Hasselblad PR: “Real World” Comparison: Nikon D850 with Decrepit Nikon 17-35mm f/2.8 Zoom Compared Without Format Equivalence or Validation of Basics
I intend to do some medium format review work fairly soon, just as soon as I can get the gear.
A sort of public note to Hasselblad: I’d love to test the Hasselblad H6D, but even the new XCD 21mm is proving a bother, as Hasselblad has changed management, and now wants site metrics. I don’t play the bean-counter game. Heck, I don’t even track metrics any more as they never delivered any value (how many fanboys?!). I want a relationship that values the work I do, and its quality. I have serious concerns that changes in Hasselblad management (early 2017) that are a turn-off not just me, but also for the best talent they have in the company (regular readers might deduce a particular name I have in mind). And most important: would I recommend investing in Hasselblad gear in the face of such attitudinal and ownership changes?
Today, Hasselblad posted this ostensibly fair medium format comparison, using UK photographer Karl Taylor to lend credibility. I find it vraiment incroyable, for those across the chunnel.
The comparison does not have any credibility for me. Which is not to say the final conclusion is wrong, only that repeated errors in methodology show that Mr. Taylor is unqualified to make comparisons, or perhaps deliberately partisan. I don’t read minds.
Let me call out just a few of the more problematic errors in methodology, which to me at least, are puzzling for a professional photographer to make.
#0 Mismatched (for format) shooting parameters
A photographer using mixed format sizes ought to understand equivalent depth of field—if not, that photographer should not be doing camera comparisons. In this case, both cameras were shot at f/16 from what I see in the video showing ACR conversion—a fundamental error.
The use of f/16 for the 35mm DSLR is guaranteed to drop overall contrast and micro contrast to marginal and very dull levels—worse then wide open for any good lens (see for example the full MTF series for the Zeiss Otus 55mm f/1.4 where f/1.4 is nearly as good as f/11!). It is why I almost never use f/16; it degrades the image massively.
One of two things should have been done: (1) the Hasselblad H6D image should have been shot at f/24*. (2) Alternately, the Nikon should have been shot at f/10.6*.
The Hasselblad H6D-100 uses a 54 X 40mm sensor, versus a 36 X 24mm sensor. Taking the long dimension (diagonal is never appropriate IMO), that means a ratio of 54/36 = 1.5. So the H6D should have been shot at f16 * 1.5 = f/24, or the 35mm DSLR shot at f16/1.5 = f/10.6.
There is no other fair way to do this. Thus the video is based on shooting parameters guaranteed to make the 35mm DSLR look bad by greatly increasing the amount of dulling, contrast-robbing diffraction.
- Format-Equivalent Depth of Field and F-Stop
- Hasselblad X1D Format Equivalent Focal Length, Depth of Field
#2 Poor-performing zoom lens vs modern-design prime lens
“This is a top of the line 35-mil camera...”—but no mention of the lens. And no mention of the camera resolution but from the video it appears to be the Nikon D850.
Taylor uses the Nikon 17-35mm f/2.8, which was marginal on a 24-megapixel camera 8 years ago, and compares it to an about $6250 designed-for-digital medium format lens. I wouldn’t even consider shooting the Nikon 17-35/2.8 for my work, yet this is the lens by which we are to compare 35mm to medium format? Where is the Otus for the comparison, which would still be much less expensive than the Hasselblad 28mm? It is incroyable to see this choice—offhand, I cannot think of any lens less appropriate. The most appropriate lens would have been the Zeiss Milvus 18mm f/2.8, or the excellent Sigma 14-24mm f/2.8.
Let’s also for giggles ignore field curvature which can hugely skew any comparison and is very strong in the 17-35. Ditto for failure to verify optimal focus. Both are topics not mentioned in Taylor’s video all while pixel peeping the results. Well, and there is focus shift too.
