Review: The Pentax 645Z, part II: compared to the 645D, Nikon D800E and Hasselblad CFV-39

The contenders. From L-R: Pentax 645D, Pentax 645Z, Hasselblad 501CM with CFV-39 digital back, Nikon D800E.

Four cameras, 166 megapixels, no sensor smaller than 36MP and 36x24mm. It’d have been nice to get the Phase One IQ250 and Leica S along for the ride too – sadly there’s no Phase distributor in Malaysia and nobody from P1 has ever replied any email I’ve sent though. So we’ll make do with four: two from the old CCD guard and two from the new CMOS challengers. Lining up on the right are the Pentax 645D (33x44mm, 40MP) and Hasselblad CFV-39 on a 501CM body (49x37mm, 39MP) against the Pentax 645Z (33x44mm, 51MP) and Nikon D800E (24x36mm, 36MP). Perhaps we should have gotten one of the 41MP Nokia PureView phones along for kicks, too. That said, the rationale behind these choices is as follows a) I had access to them; b) to build a more or less complete system would be roughly the same price; Nikon and Pentax new lenses are more expensive than the used screwdriver Pentax FA or Hasselblad V glass; by the time you add everything in, the 645Z is obviously the most expensive option – but also arguably has the highest IQ potential. Welcome to part two of the Pentax 645Z review – the first part can be found here.

A note on testing methodology and lens choices:

• All cameras were shot raw and processed through ACR CS5.5, because this is what I normally use in my workflow.
• Color profiling has been left to Adobe’s defaults. I wouldn’t do this in the real world, of course – I apply custom profiles to all of my cameras to make results visually consistent.
• No noise reduction was applied – switched off in camera and zeroed out in ACR except for the identical default 25% chroma NR applied to all cameras.
• No sharpening was applied to any of the comparison images
• Adobe’s own lens profile corrections were used for all cameras to take the glass out of the equation as much as possible.
• White balance was achieved by using the eye dropper tool on the same location for all cameras, but the 645Z came the closest to getting it right.
• Ideally, I’d have liked to have picked the best lens for each system – the 55 or 90mms for Pentax; 45 or 85mm PCEs or the Zeiss Otus or 135mm APO for the Nikon, and the 50 Distagon or 120 Macro for the Hasselblad. Unfortuantely, this would have made any comparisons meaningless because of magnification differences; I therefore went with the only lens combination that would match fields of view on all three sensor sizes: 60mm, in 35mm EFOV terms. That’s the AFS 60/2.8 Micro-Nikkor on the D800E, the FA 75/2.8 for the Pentaxes and the Zeiss CF 2.8/80 Planar for the Hasselblad.
• A best attempt at matching DOF was made – f5.6 for the Nikon, f8 for the Pentaxes, and f11 for the Hasselblad; sufficient for the lenses to be in their optimum zones, but not so much as to make diffraction come into play.
• By extension, everything said about the D800E applies to the recently announced D810, too.
• Everything other than the ISO test series was shot at base ISO (100 for all cameras) on a heavy Gitzo 5-series tripod and Arca-Swiss Cube head.
• Focusing was accomplished through live view (where possible) or picking the best of repeated iterations (CFV). My D800E and 501CM/CFV viewfinders and focusing screens have been adjusted for perfect manual focus planarity; the Pentax 645s have not – the loaner 645D is noticeably off, and my own 645Z is almost perfect (but out by a hair, I think).
• I won’t be uploading any full size files – there’s no point. Firstly, I don’t have the bandwidth; secondly, a jpeg on flickr isn’t going to tell you much about colour, and lastly, unless you’ve got a very accurately calibrated wide gamut display AND the uncompressed 16bit TIFFs, we’ll be splitting hairs without a scalpel. I recommend going by what I say, not so much what you see in the web jpegs.
• Skin tone reproduction is something I have not explored yet. Partially because portraiture and fashion aren’t my speciality, and partially because my model/muse/wife is in hospital with dengue fever at the moment.
• A note on objectivity: I own three of these cameras (and two D800Es) and my aim is always to use the best tool to get the best result; I like to believe that I’m about as close to unbiased with the relative comparisons as you’re likely to find…
• Lastly, but importantly – a huge thank you to my buddy Darren at Autodetailer for the test location and subject material 🙂 You will notice some strange EXIF data on the 645Z images of the Ferrari 599; firstly, it is a metallic burgundy NOT rosso corsa, and secondly, it’s an available light studio with fixed fluro light banks, which means either tripod or high ISO. We were both short on time, so I opted for the latter – there are a number of images here shot at ISO 3200 or even 6400 – which is pretty darn amazing, considering just how well the camera has held the car’s color and handled the dynamic range across the scene. Think of it as a worst-case torture test of sorts.

