arena.
45mm Wide-Angle Lens Comparison
120mm Macro Lens Comparison
Figure 17 above is the selected Pentax (0834) shot at f/11, ¼ sec., ISO 200. The Phase One comparator is Figure 18 below, shot at f/8, 1/5th sec., ISO 50
(There are many more bikes in Nick’s shot because he was ahead of me and in the interval a school of bikers flooded the space). Comparing these images, barrel distortion is quite similar. Center sharpness is compared in Figure 19:
Seen at 100% screen magnification, sharpness and texture resolution are comparable. Figure 20 shows edge sharpness:
Here again, seen at 100% screen magnification, the results are very similar, perhaps with a very slight edge favouring the Pentax image. It would require very large sized prints to see any real difference of sharpness or resolution between these images in print.
Figure 21 is a 100% screen magnification of a small center segment of a piece of lumber photographed at close range with the Pentax and Mamiya 120mm macro lenses. The Pentax shot was f/16, 1/4th sec., ISO 100; the Phase One shot was f/8, 1/60th sec., ISO 100. The hues of the raw images differed enough that I thought it desirable to equilibrate them to a point that the differences would not distract from examining sharpness and resolution. The equilibration was performed primarily on the Phase One image, making moderate changes to hue and luminosity, without colour or luminosity clipping and ensuring not to lose any detail in the process. Seen at 100% screen magnification, the differences of sharpness and resolution between these images are very slight, with a very thin advantage to the Mamiya lens used on the Phase One camera; I have not printed these images, but I expect these differences would be largely unnoticeable in a large print of the full image.
While this set of image quality comparisons is not exhaustive, the visual evidence suggests that on the whole these medium format camera systems perform more or less equally, with small margins of advantage attributable to the one or the other, depending mainly on the imaging conditions and the lens. Considering the price difference between them and the vast ergonomic and interface superiorities of the Pentax camera relative to the Phase One 645DF, the Pentax system looks very promising indeed. That said, two key factors potential purchasers may wish to ponder are (i) the newness of the Pentax camera means that factors such as durability and support are to be observed over time, and (ii) the range of lenses now available for this camera remains to be further developed.
Perhaps the other most striking outcomes of these comparisons are the excellence of the Leica M9 images and the comparative weakness of the Canon 1DsMkIII images. Perhaps this latter observation invites some caution about generalizing a whole condition from an image of a brick wall, because in day-to-day use, I find that my 1Ds MkIII images, captured with the Canon 24~105mm f/4 L lens do emerge from post-capture processing into print with fine resolution and edge sharpness where it matters – as damn well it should considering the cost! I have so many images which seem to underscore this point, so I thought I should provide an example of what I mean:
Every detail of the fine cotton wool, the skelton and the spider shows very sharply and well-defined in this image.
Depth of field can be a huge issue, as we all know. Here’s a case shot at close range, f/4, where I accidentally focused on the clothing rather than the eyes:
The texture of the clothing is tack-sharp, but the eyes…….pilot error, not Canon’s.
Well, we had a hiccup on location with the Canon, whose lesson of experience is simple: “Be really, really careful”; but this story is mainly about medium format. Nick and I both think these comparisons are useful indicators that the Pentax camera has definitely broken new ground in the medium format “price for quality” arena.
Mark has done such a thorough job of summarizing our findings that I am content to mostly add an observational me-too. I want to caution everyone at the outset, however, that you cannot judge anything about the colour, contrast or even density of these images from the web. I have now viewed the article on five different displays, and all are significantly different. Therefore, please follow our words, not the pictures, on these issues.
Forgive me for mounting my favourite hobby-horse before riding into battle. Mark is a relatively new Phase owner, but he has shot with the camera a lot more than I have with the 645D. Even so, it was striking how much more quickly I was able to make the camera do what we wanted: change ISO, change exposure mode, nail the histogram, flip to self-timer, bracket matching frames at different exposures through different f-stops and ISOs, change batteries, etc. (Oh wait, that was only Mark fumbling for AA cells. I had another 452 shots left to go )
While daily use of the camera would no doubt close the gap on our working speeds, the raw intuitiveness of the Pentax really shone. The accessibility of the controls and the ease/visibility of information feedback on the Pentax just make it a much easier camera to use. We both agree that the form-factor of the Pentax makes it more amenable to hand-holding, much as one prefers to use these beasties on a tripod.