Even ignoring diffraction (which is pronounced at f/16), some lenses also suffer from reduced contrast from internal reflections and this can be a serious issue at f/16 in addition to diffraction. Indeed, Zeiss created a special test lab setup when I found such an issue with one of their lenses about five years ago—and that was from f/4 to f/5.6. Contrast losses from internal reflections with stopping down are another reason the comparison is ludicrous by the choice of the Nikon 17-35mm f/2.8, hardly the state of the art in its optical design or lens coatings.
Thus Taylor pits a decrepit zoom lens design with many elements against a modern medium format prime (Hasselblad 28mm).
Thus follow statements in reference to the zoom of “looks rather 'tinny'” followed with gems like “greater aberration and corner softening... lower optical quality” and the real nugget of wisdom “Hasselblad optical glass..”, as if that were some special characteristic unique to Hasselblad.
#3 Best vs best — no mention of quality criteria
There is no mention of whether the Nikon D850 was shot in lossless-compressed 14-bit raw. That seems fundamental if the D850 is to be compared to Hasselblad 3FR 16-bit raw files. Still, I’ll give Mr. Taylor the benefit of the doubt.
I cannot find any mention of whether the cameras were shot for best quality, or for nominally same ISO. This is important, because when comparing formats, the same ISO is rarely a fair comparison, though I admit to this being a debatable point on some simplistic “same number” basis, which has no relevance to my shooting.
What is not debatable is that shooting a camera at an ISO for which less-good results are guaranteed is a poor test criteria, particularly since nominal ISO is not at all appropriate given the way Hasselblad (in particular) maps sensor dynamic range to ISO quite differently than most vendors, as I have shown in my review coverage. Moreover, to collect the same number of photons, the ISO should be lower on the smaller-format camera with a correspondingly longer exposure time. It is my position that “fair” is “best possible” or at least “same number of photons collected”. This was not done.
- Increase Image Quality by Optimal Exposure (ETTR, Dolls)
- Shootout vs Nikon D810: 4-stop Underexposure + Push (Flowers)
- Shootout vs Nikon D810 (Flowers)
When I shoot my Nikon D850, I shoot for best quality, which entails appropriate choice of ISO, e.g., ISO 64 or ISO 32. When formats change, so does pixel size, so I deem ISO 64 fair in comparison to ISO 100 on, say, the Hasselblad X1D. Given the right conditions and particularly on a tripod, does anyone out there really want to shoot for lower quality than the camera can deliver? The right comparison is “best possible” unless the goal is purely same-ISO shooting as in handheld shooting, but even there failure to account for motion blur is an error—and that relates to pixel pitch and ISO and camera mass and shutter and mirror vibration too.
#4 Failure to assess actual, recorded exposure
It is a fundamental error in methodology to not validate exposure of the raw file in RawDigger, that is, to verify that an equivalent exposure was given to each camera relative to its capabilities. See RawDigger Histograms compared in Shootout vs Nikon D810: 4-stop Underexposure + Push (Flowers). Even the lens comes into play in this regard.
Ignoring the possibility of variation in illumination in Taylor’s sunset comparison (a real risk in such conditions, as years of field work have shown me), I have seen significant ISO variations between cameras. What exposure was actually recorded relative to the sensor’s dynamic range?
Moreover, a Nikon D850 (or D810) should be shot at ISO 64, not ISO 100. However, Taylor chose ISO 100 for both, thus giving the 35mm DSLR (is it a D850?) a 2/3-stop disadvantage in terms of dynamic range and noise. It is simplistic to consider the same ISO as “fair”, for it is not, both for the number-of-photons-collected reasons as well as others. For example, the 3-second exposure in Taylor’s comparison saves the H6D from shutter-shake concerns which in reality are a major problem, and it might involve noise reduction for the D850.
Did Taylor enable LENR (long exposure noise reduction) for the 35mm DSLR? It matters to detail.
Note also that in Taylor’s video, white balance and tint and saturation are all mismatched in his ACR screen shots and that the H6D uses a 1/3 stop pull. By what objective standard did Mr Taylor establish a fair set of processing parameters for each? It is plain to see that the color is wildly different, yet late in the video he deems it just fine to comment on the warmer look of the H6D—that’s rich.