What’s not obvious in this table is the difference between the 645Z and 645D is significantly larger in practice than the numbers would suggest. Firstly, the new camera is just so much faster in operation; it’s the difference between slightly frustrating and fluid. You needed to wait at least 3 seconds per shot for the buffer to clear on the 645D before being able to view an image; even menus were a bit laggy. Review is under a second on the 645Z, and reviewing/ zooming/ menus etc. feel much the same as the D800E: effectively instant. On top of that, autofocus is much faster and more precise; it locks on with more confidence and doesn’t seem to shift around as much during repeated refocusing. Again, speed- and accuracy-wise, it doesn’t feel much different to the D800E. Battery life is also noticeably better, and despite the increase in pixel count, the camera seems to be no more susceptible to shutter vibration than the old one; I presume we have a new mirror and shutter mechanism to thank for this. There are other operational improvements too, including addition of video recording, nearly 3x the continuous frame rate, and of course a vastly larger shooting envelop thanks to sensitivities that top out a whopping seven stops higher. And that’s before we even begin to talk about image quality.

Unsurprisingly, the newer generation of cameras handily trumps the old one – on the price/ performance ratio, at least. But this has been the way for as long as I can remember in this industry, and more so when we’re dealing with very niche products. I don’t know how many CFV backs Hasselblad sold in the three or so years before discontinuing them, but you can be sure it’s probably fewer than Nikon makes D800Es in a week. There are plusses and minuses of course to going with newer options. Firstly, there’s the whole debate over colour, bit depth and tonal response; I think at least part of that is justified. And the operational experience is very, very different. For the most part, conventional DSLR and DSLR-like mirrorless designs are very well integrated; there are still some UI aspects that could be improved – there always are. Medium format digital has mostly felt clunky to me: the camera part operates separately from the digital back part, and until fairly recently, they didn’t even talk to each other. There was a point when the camera and back powering on simultaneously from the same button was listed as a feature! Finally, we have the Pentax 645 twins: their operation is more seamless than any other medium format camera I’ve used so far (though granted I have not had a chance to try the new Phase One cameras) and its handling doesn’t need much introduction because it’s much like an overgrown DSLR. That can be a good and a bad thing – good because you’ve got a lot more room for external controls, bad because it means even more weight.

Even with this in mind, using the CFV is a unique experience. You have a 100% mechanical camera – any V series body –  and lens on the front that’s of course capable of operating without batteries and independently of the back; you could even (gasp) load film into another back and swap it out if you wish. The back operates independently from the rest of the camera. The only way they communicate is via the little pin that protrudes from the camera body and used to advance the film counter; before the exposure, when the auxiliary shutter curtain rises, the pin extends into the back which wakes it up and primes the sensor; the leaf shutter is closed at this point so the whole camera is light-tight. Then the shutter in the lens fires, determining the exposure, and the pin moves back into the body, ending the exposure. So in effect, the exposure time of the digital back is always much greater than the actual exposure time determined by the lens. Presumably, there’s some clever software or noise reduction circuitry in there to prevent long exposure shadow noise – though this is already slightly higher than for most cameras, it’s not easy to determine if it’s the result of each actual sensor-on time being at least 0.5-1s or something else.

Regardless of all of this, if you enjoy the experience of shooting with an older, fully mechanical/ manual camera like a film V-series, this is as close as you’re going to get in the digital world. Remarkably close, in fact, other than the crop factor (1.1x, 645 – not square, sadly, unless you don’t mind a 37x37mm sensor area). I still think the V series design is one of the most fluid in use – all of the critical controls (focus, aperture, shutter) are all on the lens itself, easily preset, and viewable at a glance. And you even have program mode because the aperture and shutter rings are coupled on most of the lenses. However, this system as a digital option has one enormous catch: because the film capture area was meant to be square, the viewfinder and controls were laid out in such a way as to only work in one orientation. However, the digital sensor is a 645 aspect ratio: in order to shoot verticals, you need one of the 90 degree prisms and a grip, or a tripod. Leaf made a series of backs with an internally rotating sensor, which was a clever solution; Phase One P backs could mount in either direction (but run the risk of dust and moisture ingress) – either way, you’d still need a special viewfinder mask or focusing screen to show both vertical and horizontal orientations. Personally, I found the prism + grip solution much better for stability and general usability.

