This is both the first extended review I’ve done in some time, and the first one of a Canon product. Regular readers will know that I only review things that are interesting, and to be interesting, they have to expand the envelope somehow. I am curious as to whether the 5DSR will do this –
I sold my 645Z a few months back because I could not justify ow ning a second system against the need for a larger home for my family; but I won’t deny that I missed its resolution and print flexibility, especially for making larger Ultraprints.
I come to this review as primarily a Nikon D810 shooter. I make no secret of the fact that I have not that much experience with Canon other than a serious evaluation period in 2007 where I decided if I should switch (pre D3); I didn’t because local support at the time left a lot to be desired. However, I also come to this review with an open mind: I’ve tried many other systems previously to expand my toolkit including Leica M, Pentax 645, Hasselblad V and M4/3. And I can guarantee you that I have no self-interest either way, since these experiments have come out of my own pocket. The 5DSR was also purchased at retail and is NOT a loaner. I don’t care what the name plate on my camera says: I care only that it lets me make the images I want to make and doesn’t get in the way, and my clients are the same. If anything, I am biased towards image quality. That said, I’m sure it’s only a matter of time before some forum keyboard warrior ‘expert’ asserts I’m now on the Canon payroll.
Note: I will present few full-size samples because I simply don’t have the bandwidth, nor do I trust that they won’t be reused without permission – it’s happened before. Given the limitations of web JPEGs, please go according to what I say and not what you see. Any perceived oversharpening is thanks to Flickr’s overzealous resizing algorithms and not blindness on my part. Read on if you have an open mind. Finally, a special thank you must go to my last client for allowing me to share some of the images from that assignment at the Crawick Multiverse and Garden of Cosmic Speculation. I literally collected the camera a few hours before heading to the airport and this was the first time I’d shot with it properly – no pressure…
Nikon’s dominance of image quality at anything less than medium format (and actually up to the 50MP medium format cameras) for the last three years has been pretty much unchallenged until recently. The niche Sigmas were the first, but they don’t form a complete system; then came the Olympus E-M5 II and its sensor-shifting capabilities, followed by Canon’s 5DS/5DSR and the Pentax K-3II which also employs similar pixel-shift technology. (I’ll be comparing the D810, 5DSR and E-M5 II in part II of this review – the Pentax was not available, and I don’t consider the Sigmas a serious replacement due to system and workflow completeness.)
Here’s why I think the 50MP Canon 5DSR is interesting from the standpoint of a D810 shooter:
– You are probably already using the best lenses possible to extract the D810’s performance, otherwise you might as well save cost and weight and use a D750. These lenses are manual focus and have mechanical aperture mechanisms.
– The Nikon F flange distance is greater than the Canon EF mount. Therefore, it is theoretically possible to use F mount lenses on an EF body. Not only that, it is possible because high quality adaptors like the Novoflex exist – and they even permit aperture control with G lenses.
– You already have the necessary ancillaries – solid tripod, adequate computer, storage, some means of displaying the output, etc.
– Your cost of acquiring resolution is some muscle memory and just the cost of the body…
I have said much in the past about bad results with adaptors on mirrorless: this is because when you have a given machining tolerance, those tolerances will affect shorter flange distances much more than longer ones. Medium format adaptors are very forgiving, for instance – but Leica M to mirrorless are not. For the purposes of this review, even if there are planarity issues we can ignore them because it is only meaningful to compare images at the focal plane since that critical focal plane itself is now very, very thin. We are not assessing lens performance. That said, it is now also even clearer to me that the vast majority of adaptors simply are not built to sufficiently high tolerances. I bought several copies from multiple brands and selected the best of the lot to use. Even so…I am thinking very, very hard about which lenses I should acquire in Canon mount in future.
Practically, not much changes in terms of workflow. For ultimate image quality, you’re working slowly off a tripod, base ISO, manual focus in live view, mirror lockup. It’s basically a sort of medium format/ view camera type approach with a bit less perspective control, lower weight and more weather sealing. And if you’re going to shoot handheld, then the extra resolution is really not going to be seen most of the time because your shot discipline is going to have to be superhuman (or you live near the sun) and you might as well use your D810. For all intents and purposes, the 5DSR is just like buying a new body: whether you shoot Canon or Nikon. Clever move, Canon; I’m sure I cannot possibly be the only one in this position.
