Another reason why we may not see a medium format Nikon

Flashback: “The D3x gave medium-format looking files with all the flexibility of the Nikkor Lens system”

Dxomark tested sensors from several medium format cameras: Mamiya ZD Back, Leaf Aptus 75S, Hasselblad H3DII 39, and Phase One P45+. Now medium format can be compared to the D3x (in theory):

Related article: "Especially at low ISO, the Nikon D3x offers the cleanest signal path yet seen in a DSLR, as good or even better than some medium format systems I’ve previously tested".

I think it is time to remove the Nikon MX tag.

Screenshots: cnet

This entry was posted in Nikon D3x and tagged . Bookmark the permalink. Trackbacks are closed, but you can post a comment.
  • Dan

    I guess higher megapixels does not offset noise or does higher pixel pitch mean better low light performance in this case!!!!!!

  • I think it’s a very interesting comparison between the formats, but I have to say that straight comparisons like this may not tell the whole story. It’s a comparison of sensors, and I think that this does show that the smaller D3X sensor does indeed compete with the much pricier, much larger and much more difficult to fabricate sensors of it’s larger cohorts.

    The problem I see, however, is the quality of glass to seriously drive a sensor of that resolution. Take your average 35mm equivalent DSLR and strap your average quality glass onto the mount at approximately 50lp/mm, take a shot and you’ve got a recorded image that’s about 24mmx35mm at 50lp/mm assuming the sensor is capable of meeting the full resolution capability of the glass. At about 5lp/mm a sharp eye will begin to notice the resolution of a photo breaking down. An average eye will see resolution breaking down at around 3lp/mm. So, with average glass at an average rating of 50lp/mm, one could enlarge that original 24mmx35mm up to either 10 or 12 times its original size before noticing degradation in image quality. That leaves us with a maximum print size of 24cmx35cm prior to noticeable resolution degradation using a 35mm sized sensor that takes advantage of an average resolution lens. Obviously, if you had a piece of glass that could kick out 100lp/mm (good luck) you could extrapolate a much greater possible limit of print size prior to image degradation.

    Your standard Digital Medium Format system, operating on an equivalent 6cmx4.5cm sensor (actually, the sensors are a little smaller than this), with the same 50lp/mm glass might result in an image that can be blown up to the equivalent of 60cmx45cm, or nearly 3 times as large prior to a break down in overall visible resolution degradation.

    The problem as I see it, is that the 35mm class of sensor is now outclassing the available 35mm glass, while the Medium Format sensors, probably due to the sheer difficulty of producing large enough pieces of silicon at relatively low prices, has not reached the same degree of resolving power to meet the glass available for the system.

    Sure, the D3X is leaning into MF’s territory, but without improvements in glass, it will eventually be greatly outstripped by MF, assuming MF systems devise sensors that take natural advantage of their size and their lenses’ resolving power.

    For now I think the D3X is a viable alternative to current MF solutions. In the future I don’t think it will be so.

    And, without vast improvements in glass, the comparison in sensors is, at least in my reckoning, purely academic and the comparison across media is a difficult one.

    • Jason

      Do you have any figures for the resolving power of various popular and/or pro-quality Nikon lenses? If not, does anyone else out there?

      • rhlpetrus has all lenses tested

        • Jessica

 has 54 Nikon brand lenses tested. Nikon’s current compatibility matrix runs into 216 lenses, so it’d be unfair to say has tested all lenses when they’ve only tested 1/4th of the current compatible lenses.

    • rhlpetrus

      I agree, Nikon needs to launch an ultra HQ prime lens line. The smaller sensor actually requires better lenses than the larger ones, since the magnification to reach same print/viewing size amplifies lenses’ problems. This has always been the case, MF and LF lenses are actually simpler technically, less number of elements, etc., than 35mm lenses of similar quality.

    • RIW

      Using a D3x my impression on charts and practically is that in landscape and Macro use the Micro Nikkor 105 mm 1:4 far outstrips the sensor. The PC Micro Nikkor 85 mm 1:2.8 D is nearly as good. The AF-S Micro Nikkor 60 mm 1:2.8 G ED is slightly less good – on a par with the sensor except at the edges? However the remaining lenses compared (Macro, Prime and ‘Pro’ zoom) are not in the same class. I do not have any long exotics to consider.