When seriously considering one of these imaging monsters, it’s also important to look at the types of subjects you shoot, and the lens choices and general system accessories available. Whilst the V series is perfect for studio work because the leaf shutter will give you flash sync all the way to their 1/500s maximum speed, and relatively small maximum apertures (f4 for all of the leaf lenses except the 80mm, which is f2.8) and manual focus only aren’t very important, they’re not so good for outdoor reportage and available light work – precisely because you’re restricted to f4 and 1/500s. In the tropics, or in any kind of bright sunshine, that means using ND filters (focusing becomes tricky), having almost no depth of field control and running into diffraction issues, or overexposure. By the same token, the Pentax 645 system is not the best choice for studio shooters: flash sync is just 1/125s because of the focal plane shutter and a lack of leaf shutter lenses; however, you do get a 1/4000s maximum overall speed, and most of the lenses are f2.8. The Nikon D800E splits the difference – 1/8000s maximum shutter, 1/250s sync speed, and lenses down to f1.2 (mostly f1.4). It’s easy to see why the D800E has been making serious conquests over both documentary and studio professionals – I deploy mine under both situations to equal effect. However, it still renders like 35mm. The difference is seen mostly at normal angles of view and below: it’s difficult to describe exactly, but wides look wide. There’s a definite difference in the way the Zeiss 2.8/21 renders on the D800E as opposed to the Pentax 25/4 on the 645Z – even though the Pentax is actually wider at an equivalent 19mm horizontal field of view, it doesn’t feel as wide as the Zeiss. Something to do with geometric projection, perhaps.

But I digress: all of these three options are mature systems. So mature, in fact, that legacy lenses for the Hasselblad V – proper made in Germany Zeiss glass, no less – go for an average of US$500-1000 a piece; non-SDM FA autofocus lenses for the Pentax are even cheaper, at somewhere between US$300-500. And there’s absolutely no shortage of Nikon glass whatsoever, though not all of it will be able to perform to the expected level on the D800E’s sensor. This is an important point to note: all of these sensors are relatively high density and will test lenses to their limits, especially the D800E; a lack of good lenses is its achilles’ heel. The others suffer to the same thing to more or less the same degree, so it’s very important to know that the focal lengths you need are adequately served. I would say that there are actually very few lenses indeed that can be used on digital without reservations to distance or aperture on all three of the systems (there may be others which I haven’t tried):

• Hasselblad V: 4/40 IF, 4/50 FLE, 4/120 Makro-Planar, 4/150, 4/180, and the Superapochromats
• Pentax 645: D-FA 25/4 SDM, D-FA 55/2.8 SDM, D-FA 90/2.8 SR SDM Macro – and perhaps the FA 200/4 if focused manually with live view.
• Nikon D800E: Nikon 24 PCE (beware sample variation), 45 PCE, 85 PCE, 85/1.8 G, 70-200/4 VR and the superteles; Zeiss 2.8/21, 2/50 Makro-Planar, 1.4/55 Otus, 2/135 APO

You’ll notice this is not a very long list at all, and only Nikon has anything tilt shifts. Hasselblad has the clunky Flexbody and somewhat less clunky Arcbody, but operation of both with digital backs is fraught with risk as you must remove the digital back to focus and frame between every exposure. On top of that, most of the lenses will have edge CA and purple fringing issues when tilted or shifted. Realistically, this means that architecture still remains the preserve of the Nikon system for me. Product photography too either requires stitching or tilt shifts; it’s easier to get it right in camera, so again – we’re back to Nikon for most things. For portraiture, fashion, reportage/ documentary, travel, landscape etc – pick your poison. That actually means for these things most people use their cameras for, you’ve got a lot of options; this is not surprising since all of these are mature systems which have had the benefit of many decades of professional photographers to refine.