So: this review is therefore not going to cover a few things: native lens selection (academic, since Zeiss lenses are available in identical optics for both mounts, as are Sigma Arts and Voigtlander APOs) or autofocus. For very critical applications, any autofocus system is simply not precise enough. No viewfinder focusing screen is going to let you differentiate with sufficient precision at f2, let alone f1.4. And definitely not in the corners. So, I will be shooting the 5DSR like I do the D810: with [Zeiss 55, 85 Otus] and 135 APO lenses, a Zacuto and magnified live view if there’s enough light, or on a tripod…and magnified live view if there isn’t. I suppose this is a rather unusual point of view, but practically, I cannot see any alternative way to deploy this or any other high resolution DSLR. And no, the Sony cameras are not an option because they have other serious compromises.
Unfortunately, my experience with the Canon viewfinder is little better than Nikon’s: it’s brighter, but this comes at the expense of not being able to read the LCD info bar because that itself isn’t bright enough. The second row of information on the screen itself is useful but difficult to see if the subject is dark because it’s a black LCD mask over the view. Like the D810, the mirror or finder screen require shimming because what live view tells me is in focus isn’t the position of peak sharpness in the optical finder. Except there’s no way to move the mirror zero position because it’s motor driven (good, because the mechanism appears to have even less vibration and slap than the already good D810). On top of that, my viewfinder mask appears to be crooked relative to the sensor: there’s a decided clockwise skew to my images if I align straight edges with the finder edges. Perhaps this is why it isn’t a 100% finder…
There’s still a fundamental lack of innovation in the camera market, though: more pixels, yet again. We were already long past the point of sufficiency and more pixels aren’t necessarily better; they often bring more frustration than improvement in results. Extremely tight [shot discipline] and the best lenses are required to extract all of that performance – that is not to say you won’t see an improvement on all lenses, but you just won’t be maximizing the potential of the sensor. The 5DSR is a 5D Mark III with a new sensor and mirror assembly, which is itself basically a slightly upgraded 5D Mark II – which dates back six years to 2009. I suppose if it ain’t broke, don’t fix it and all. However, we are also past the point where most people can still make images that look sharp, let alone when magnified. Though Nikon got the jump on the medium format market three years ago, it’s taken Canon this long to respond – with a disappointingly conservative offering. One wonders why they couldn’t at least have kept the mount, removed the mirror, given the thing an EVF and a focusing system/ magnification that would actually make that performance more easily accessible. There’s really no point in being able to do 200mph if you need special tires, race fuel and 5km of runway.
At the top end of the market, the cameras are basically all the same feature-wise: effectively, you get everything – interval timers, HD video (though the 5DIII or D750 are probably better video choices because of frame rate limitations and headphone monitoring jacks), weather sealing, mirror lockup etc. The difference is in how they are implemented. I have not been able to confirm anywhere whether the 5DSR has an electronic shutter, but it appears to do so in live view because I cannot hear the first curtain move – it isn’t switchable or selectable if it is, however. I expected to have to be very careful about critical shutter speeds and vibration induced from the camera’s own mechanics – thankfully, this is a non-issue. That same motor-driven mirror controls travel and provides some deceleration at the top of the mirror’s arc to minimize recoil. You can handhold it to the same or slightly longer shutter speeds than the D810 – I’d say 1/2x-1/2.5s is safe – and there’s no shutter-induced vibration even on a tripod without use of mirror lockup. That said, you simply cannot cheap out on your support system and still expect to fulfill the potential of the camera.
In the hand, the body is solid and confidence-inspiring; buttons fall naturally to hand and other than having to get used to a few different placements and dials moved around, it felt pretty comfortable. In fact, I’d argue that the grip shape of the 5DSR is actually a bit better for me than the D810, and the build is slightly more solid (perhaps due to lack of a built in flash) – the extra depth avoided pinching of the little finger. Your hands and opinion may differ, of course. However, using the cluster of buttons for AF/AE lock and point selection off on the upper right made the grip less secure than the D810’s equivalents, which are under your thumb in the natural grip position. Even the different default button positions and functions aren’t really a big deal – you can always remap them to something approximately similar, which is what I did. The sole exception is the front and rear command dial functions, which cannot be swapped – oddly, this is possible on the 7DII. Firmware update, please.
There are a few other usability niggles – you have only a very limited playback mode in instant review; you can see/delete/rate/protect/zoom the last image, but not the preceding ones. You have to enter full playback mode for that. Furthermore, battery life isn’t as good as I’d expect from a DSLR of 2015: I saw at most 600 images per charge, with no review and shooting bursts. If I use the D810 the same way, I can easily achieve three times that. Metering in matrix/evaluative mode is a crapshoot – it isn’t clear to me when it chooses to bias exposure one way or another, nor is it clear whether it’s linked to the area underneath the focusing point or not – sometimes it is, sometimes it isn’t. I’ve resorted to using centerweight or manual.