As an opening note, it’s also worth mentioning that the built-in levels on Mark’s $1700 Arca SwissCube tripod head [aka the “the Borg Mothership”] was a passable susbtitute for the electronic level feature on the 645D J.
The Basic Brick Wall
The bottom line on our tests was that the Pentax 45-85mm and 80-160mm matched up pretty well to the 75-150mm Phase One lens (at least at around 80mm). To my eyes, the 75mm f2.8 Pentax beat both of them. We’re talking about counting eyelashs on a cat here, but there was a slight edge to the fixed focal length Pentax on this test, definitely confirming that it was the best $175 I have ever spent on Ebay. The detailed segments are found in my original review, for those who want to take a closer look.
However, something curious was going on with both cameras for a number of the frames. With the 75-150 on the Phase, and the 80-160 on the Pentax, I found that each was sharper in different parts of the frame. In my examination of the images, the Pentax was sharper on the bottom of the image and on the right side, whereas the Phase showed the converse of this. Because we are seeing uneven results from one side of the image to the other, on both axis, I don’t believe this is an optical issue. It would be too much of a coincidence if these two optics had exactly complimentary de-centering (if that could even cause this).
I am guessing that this may be a function of the camera backs not being perpendicular to the wall. The “sharp” Pentax corner detail can be seen here in my original review. Unlike Mark, I think the Pentax 100% image in his corner comparison is visibly softer, and I don’t want this image to start a war. I suspect this has something to do with the plane of focus on both axis. My conclusions is buttressed by the fact that in the bottom right corner (visible in Mark’s illustrations on High ISO), the Phase has sharp detail on the front of the bench, whereas the Pentax is sharp at the wall. This, I hypothesize, is due to the “swing” on the film plane of both cameras, noted above.
Mark, I should emphasize, neither agrees with me that the image is not “sharp”, nor that ‘film plane’ angle could be causing this. Bottom line, who knows? We both agree that the difference in a printed image would be minimal.
The fact that such slight variances could have such a dramatic impact was a shocking result to me. Neither lens is defective, this was a relationship-to-subject error. The moral of the story is that an astonishing level of precision is needed to get 100% quality out of these instruments. The same is probably true with smaller formats, to a smaller degree, but we are not used to seeing the subjects in such breath-taking clarity at 100% magnification as to really notice the difference. I suspect that it’s is a form of ‘user error’, though not one for which I can perceive any ready fix, given we are talking a swing off-axis of a few millimetres. It’s like using a view camera again (?!) While DOF should cover off an minor angularity on the plane of focus at these focal lengths at this distance, it has been suggested that DOF with digital is not the same in practice as with film. If anything, these images might confirm that.
Other lens match-ups
The 45mm Phase did not beat the 45-85mm Pentax zoom set at 45mm. This was a bit surprising in principle, but not entirely unexpected, since the 45-85mm had already proven itself in my testing. The fact that it had an ever so slight edge over the Phase lens (Mark and I agree on this) was curious. This could be down to some sort of focusing error, but at that distance on a 45mm lens at f11, I doubt it.
As between the 120mm macros, there was a single frame on which the Phase lens was better, but on the rest I could detect no meaningful difference – and we are talking about 100% peeping at a micro-detailed subject. I processed the files in LR3, and found that the file looked better than it does in my output than in Figure #21, above. My version is below. This shows how much of “apparent sharpness” is a function of micro-contrast relationships as varied in post-processing.
All in all, I am extremely pleased with the performance of the Pentax lenses in this test. For the price, it’s an outstanding result.
I would pause here to observe that the price of MF lenses is getting a bit off the hook. Any zoom, macro or super-wide angle lens will now run you north of four grand new, with fixed lenses closer to $2,000. That seems a bit nuts to me, considering that Pentax was able to sell 645 FA lenses for 1/2 of that not that many years ago. Perhaps the small quantities of manufacture and the decreased tolereances required for digital account for the gap.
The 35mm wild cards in the mix
Will anyone flame me if I say the Leica is near perfect? You see, it kind of is. Put a 50mm Summilux ASPH on an M9 on a tripod, focus carefully, and oh-my-gawd. Aperture is largely irrelevant beyond depth of field.
The two 40MP cameras were obviously visibly “better” at 100%, because they have double the data and thus render more detail in the subject. On screen, at max-peep, there is no denying the quantitative superiority of the larger image. But qualitatively, the Leica was just lovely within its own megapixels.