A note on Hasselblad-Zeiss V glass on the Pentax 645: it can be done, because the V has a longer flange distance than the P645. I conducted some informal tests but didn’t have a chance to present them properly. In short, the Zeiss V glass outperforms the legacy FA Pentax lenses (75/2.8 vs 2.8/80; 200/4 vs 4/150) by a noticeable margin, wide open or stopped down. The Zeisses tend to start off a bit softer but improve more through f8-11; the Pentaxes start off a bit better and have better near performance, but don’t improve as much when stopped down. The difference between the 200/4 and 4/150 was so great that I’d probably pick the 4/150 over the 200/4; despite 33% more focal length, extra resolved detail is negligible. You’re going to be using both of these on a tripod anyway, so stability is a non-issue for the most part. Microcontrast and color rendition are better, and the output simply appears to have more ‘punch’ without being harsh. However, the new Pentax SDM lenses are all on par or better – the 55/2.8 is slightly sharper than the 4/50 FLE, which suffers from some focus shift as a consequence of the FLE group; the 90/2.8 SR blows the 2.8/80 out of the water, and is slightly better than the 4/120 Makro-Planar all over. There is no Zeiss equivalent to the 25/4. Food for thought. Note also that because of the much longer flange back distance compared to say 35mm and mirrorless, production tolerance issues are much less noticeable. 

At this point it’s probably worth mentioning the competition – Hasselblad’s own H5D-50C, and the Phase One IQ250. Both of these have more extensive lens systems with some tilt shift options (or the 1.5x HTS tilt-shift converter for the Hasselblad) and autofocus, but a significantly increased price – north of US$27,000 for the H5D-50C, and $32,000 for the IQ250. In today’s strained commercial environment, such expenditure is extremely difficult to justify by most pros – including myself. In comparison, if you need another Hasselblad V body, you can probably find a nice one for about $800 – including a film back, waist level finder, and possibly an 80mm lens, too – they tend to go as kits. The body only might be $400-500. I can make medium format digital pay off for me, at that level the ROI is much lower than I’d like – and that’s before we even count the cost of adding lenses. There are two enormous reasons why they can ask for that much money and still find buyers, though: autofocus + leaf shutter/ high speed flash sync, and professional support – at least in the rest of the world. Phase One is pretty much nonexistent in Malaysia, which makes it a non-starter for me; I don’t need high speed flash, and even if I did, I prefer to use speedlights, so running a separate NIkon system is both cheaper and more flexible. However: your requirements and local options will almost certainly differ, and this should be taken into consideration, too.

Let’s move on to the more quantitative part of this comparison. Again, I emphasize that you should go by what I say, not the small web jpegs. Since the whole point of the exercise is to chase the last fraction of a percent of image quality, viewing these images at anything but maximum quality – uncompressed full-fat TIFFs on a calibrated monitor – is rather meaningless, and I simply don’t have the bandwidth to serve them. I’ll break down the assessment into a few categories. Even so, at base ISO, the results are much closer than you might think. Click through to the images on flickr for larger versions.

For reference – applies to all of the images which are divided into four panes.

First test scene, processed to final.

Second test scene, processed to final.

Resolving power
Our ‘lowest’ resolution camera here weighs in at 36MP; the highest, 51. Though this is an area difference of some 40%, the linear difference is much less – more like 20%. It’s basically going from a 10×15″ to a 12″x18″, or more like 12×16″ due to the difference in aspect ratios. Pixel pitch also plays a secondary role here: the smaller the pixels, the less acuity there is because all other things being equal, the signal to noise ratio is lower and this affects acuity. (Of course, there is some oversimplification here as lenses also play an important part in the equation.)  What’s telling though is that only the D800E and 645Z don’t appear to have any significant resolution compromises with increasing ISOs; edges don’t start degrading until very late in the day. Bottom line though: to see a consistent difference, you’re really going to have to be employing optimum shot discipline all of the time; otherwise that 20% is going to disappear very fast indeed.

From the 645Z. I think there’s no question that there’s a frightening amount of resolution here – note the reflections in the bolts, and the textural differentiation in the brake calliper. Remember, this image has not been sharpened. More impressively, most of this detail holds through even ISO 6400.