      It is a pity the 105 f 4 Micro is difficult to focus with old eyes except with intense light. I look forward to learning which other new or old lenses practical photographers find really perform with the D3x.

    • Erik

      Can someone explain how a compact camera can resolve anything when you say that the size difference between FX and MF is so big and makes such a big difference?
      Compact camera lenses must have a very high number of lines per mm in other words superior optical quality?

      please correct me 🙂

  • Anonymous

    imagine if you stitched 2 d3x sensors side by side? then you would have an mx camera

    • Pablov

      They do that in Astrophotography (not D3x sensors, but a lot more expensive ones)

  • WOW, I’m totally impressed, Nikon did it! MX quality, but with the decades-old F-mount! This calls for a blog post!!!


  • Yet another winnig award for Nikon… thanks mum!

  • Please, take a look here, before to start screaming of joy.
    DxO test is just a “sensor-sided” test. It has not that much to do with the real image quality.

    • rhlpetrus

      Reichman’s comments are interesting, but some of his claims are not correct or have ane relation to issue. For example, he says CMOS requires in-senro NR circuitry that could reduce IQ. Where’s the evidence. The D3x has better IQ in all aspects compared to any dslr using CCD. And there’s a test 1DsIII x P25 right there at LL showing the Canon has at least same detail (and looses re DR and color by a hair, exactly what the DxO mark is showing). The D3x is better than the 1DsIII both re DR and color, and, from measuremnts, better than all MF backs tested. So, the D3x is clearly in the MF league re IQ aspects, and miles ahead in everything else. One issue though: primes for HQ work are not available in equivalent terms with MF lenses, so Nikon should invest in that sector, a have an ultra HQ series of primes to go with the D3x. Great zooms are ok, but not the type of lens a studio or a landscape shooter would use frequently. PC lenses are ok but one still needs great basic primes.

  • Pete C

    I wonder what the test results would have been using a MF adaptor and using the same lens on the Nikon as the MF bodies?

    I’m curious to see when DxO will start using the same body to test the resolution of different lenses.
    That would be a wake up for a lot of fanboys.

    • rhlpetrus

      The test does not use lenses, it’s just a sensor RAW test, essentially a specification test.

      • Pete C

        Teach me to not read.

  • markdphotoguy

    I’ve used MF lenses on 35mm bodies and quite frankly they are about the same as a high quality 35mm prime. The reason is simple as you go up in capture area (size of film or imaging sensor) the resolving power of the lens does not have to be as high for the same size of print. This doesn’t mean MF lenses aren’t good, they have superior flatness of field (corner to corner sharpness) wide open as well as better contrast in crapy low lighting situations.
    The big difference between MF and 35mm in my experience has always been the capture area not resolution. The larger your capture area the better tonal gradation across frame (not to be confused with bit depth) you will have, giving the image a more three dimensional look over the same shot on 35mm (FX), DX or a compact camera. This is the reason why a 12Mp compact camera will not match the IQ of a 12Mp DX camera. Sure the pixels are larger and better at gathering light on the DX but there will still be a difference in IQ under good lighting that gives the best result with the compact. That is one of the reasons (perspective correction being the other) catalogue photographers used 4×5 over MF in the film days, it was to get the smoother tone, they didn’t need the resolution for the catalogue where the max equivalent print size was about 9″x13″.
    I for one am sad that the economy tanked the way it did because we are not likely to see an MX camera now which is too bad, It would have been the first step into an affordable MF system for the rest of us (read: not pro with $$$, not wealthy) serious photogs who are unable to budget for the MF digi backs.
    DxO rankings are not an indicator of photo quality but of data quality (some would even argue it doesn’t even give that) which is not the whole story of what makes an image good and should be taken with a grain of salt.

  • WTF?

    This is an awful post. You need to read what DxO does before jumping to a conclusion like that.

    Hint: They do something to even out the resolution advantage.