Fortunately, outweighing the disadvantages are the things which the 5DSR gets really very right: live view implementation is better than the Nikons, for starters. You have two center buttons, one of which can toggle zoom and the other re-center the magnification box. The Nikon has only one, and you have to pick one or the other. There are custom C1-3 modes on the dial, which means it’s easy to switch between say a tripod-optimized setup and a something responsive for a quick opportunistic grab. You don’t get that with the pro Nikons (the mind boggles, because it’s on the lower end ones) – that alone has often necessitated me carrying two bodies on assignment set up differently in case there happens to be another shot (and it’s paid off on several occasions).
We’ve also got a mind-boggling array of AF modes and setups designed to alter the responsiveness and tracking tenacity of the system and which points it uses depending on the subject; I honestly didn’t have to use it much because I mostly shoot static subjects, so I’ll leave exploration of that to somebody better qualified. That said, my experience with AF on the 5DSR has left the impression of being more positive and faster than the D810. It is worth noting that not all options and points are active all the time; it depends on the lens you’ve got attached, too. Continuing the list of the good is a crop mode that saves the entire file but tags the crop metadata to the file – the Olympuses also do this, but the Nikons junk the bit of the file you decided you didn’t want at the time of capture. The biggest problem here is again that the crop markings in the finder are not 100% precise, so some adjustment room is definitely welcome – saving the entire file is better. I’ve come to use that mode quite a lot – the 5DSR makes a very good travel companion paired with a Leica Q and the 40/2.8 STM, which can also be seen as a 52/63mm lens if you use the 1.3 and 1.6x modes respectively.
Note that I am not advocating cropping in the traditional sense – i.e. shoot everything and try to compose a scene in post – I am now able to use a compact lens, visualize something longer, and still have enough pixels to make a meaningful print. But beyond that, the interesting thing is the ability to see say a square composition, but shoot it with the camera vertical and then adjust it later (reminded by the tagged crop) to the right composition but without any geometric convergence – in a sense like using a shift lens, which is obviously going to deliver better image quality than keystone correction in Photoshop.
Enough of the handling stuff – suffice to say it’s more than good enough. What of the main reason anybody would buy a 5DSR: image quality?
Whilst I’ll save the direct A-B comparisons for the next part of the review, it is clear to me that under ideal circumstances, the 5DSR clearly outresolves the D810. Under less-than-ideal circumstances, the difference is there, but it might not be as large as you expect. The 5DSR has an AA filter, but it has a cancelled one like the D800E – not a completely filter-free design like the D810. Similar to how the pixel-level acuity of the D810 is an notch higher than the D800E, the 5DSR – to my eyes at least – still shows some trace evidence of AA softening, and is comparable sharpness-wise and acuity-wise to a D800E. Put quantitatively, I’d be using 250/0.2/0 sharpening instead of 160/0.2/0 for an optimal image (a ‘good’ camera with AA filter like the D750 would be 250/0.3/0, for comparison).
My biggest surprise was that I found color for outdoor and landscape work to be noticeably superior to the Nikons; grass did not go radioactive but instead required very little correction after profiling the camera using the process shown in Photoshop Workflow II. I suspect those reporting strange color from ACR have not profiled theirs, and I agree that the current defaults leave a lot to be desired. Dynamic range, however, is a different kettle of fish. Canon claims the 5DSR performs similarly to the 5DIII, which of course has much larger pixels – but is made of older technology. It is known that the current Canon offerings lag behind the pretty amazing D810, which from experience has somewhere around 14.5 stops when used optimally. I’d say the 5DSR is closer to 13-13.5; it’s certainly on the right side of ‘good enough’ but of course I’d rather have a bit more.
Despite the small pixel pitch, noise performance is better than expected. I would say subjectively it is almost on par with the D810, with a quality that is very easy to clean even just using the limited sliders in ACR. That small pixel pitch also unfortunately means that with ideal lenses, f5.6 delivers optimal sharpness with f8 being ever so slightly less, and f11 clearly exhibiting diffraction losses. This is a camera you’re clearly going to be using tilt shifts with in order to extract all performance. I cannot help but wonder if there were some electronic/design choices made that resulted in a tradoff between noise performance and dynamic range. Overall though, I think we can safely say the camera delivers to my (already very high) expectations on the image quality front. I didn’t feel at any point ‘oh, I wish I’d used the Nikon’ – in fact, vice versa in some cases.