Dynamic range
The first time I used the D800/E, I was surprised by firstly how flat the raw files looked, then by just how much latitude they held – in both shadows and highlights. There’s probably an easy 1.5-2 stops at both ends; practically, the D800E seems to have around 13.5 usable* stops at base ISO. Put it this way: if exposed properly, it rarely clips. Being used to the D800E, I found the 645Z even more impressive: it almost didn’t clip at all, despite being shot under extremely high contrast situations during the test period. I’d say there’s at least a stop more dynamic range in there, perhaps slightly more than that. However, it’s worth noting that at least some of this (for all cameras) is attributable to the raw conversion since we only have 14 bit files to begin with. This is not to say that the 645D and CFV-39 are slouches in this department: far from it. I think both top out at around 12-12.5 stops; they certainly clip with more regularity than the D800E, let alone the 645Z. This is more of an issue for the 645D, because it doesn’t appear to handle overexposure with as much grace and smoothness as the CFV; pixel pitch probably has something to do with it. The CCD cameras lose dynamic range very quickly as sensitivity rises; basically, one stop for every one stop gain in ISO. This isn’t quite the case with the CMOS cameras, which only start to visibly lose dynamic range over about ISO 800 (D800E) or 1600 (645Z). The difference is further compounded by the fact that the two CMOS cameras had more dynamic range to begin with.

*The definition of ‘usable’ is also highly dependent on how tolerant you are of shadow noise.

Highlight dynamic range, ISO 100. Ignore the slightly different focal planes/ depths of field.

Highlight dynamic range, ISO 800. You can see all of the cameras starting to block up slightly, the D800E appears to be handling this slightly better than the other cameras. Still impressive just how much of the highlights they’re holding, and how smooth the rolloff is – despite being a very hard reflection.

Midtone dynamic range, ISO 100 – none of the cameras is having any issues here.

Midtone dynamic range, ISO 800 – you can see the two CCD cameras starting to struggle with both noise and shadows crushing; the tonal response is starting to become a bit weaker, too.

Shadow dynamic range, ISO 100 – all of the cameras have no problems holding the entire dynamic range of the scene without clipping; however from the density of blacks it’s clear that the 645D is the weakest of the four, followed by the CFV (though it has quite a high noise floor compared to the other cameras); the 645Z leads the D800E by about a stop.

Shadow dynamic range, ISO 800 – If we take a look into the shadows, you can see the CMOS cameras pull far ahead – the 645Z looks to be a bit lighter than the D800E, despite an identical highlight exposure, which suggests that most of its extra dynamic range lies in the deep shadows. The CCD cameras, on the other hand, are starting to lose detail and dynamic range; in the case of the CFV, we’re losing pretty much everything below Zone IV.

Noise
The 645Z and D800e walk away with this one; whilst all of the cameras look pretty much the same at ISO 100, you’ll find that the noise floor is much lower on those with CMOS sensors. Beyond that, the CFV starts to fall apart at IS0 400, and earlier in the deep shadows – notice how quickly luminance noise starts to set in. We get another stop out of the 645D to about ISO 800; even 1600 is actually usable at a pinch, though expect to have lost many stops of dynamic range in the shadows. The D800E still looks much the same as the 645Z, and both don’t appear much different from base ISO. That said, though the highlights and overall histogram appear the same, the 645Z’s lowest zone is a bit lighter, which suggests that there’s more dynamic range than the D800E in here. In fact, the D800E remains relatively clean at ISO 3200, and ISO 6400/12800 can be used in a pinch with some postproduction NR work. Under this kind of (fluro) lighting, the 645Z appears to perform much the same.

Subjectively, this is somewhat different to my observations under more neutral white balance; the 645Z has a bit of an edge over the D800E in those situations – by as much as a stop.  I’d say the 645Z’s higher 25k and 51k sensitivities can be used if you don’t mind downsizing. In fact, a 645Z file downsized to match my D4 blows it out of the water in noise, acuity and dynamic range – by more than two stops. Remember, for a given output size, the increased amount of information to begin with in the 645Z’s files are going to mean that even if noise at the pixel level is the same as say the D800E, you’ll still get a noticeably cleaner result. And that’s not quite the case, because this camera is about a stop or so cleaner at the pixel level, too. It’s also worth noting that the CFV-39 has no microlens array over its sensor; this means that its light gathering ability isn’t quite as good as the other cameras. Its actual ISO seems to be about 2/3 stop less sensitive than the others despite having the same numerical rating – important when taking into account flash and other exposure calculations. Also, the CCD cameras will punish underexposure and post-processing recovery quite badly with noise and colour shifts in the shadows; this penalty doesn’t seem to be anywhere near as harsh with the CMOS cameras.