Different cameras of the same format size often have very different requirements for sharpening (e.g. Nikon vs Sony), not to mention different cameras of different format sizes. I can’t find any mention of this in Taylor’s video. It comes down in essence to “best” versus a simplistic idea of “same sharpening”—that never works.
“We can see a much greater level of detail in that rock...”. So—we have a 35mm camera degraded substantially by diffraction at f/16, shot with a poor-performing lens with no validation of focusing or field curvature or depth of field considerations as compared to a 100-megapixel camera using a modern prime lens that should have used the equivalent aperture of f/24. It does not seem to occur to Mr. Taylor how ridiculous that is.
“I’m going to look at corner sharpness...”.... Taylor finds that the Nikon 17-35mm f/2.8 is lower optical quality, and I agree wholeheartedly. He concludes that moment by quietly murmuring “... compared to the Hasselblad optical quality glass...”), as if the Nikon 17-35 uses non-optical quality glass? Mr. Taylor does not discuss why he chose a low-grade lens in the first place. What can possibly be the point of such a comparison? I cannot read his mind, so I don’t have that answer, but for me it invalidates his entire test.
#6 Lens corrections
Having shot a grossly inferior lens no better than today’s $100 kit lenses and probably worse (the Nikon 17-35), Mr Taylor proceeds to comment on the chromatic aberrations of the Nikon 17-35.
Mr Taylor makes no mention of lens corrections for either lens. In particular, he does not comment on whether the H6D has already baked lens corrections into the Hasselblad 3FR raw file (or whether Adobe Camera Raw does so), emphasis added from the Hasselblad spec sheet:
Image quality is refined with integral use of Digital Lens Correction which perfects the raw image by digitally removing any color aberration, vignetting and distortion.
It is interesting that towards the end (around the 08:30 mark), Mr Taylor conflates detail with dynamic range—I can’t follow that argument, which is a non-sequitur for the image content in question, and unsubstantiated by any objective evaluation of dynamic range, e.g., RawDigger.
Also towards the end, Mr Taylor also emphasizes an “equal platform” (ACR), failing to grasp the issues of lens corrections, appropriate sharpening for the particular camera, the fact that he has strongly biased the comparison by using non-equivalent apertures, etc. His last comments on a “slightly warmer natural tone in the Hasselblad” is ludicrous (!) at best, with his screen showing a huge difference in white balance—as if pseudo-random “As Shot” white balance has any relevance to raw image conversion.
Other curious issues
What about Live View?
Hasselblad Live View is poor on the H6D as I understand it (if it even has a usable mode), but this is not discussed by Mr. Taylor.
“Just looking through the medium format camera is a pleasure because I’ve got a larger brighter viewfinder...”. True enough I presume, but working on a tripod as shown why would anyone sensible use the optical viewfinder for that kind of shooting? Using the OVF also implies failure to establish optimal focus, that is, using conventional autofocus instead of the far more accurate magnified Live View.
Is Mr. Taylor unaware of Live View and a Zacuto loupe? That is all I use these days for my tripod-based work. Large and bright and supremely accurate for focusing. When comments like that are made out of context and with context dropped, I take issue with it.
Having rigorously evaluated a number of medium format cameras, I do believe that medium format has advantages in many cases, though I also have proven that a skilled photographer can minimize the gap or even beat medium format in some ways, particularly if top-grade lenses like Zeiss Otus are used.
However, Mr. Taylor’s ostensibly fair comparison is swiss-cheesed with fundamental errors in methodology (equivalent aperture and ISO, validation of exposure and focus) along with a absurdly marginal lens choice (the aged Nikon 17-35/2.8). It cannot be taken seriously as a comparison exercise.
I worked professionally with the 500V system, along with Canon and Nikon SLRs and the digital counterparts. So, I stopped watching after the ill-fated question of the video… Lets see, how? my trusted Nikon will behave next to a $50,000 set up of a vastly different format.
You just unleashed the Internet Beast, here...
Congratulations though, for clearing up this mess of a post on Hasselblad’s page. And “defending” the integrity of the endless hours of dedicated work you devote to your site. Not that it is needed, defending!,- I was going to say, - yet posts like these, demand a shoulder check from all of us about the junk news that has infested most every subject ever highlighted on the NET.