By comparison, the Canon image was really weak when viewed at 100%. In fact, it was so much worse than the other images, that I echo Mark’s concern as to whether something was amiss beyond optical and imaging mediocrity. The image looked like it had a thin layer of haze on it, and the most aggressive post processing could not bring it to approximate anything near the accutance in the Leica frame. The difference was both qualitative and quantitative. Just. Plain. Bad. Even as a disgruntled former Canon user, I expected much better. I know the system is capable of better than what we saw in this particular test, as Mark has illustrated at the end of his article. That said, I am not prepared to make as much of an allowance for the Canon as Mark is. We pointed this state-of-the-art camera at a highly textured wall in near-daylight, pressed the focus button and fed it a full serving of photons. Why this garbage-out? If nothing more, the Canon result should lead everyone to cut MFSLRs a little slack when things don’t always go right. Or just buy a Leica. (kidding!)
Still, in a quick but careful test, there were a lot of anomlous results. Real photographers in the real world work the way we shot this test: carefully, but not to any standard of scientific variable-control. These results are a salutary reminder that, while digital has made it that much simpler to capture an image, it has not made it particularly easy to capture a technically perfect image.
A print is worth a thousand pixel peeps
But the story does not end there. I printed each of the shots at 13×19″. In print as on screen, there is little or nothing between the Phase and Pentax shots, besides a slightly different rendition of the brick’s colour pallete.
The real surprise, however, was the Leica print. At this size, at 300 dpi, we are just outside the M9’s native resolution. Even on close inspection, the Leica achieves upwards of 95% of the quality of the MFSLRs. Make no mistake, the Leica will fall steadily behind as enlargements increase. I have found that 24×30 inch prints look nice off the Leica, but are definitely showing the strain of serious enlargement, whereas the P40+ and 645D won’t break a sweat at this magnification.
The take-away is that the benefits of MF, setting aside perspective and “look” in real-world work, are not enormous at most working output sizes for most photgraphic applications. Our particular test subject played into this equation as well. Being highly detailed, it offered an optimal test of resolution. However, its highly varied tonality was more forgiving of tonal transition. In a tougher situation, such as portraiture, where the subject features large areas of relatively similar tonality (eg: skin), the MF cameras will separate tonality more smoothly, just as they did with film. But again, the margin will not be that great at smaller sizes.
The other interesting result from a print analysis is that the ‘oh-so-awful’ Canon frame was really quite acceptable on paper. The difference was certainly more apparent than between the Leica and the MF prints, but by a much slimmer margin than the 100% screen views would have predicted.
Small images mask a myriad of sins. Though in fairness, saying 13×19″ is “small” is a relatively recent digital conceit. I well remember that in the days of film, cracking out the 16×20 trays in the darkroom was a real event. The ease of printing in the digital age has spoilt us, just as has the gloss of an iPad. Maybe IQ matters less than we want to admit?
The Phase back renders ‘richer’ colour than the 645D. Now, that’s not to say that this is more accurate. But think of it as Provia vs E100s. It’s a much finer difference than that, but you get the idea. Good post-capture processing can take either image to where the other is, mostly. A good camera profile could probably eliminate the difference. Still, I give the Phase a slight edge in colour. Mark correctly points out that the Phase is just more saturated, not more accurate. But hey, I’m a Velvia kind of guy at heart, so I prefer the Phase colour.
Unlike Mark, I can draw no comparative inferences from the files we shot. Without dead-standardized exposure matches, I don’t feel comfortable saying one camera is better than another. When we are talking about deep shadows, a little bit of extra exposure means a lot. One can double the amount of data in the bottom bits in a hurry, so until we get to re-shoot this, I don’t want to make any pronouncements on this issue. I am, in any event, highly skeptical of the average users’ ability to test DR. We depend on a JPEG of the capture, produced in-camera, to show us a histogram, which limits the accuracy with which we can achieve optimum exposure. Adding or subtracting a little exposure can make such a big difference that what we get out of the camera is as much a comment on technique and user interface as it is on DR. I’ll just go with DxO’s conclusions on this issue, since they put the cameras on a proper test-bench.
I would only add that that, in my experience, Phase backs are the reference standard in DR. They capture tremendous tonal richness and ‘deepness’. If the Pentax comes close I will be happy.