ISO 100. What follows is a series of crops at every ISO; the CFV tops out at 800, which we’ve left in for the last full comparison at ISO 1600. Thereafter it’s down to the two CMOS cameras. I stopped at 3200 – this is probably about the limit for what I’d consider very good quality after some noise reduction/ cleanup; 25k was thrown in for kicks and entertainment.

ISO 200

ISO 400

ISO 800

ISO 1600 (CFV at 800)

ISO 3200 – 645Z on the left, D800E on the right. Notice the 20% linear difference in image size isn’t quite as much as you’d expect.

ISO 25600. Apply PP NR, downsize and you’ll actually have a surprisingly usable file.

Color and tonal response
Firstly, it’s important to know that all of the images from each set had their white balance set on the same object in the scene, using a 5×5 pixel sample eyedropper; this should be about as close to consistent as possible. I used the default Adobe conversion without custom profiling. I’m sure somebody will say at this point that there is better software for each camera; I agree, but from a consistency, comparability and workflow point of view, ACR is still very much the way to go – especially if you happen to be mixing multiple cameras on a job, which I will probably land up doing. There is only one 16-bit camera here; does it produce more accurate colour than the rest? No. More pleasing? Perhaps, but that’s subjective. The only place where the CFV is demonstrably better is in reproduction of the reds in the tail light; it’s much closer to accurate than the 645D or D800E, which are far off; interestingly, the 645Z is closer to the CFV despite having architecture and processing that’s much closer to the other two. Skin tones, blues, greens etc – yes, they’re missing from this comparison, but there’s only so much I can physically do in two days. In any case, it’s not a big deal to profile the colour of any one of these cameras (or all) to be the same as the CFV, or consistent with a fifth reference. I know for a fact that my custom profile for all of these cameras will be more like the latter. I just want to make one comment about all of the cameras: there’s remarkably little hue shift as the sensitivity increases even to 6400 or thereabouts (with the exception of the deep shadows of the CFV, which have been polluted with chroma noise); we see some desaturation but this is to be expected.

ISO 100. Strong tonal differentiation of fine gradations even into the shadows from all cameras, as expected. Color reproduction is actually remarkably similar here.

ISO 800. Not quite the same story; the CFV has acquired a green hue shift, the 645D has increased in contrast as its dynamic range contracts, and the two CMOS cameras look almost the same as at base ISO.

ISO 100. This is a bit more interesting: one of the trickiest hues to reproduce accurately are deeply saturated reds; Ferraris, transparent clear plastic tail lights, maple leaves and the scarf of my polar bear all fall into this category. Early digital cameras would blow the channels very quickly, to the point of being unusable unless you underexposed by 1-2 stops – at the expense of the rest of the image; we’ve come a long way since then. Here, the same point on the car’s bodywork was used as a white balance reference. The 645D and D800E are both too orange, with the 645D being the least accurate. The Hasselblad and 645Z are closer to the mark, but neither quite nails it – the Hasselblad is too pink, and the 645Z still a bit too orange. I think the truth lies somewhere between those two.

ISO 800 – and remarkably little hue shift or noise in this zone, even for the CFV.

The final image is to show a few things: at the same output size, how much depth of field separation you get from a given angle of view and real aperture – in this case, 60mm (35mm EFOV) and f2.8. The CFV’s format is double the size of the D800E, and as a consequence there’s a noticeable amount more defocus – as expected. There’s also slightly less depth of field (60mm plays 80mm). The second point I’d like to mention is colour again: the CFV has perfectly nailed the colour of the Ferrari, but the 645Z is a bit too dark and the D800E too orange. Still, the differences are relatively slight and easily correctable afterwards.

Shooting envelope
I think it’s no surprise here that the two newer cameras have the largest shooting envelope of all; I’d put the D800E and 645Z on par because what the latter gains in high ISO performance and resolution it gives up to faster lenses and a more flexible system to the former. It’s quite possible that the slight high ISO gain for the D810 may edge this somewhat in Nikon’s favour. None of these can exactly be considered light or compact, though the 645s with the diminutive 75mm aren’t too bad, and if you opt for the f1.8G primes for the Nikon (28, 35, 50, 85) you can pack a lot of image quality into not much space at all. I think the 645D lands up somewhere in no-man’s land; it’s really the low-cost starter option to see whether medium format is for you or not. And finally, the CFV combination is really a scalpel: it excels at some things, but definitely isn’t for every situation.