This brings us to the question of lenses. So far, I’ve tested the camera with the EF 40/2.8 STM, EF 17-40/4 L, EF 70-300 DO, EF-70-300 L, EF 24-70/2.8 II L, EF 24-105/4 L, EF 100/2.8 IS L Macro, the Zeiss Otus 55 and 85mm lenses, the 2/135 APO, the Voigtlander 90 and 180mm APO-Lanthars and the Contax-Zeiss 2.8/35 PC Distagon and 100-300 Vario-Sonnar. Of these, the zooms were the weakest performers, in some cases, by a significant margin. The 24-70/2.8 II L was excellent, however, and the 70-300 L surprisingly very good – comparable to my Nikon AFS 80-400 VR and significantly smaller/lighter. The 17-40 and 70-300 DO were both a disaster, though I don’t know how much of that is down to sample variation – I didn’t see any decentering or astigmastism on either copy, though. The 40 STM is a pleasant surprise optically and basically functions as an AF body cap – I can see no reason not to have one. The Contax-Zeiss lenses were a pleasant surprise, with the 2.8/35 PC being really excellent if shift is kept below 6mm or so. Otuses still reign supreme, though the 55mm is starting to show a bit of weakness in the corners and at f1.4 that weren’t visible with the D810. I suspect the Sigma ART series will similarly continue to perform as expected.
In this review I haven’t said anything about the AA-filtered 5DS for the simple reason that the AA-cancelled 5DSR is the better choice to counteract the effects of diffraction. Perceptually the loss/cancellation of the AA filter is good for another stop; that may make a difference in depth of field or corner performance of your chosen lens. I see this as being the de-facto preferred choice unless you frequently photograph things liable to moiré like fabric or fine architectural patterns; moiré is removable, but no amount of sharpening is going to replace the microcontrast lost from an AA filter.
I used both the 5DSR and D810 in parallel on my last assignment, which was landscape architecture. The 5DSR produced better color and was more fluid to use in some ways (you can set the delay duration for mirror lockup, for instance) but was frustrating in others – battery life in the cold, for instance, meant that I’d go through nearly three full batteries in a day shooting live view and not more than 400 or so images. The Nikon was still on 50% at the end of the day being used the same way (in this case, with the PCEs). However, I can honestly say I enjoyed shooting very much with both; to the point that I am considering breaking my initial intention and having some overlaps between the Nikon and Canon systems; the allure of extra printable latitude is just too much to ignore. This despite my initial logic of the canon serving to fill holes in my Nikon setup such as very wide tilt-shift or a light, high quality tele. I may well add the 24-70 L to the initial buying plan of 17 TSE, 24 TSE II and 70-300 L.
I honestly think the reasons to jump system are a lot less than they used to be. The 5DSR is a very specialized tool that honestly the vast majority of people won’t need, probably can’t handle in a way that maximizes performance, and almost certainly do not own the right glass for. Is it going to stem the want factor? No, and that was never the intention. But even if you’re a Canon shooter, and as good as the 5DSR’s image quality is I’d highly encourage you to think very carefully about how you’re going to actually deploy those output pixels – it’s thorough overkill for most purposes. If you’re a Nikon user, unless you have extremely specialized requirements like me, it’s even harder to justify the purchase. The only good news is that thanks to adaptors, the cost of lensing up isn’t necessarily as bad as buying a whole new system.
Is it better than medium format? Can it really challenge the current 50MP CMOS MF cameras? I don’t think ultimate image quality is as good on noise and dynamic range fronts – but it’s closed the gap even further. I no longer have the 645Z for a direct comparison, but I have to say that the answer is the reasons for MF definitely just got even fewer, and the number of situations in which the price difference is justifiable is virtually nil for the majority of users. For pros, it also doesn’t help that the number of clients willing to pay rates that support MF equipment use are an alarmingly endangered species. In many the 5DSR represents a little shuffle further up the curve of diminishing returns – though the maximum possible performance has increased by a bit, so has the amount of effort and cost required to get there. If you are a casual shooter, it’s overkill and has been for a long time. If you are like me and chasing the frontier, you’ll welcome it (if not the cost). The DSLR isn’t going the way of the dodo anytime soon, and the theoretical maximum performance bar has just been moved another incremental notch by the 5DSR – but so has accessibility. MT
If you found this review useful, please consider supporting this site by purchasing one of my workshop videos (and upping your own game simultaneously!), making a donation using the link in the right sidebar, or by purchasing your equipment from one of my partners below. Unlike other sites, I don’t charge subscriptions and I don’t get flown to exotic locations to test new gear – in fact, review equipment has to be bought retail or import taxes paid on loaners. Your contributions help to keep the economics of being objective working. Thank you!
Canon 5DSR (B&H, Amazon)
Nikon D810 (B&H, Amazon)
Zeiss Otus lenses (B&H, Amazon)
Arca-Swiss Cube (B&H, Amazon)
Ultraprints from this series are available on request here
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!
Images and content copyright Ming Thein | mingthein.com 2012 onwards. All rights reserved