  • Jj

    Assuming that DxO stats are “true” in the sense that difference in their numbers corresponds to proportional difference in image quality ( that is there are no other – not covered by DxO – factors that have influence onto image quality) then we can make following observations:
    according to DxO, nikon’s 35mm camera sensor is strictly better than competitors’ *larger* MF sensor by all means.
    MF cameras (generally) when compared to 35mm have bigger aperture (“input hole”), thus collecting more light, thus having more data (fotons) on their input.
    So at least theoretically MF cameras should be capable of getting better results by distributing that larger (than in 35mm) number of fotons over either larger number of pixels (of same size as in 35mm) getting higher res, or – with larger pixels – get better fotons-per-pixel ratio resulting in better dynamics/ tonal range/ whatever.
    // I suppose the issue of optics quality is of less importance here – even if you have crappy lens that can do only 10 lines per mm in the center – it will not increase sensor’s noise or worsen color redention (which are key factors evaluated by DxO).

    so if both 1) and 2) are true, and nikon can get more from smaller sensor it means one thing: nikon’s technology is *far* better than that of leaf/mamiya/hassy in terms of sensor’s preceived optical density, sensor’s noise level, low noise amplification of signal, low noise AD-conversion and noise-removal in postprocessing.
    But it also means that, if nikon was manufacturing digi backs for MF, they could take their current D3x sensor, scale it up to 645 format and get the same per-pixel quality as in d3x with proportionally larger number of pixels (and maybe smaller fps).
    So altogether it doesn’t mean that 35mm format is “as good” as medium, and that medium will eventually die in uneven battle with fanatstic MX. It means that Leaf/Mamiya/Hassy/whoever_else should improve their technology to get the same quality-per-square-inch-of-sensor as nikon has right now. With the same technology as nikon and bigger sensors they should always be able to get better result form bigger camera. (IMHO 😀 )

    • Anonymous

      Unfortunately (1) and (2) both fail to make sense.

      (1) “Strictly better” doesn’t mean anything.

      (2) Larger format cameras don’t tend to have lenses with wider apertures.

  • DxoMark isn’t necessarily an absolute truth, you know. A lot of critical MF shooters would take exception to your conclusions. I’d say keep the MX tag around at least over the next few months.

    DxoMark has been around, what, six months? Give it some time – and competing benchmarks. Maybe we’ll see a more broadly accepted, proven benchmark a few years down the road. It’ll take a while.

  • MB

    I haven’t found exact description of methodology used by DXO in their testing, but from what I have found it seams that they are using RAW data as a source for a comparison and that implies that the results are affected by the quality of a decoding engine used and not only by the quality of the sensor (nothing else but the sensor data is really compared as far as I can tell).

    Nevertheless we must admit that Nikon has made a great job on D3x sensor, even if compared only to other FX format cameras. If only they haven’t tried to charge it 4000$. At that price it also makes something with that sensor, like D700x, very unlikely because it would cost couple of thousands more than D3.

  • Looking at the prices of medium format systems, the D3x looks a trifle under-priced considering the price vs performance. Calumet has the current PhaseOne with an 80mm lens for $45,990.00. The back alone runs $42,990.00.

    Of course, DxO is testing sensors ONLY. Were it to consider the whole system, there is even more to like. We bitch about wanting Nikon to come out with more lenses, but no medium format system offers anything close to what Nikon or Canon offer.

    On the same site, only 15 are offered by Hasselblad, compared to 82 for Nikon and 62 for Canon. Furthermore, there is a lot of redundancy in focal lengths, since Hassy lenses tend to be camera-line specific. B&H shows ten lenses for the H series, which is the one primarily built for digital photography.

    My “normal” lens, the excellent f/1.8 50mm runs a bit over $100US. The normal lens for Hasselblad, an f/2.8 80mm is $2040.99. The cost of being nicely equipped is going to be a small fraction of what a similar setup would be with medium format, if the lenses were even available. For one wanting a high pixel count, it seems clear that D3x is the best and cheapest way to go.

  • Paradise

    Some MF back tested are 4 years old ! Do you compare them to the D2x ? No.

    And what about sharpness ? Dxo don’ t test it. D3X has a Anti aliasing filter so the picture are softer… That’ s the only reason I just bought 3 weeks ago a Leaf Aptus II 7 ( build in 2008 thanks )

  • Paradise
    • Anonymous

      Um, you’re looking at resolution. DxO mark is looking at other parameters like DR. If resolution is all you care about, the DxO mark would not matter to you.

  • HDZ

    Hope this king was kill every-one.

  • Back to top