Of course, there has to be a huge amount of personal subjectivity to be taken into account. It may well be the operation of the CFV is so much more pleasurable to you than a digital-DSLR style interface, in which case the rest are non-starters; or price, which would favour the Nikon; or perhaps you’re allergic to manual focus or have vision issues, in which case the Hasselblad would be a non-starter – and you’d have to think very seriously about focusing the D800E, because it seems to be the most intolerant to focus errors of the lot  – partially because of the pixel pitch, and partially because the focusing screen is nowhere near as differentiating as the others between in and out of focus areas. Haptically, none of these cameras are bad choices: they’re all mature, evolved designs. Even the blocky-looking Hasselblad V bodies are actually very comfortable to use; I actually find them very, very fluid in use and beautifully simple. I put my left hand on the lens controls, and my right hand cradles the bottom the camera body with my index finger on the shutter release; I had to be a bit careful when doing the comparison because the lens release is in the same place on the Pentax! On top of that, the camera itself (and the back, styled to match the chrome body) is a work of art – unfortunately, I can’t say the same about the 645Z or D800E; at best they are ‘functional’; quite frankly, the 645Z’s external design is something only its engineers could love. Still, it gets the job done. In any case, I highly encourage you to try them out in person before making a decision; ultimately, it will come down to personal preferences. Note: the film Hasselblad V bodies handle exactly the same as one with a digital back; the digital back is a little bit heavier and larger than a film magazine, but not by much.

It’s difficult to come to a clearly meaningful ‘which is better’ conclusion here, other than it’s clear that the 645Z succeeds and improves on the 645D by some margin, and unless you’re on an extremely tight budget (in which case you’re probably going to be better off with the D800E or D810 anyway) is really a non-starter option unless you must have that medium format look. If you only have the budget or weight/ space allocation for one system, I have to say go with Nikon – it’s the most flexible and will let you do probably 99% of what the others can. It’s also the only option if you have special purpose requirements such as tilt shifts or supertelephotos. Splitting the 645Z and the CFV is much harder; though there’s a definite resolution advantage to the 645Z over all of the cameras here – between 27% and 40% in area – I think the CFV trumps it very slightly in pixel acuity, though this may well be down to the lenses. I personally prefer the slightly biased warm colour native to the CFV, but this can be easily be adjusted for in post. I think there are two deciding factors here: firstly, whether you’re going to use it in studio or with available light, and secondly, the rendering. The former has to do with flash sync and high iso capabilities; effectively you’re trading off one for the other. The latter is something a bit more subtle: the larger the sensor, the more natural the rendering appears to be – at least to my eyes. I’m not talking about the super-shallow depth of field that’s a property of very large formats, but the way the focus transition rolls off, even at f8; it tends to be much smoother and more gradual than the abrupt ‘planes’ we’re accustomed to with fast lenses and 35mm DSLRs. I guess what I’m saying is that I can see a reason to keep both the CFV and 645Z around – unfortunately, my accountant is telling me otherwise…MT

The Nikon D800E is available here (B&H, Amazon). The D810 is available here (B&H, Amazon). The Pentax 645Z is available here (B&H). The Pentax 645D is available here (B&H, Amazon) – though I suggest hitting up Japan Camera Hunter to find you a nice used one from Tokyo instead at significantly less cost. The Hasselblad CFV-39 is sadly discontinued, but the CFV-50 (50MP) is still available here from B&H.

__________________

H2 2014 workshops now open for booking – Making Outstanding Images San Francisco, Chicago and Venice; Masterclass San Francisco and Venice – click here to book or for more info

____________

Visit the Teaching Store to up your photographic game – including workshop and Photoshop Workflow videos and the customized Email School of Photography; or go mobile with the Photography Compendium for iPad. You can also get your gear from B&H and Amazon. Prices are the same as normal, however a small portion of your purchase value is referred back to me. Thanks!

Don’t forget to like us on Facebook and join the reader Flickr group!

Images and content copyright Ming Thein | mingthein.com 2012 onwards. All rights reserved