And the Pentax does a very good job in and of itself. In my main review there is an image shot at dusk in New York, which includes both brightly lit clouds and deep shadows at street level. Shot at ISO 800, the image captures both ends of the scene. Although the shadows are undeniably noisy, you can almost read the writing on the cards on the memorial attached to the fence. That’s pretty convincing to me.
The Pentax wins the ISO race handily. To my tastes, the Phase back tops out at ISO 400 for usable noise. The Pentax is entirely usable at ISO 800, and even at 1600 under the right conditions. The illustrations above have been brightened, and to my eye show the noise more pointedly than it appears when I process the files, due to what appears to be increased contrast. As Mark mentioned, there has been almost no noise reduction applied at all to these, and the noise would largely disappear from the ISO 800 Penatx file with just a little treatment.
This is less of a ‘big deal’ in terms of the comparison between these cameras than many might think. In simple terms, the P40+ is not a speedster camera, nor was it meant to be. The native ISO is 50. It is meant to be used on a tripod or in studio. In those conditions, there might well be a real advantage over what the Pentax can produce at ISO 200 (or it’s “pull” setting of 100). The Phase is not a street or hand-held editorial camera. Weak high ISO isn’t that much of a big deal when working with MF in most situations.
That said, the Pentax kicks it in the ISO department. I would be interested to see how the H4, which features the same Kodak chip, and the S2 do by comparison.
Within the parameters of this test, I was not able to discern an advantage to the Phase files at ISO 50 over the Pentax at 100. My gut tells me there likely is some advantage, but nothing in this test made that apparent.
On identically exposed Pentax files, ISO 100 shows about a 1/3rd stop increase in exposure on the histogram compared to the native ISO 200. When subjected to a processing torture-test, the ISO 100 file held up just ever so slightly better, which accords with there being an angel’s breath worth of extra data. More significantly, however, the ISO 100 file exhibited noticeably less noise in the deep shadows – more than the 1/3rd stop of data would seem to account for.
The increased exposure has a concomitant impact on highlights, so extra caution is required when using ISO 100, but I think it that will be my preferred working ISO.
Because of our failure to equalize exposures, I can’t say whether the Phase has even less noise at ISO 50 than the Pentax at ISO 100, but I would bet some money that it does.
I proposed this test to Mark hoping to prove that the Pentax was at least in the same league as the Phase – an instrument I have the utmost respect for. The final results are more than I had hoped for. In this field-test at least, there is nothing to separate the cameras beyond the 645D’s superior high ISO performance.
In studio, I suspect that the Phase would be better, but that remains a supposition. Since the Pentax does not allow a professional tethering solution, there is little danger that it will find many adopters amongst high-end commercial studio shooters. In its native habitat, however, the Pentax more than holds its own.
UPDATE: November 16, 2010
Ever since we examined those images it’s been on our minds – what is really causing these – relatively minor – differences in sharpness/resolution from one corner of an image to another? We were careful: we lined-up our cameras as parallel to the wall on all dimensions as we could assess it using levels and instruments; we used mirror lock-up and time delay. We used apertures we thought optimal in terms of the depth of field needed (not really much for these shots) and the “sweet spots” of the lenses, we used high quality Gitzo tripods, low ISO, our MF cameras both have crop factors so we aren’t relying on the outer edges of the lenses themselves, etc. So what’s left on the table?.
Has it ever happened to you that after being immersed in an exercise for days, working your way through the trees to see the forest, you go to print knowing you missed a twig or two somewhere, but you’re not quite sure what? Then it suddenly dawns on you as you wake up one morning – in my case just now – oh gheez – what if the cameras aren’t perfectly manufactured, so there is an allignment issue between the optical axis of the lens and sensor. This of course would be a far more sensitive factor than anything happening in front of the lens at that aperture and camera-to-subject distance. Alternatively, using zoom lenses, there is always scope for minor alignment issues between the groups of moving elements.
And bingo – we received an email from Mark Dubovoy suggesting exactly the issue of lens to sensor alignment. If this were really the issue, what it means of course is that we are paying large amounts of money for incredibly well-manufactured sensors and lenses, only to be foxed by inadequate precision of the camera bodies. Mark Dubovoy reminded us of this reason why each Leica is hand-shimmed before it leaves the factory, and why he likes mating his Phase One back with an Alpa – individually shimmed. It all begins to make sense, and it all delivers a message to the other camera manufacturers to get their act together: in this pricing stratosphere, there’s no excuse for not going the whole nine yards on technical excellence.
Mark Segal & Nick Devlin