I am not even sure why Hasselblad thought that it was STILL necessary to feed the MF debate.( Click Bait? ) If there was ever one worth noting. It has sailed to a Hawaiian beach and thankfully was covered over by a lava flow a very long time ago…
Have a great day! Having had a major bike accident, I followed your journey of recovery with great interest.(and that sounds funny written down) I am grateful that you are now able to pursue your work in photography and passion for biking.
DIGLLOYD: regarding the bike accident, I am getting there, improving. I can now work for 3-4 days with high efficiency but as several times before, I just suffered a sort of relapse for 4 days in which I found it challenging just to do simple tasks, like package up some lenses in boxes for shipping and seeing sleep needs go to 14 hours a day. I can exhaust my brain with no warning from one day to the next—alert and strong one day and toast the next—and I don’t know how to gauge it yet. Just six years ago I could work 14 hours a day for two weeks, back off to 10 hours a day for 2-3 days, and do just great.
Howard C writes:
1. I completely agree with your comment about the rank amateurish quality of the comparison in the three videos Hasselblad has put up on its website comparing the H6D-100 with the Nikon D850 with a crappy zoom lens. At the end of your analysis of the flaws in that comparison, you made the following comment: "It cannot be taken seriously as a comparison exercise.” I think anybody with a drop of sophistication about these matters would have reached the same conclusion you did. That begs the question, which is why did you spend some much time and effort in your blog describing in detail all of the flaws in Karl Taylor’s comparison videos? I personally laughed at the videos and chalked them up to ill informed salesmanship.
2. You state that there were recent changes in the management of Hasselblad that are tied to recent changes in the ownership of Hasselblad. As far as I know, the changes in ownership of Hasselblad occurred around January, 2017, when DJI is reported to have acquired control, and there was a change of CEO at that time. However, I am not aware of any changes in ownership OR senior management since that time. Ming Thein was appointed Chief of Strategy after the change in ownership, and I believe he still occupies that role.
3. I thought you received your equipment for reviews from B&H, not the manufacturers. I have noticed that Hasselblad (and Fuji) have been very busy circulating demo versions of the X1D and GFX bodies and lenses to people who are nominally reviewing the equipment, but have no idea what they are doing.
DIGLLOYD: this post had several justifications, some of which I won’t discuss, but can be inferred given my comments. As for “anybody with a drop of sophistication”—I would not put it that way. Not everyone understands the issues and that has nothing to do with sophistication, particularly those new to medium format. I chose to spell out those issues for anyone considering medium format.
I’ve lost track of the changes in management at Hasselblad, but my impression is that DJI ownership of Hasselblad as of 2017 is not a good cultural fit.
I do get most equipment from B&H Photo for review, but any time I have some hope of getting gear in advance, I at least try (usually without success). Plus, gear like an H6D is just too expensive to get on loan from B&H Photo which is why I rarely have the opportunity to review such gear.
Chris R writes:
I’ve read your article earlier about the accurate Hasselblad vs Nikon video!! and thought, ‘Ah, Lloyd has just got a bee in his bonnet about some technical article’ But then I watched the video and found myself too virtually shaking my monitor as the promo video was so bad! and found that all your comments were more than well founded!
I mean, it’s a perfect marketing tool for rich none technical enthusiasts who know next to nothing about camera equipment!
Also, he’s even got filters on the front of the lenses too which will not help whatsoever, especially with an old wide angle lens, it’s not going to do it any favours at all. As you’ve said, there’s more holes in that test than a Swiss cheese! and Hasselblad should be embarrassed for putting such a one-sided marketing video out to try and pull the wool over punters eyes!
DIGLLOYD: I noted the filter issue but forgot to mention it: as I recall, using filters on the Nikon 17-35mm f/2.8 could be a problem in the way they interact with the unusual front lens element (type of element and its spacing). The large non-screw-in filters are not at all likely to interact well, or even be exactly planar to the lens optics.