The Nikon D90’s Annoying Flaw

Nikon’s long-rumored and just-announced D90 is a pretty wonderful piece of kit by all accounts, but with one notable and annoying limitation buried in its tech specs. It has a 12-bit A/D converter whereas Canon offers 14-bit support in its similarly priced 40D (indeed I can find 40D’s selling from reputable vendors for under $950) and considerably cheaper EOS 450D. While providing HD video capabilities in the D90 is certainly a radical, bold, and perhaps even compelling move — skimping on still picture quality (tonal* range) is pretty annoying.

Perhaps Nikon is trying to differentiate its “pro-ish” D200 and D300 from it’s “enthusiast” D80 and D90 with tonal* range. If so, this seems to me to be a grave¬†misjudgment. Serious photographers will be intrigued by the video shooting capability, but ultimately I think that — effectively — limiting the D90 to inferior quality “film” is very, very bad idea. Nikon never did similarly dumb things with film cameras — why do it with digital?

Nikon is now in the rather odd position of having a huge gap between the D60 (which retails for around $600) and the D200 (which retails for around $1300) which has two Canons (the 450D and 40D) in it, both of which shoot 14-bit RAW.

Now, just how big a difference 14-bit makes over 12-bit in practice is unclear. Pixel-peeping review sites such as dpreview.com haven’t really started looking hard at it, but it’s obviously important enough that both Canon and Nikon have added support for it in some of their cameras. In theory, each “bit” should equate to an f-stop of dynamic range, which would make it ridiculously important.

Anyway, leaving 14-bit RAW support out of the D90 seems like a foolish oversight. It gives people a reason not to buy the D90. The D80 simply did everything it could reasonably be expected to do and then some. It was incredibly successful because there was simply no reason not to buy it. The D90’s lack of 14-bit support will give me, for one, pause. Of course, Nikon may release a D90x with 14-bit RAW once it ramps up production … in six months.

Second Thoughts

If you visit the Nikon’s D90 web page you can see a sample of the video quality, which is stunning. Autofocus is disabled while shooting, so it’s not going to be an all-round camcorder substitute, even with a separate audio recorder (which is a good idea even if you’re using a real camcorder) and an external battery grip. So, video is definitely not a half-arsed feature.

While the 12- vs. 14- bit issue is galling, it’s worth noting that the D90 excels the 450D (and in several cases the 40D and 50D) in some rather more important ways, such as continuous shooting speed (4.5 vs. 3.5 fps), 3D predictive autofocus, autofocus points, metering, and viewfinder quality. So, while the lack of 14-bit NEF support is still grating (reminiscent of Apple blocking dual display support from iBooks to differentiate them from Powerbooks way back when) Nikon is delivering an all-round superior camera.

The street price of the D300 is dropping, too, so I guess Nikon is bracketing the 50D with the D90 and D300, and letting the D90’s superior usability deal with the 40D and 450D.

Correction

As per comments from an alert reader, 14-bit refers to tonal range not “dynamic”. The two aren’t completely independent (as per my responses in comments) but I stand corrected.

Post Script

dpreview has posted sample shots from the Nikon D90 and also the Canon 50D. The first thing I look for is good low light performance, and both produce very usable output even at ISO 6400 (if you pixel peep, it looks like the Nikon produces “clusters” of noise while the Canon produces more outright artifacts, but this is just one photograph per camera of different scenes).

There’s no question that the $1000 D90 trounces the $800 450D. Although the Canon has 14-bit RAW, it is built like a toy, has a pentamirror* viewfinder, maxes out at ISO 1600, uses its flash for low-light focusing, and offers lower shooting speed (3.5fps vs 4.5fps). The 50D barely trumps the D90 for considerably more money with compact flash, 14-bit RAW, and slightly higher resolution… but the D300’s price is now in the same ballpark as the 50D.

Canon has finally released an image-stabilized 18-200mm lens (quite reasonably priced too) which will at least remove that as a reason to go Nikon.

* Every film SLR I’ve ever owned or used has had a pentaprism with excellent coverage, and not one cost more than $350, so why on earth do we find pentamirrors acceptable in $800 digitals?

Canon 50D at 6400 ISO
Canon 50D at 6400 ISO
Nikon 90D at 6400 ISO
Nikon D90 at 6400 ISO
  • Giorgos

    Well, dynamic range has nothing to do with the A/D converter. A 14bit A/D converter may(again, may)give smoother gradiations between higlights and shadows, but will definitely not affect the dynamic range of the sensor.
    cheers,
    George.

  • Technically, you’re right. Nothing in the A/D converter affects the sensor. The question is what you get out of it. What you refer to as “smoothness of gradations” amounts to “dynamic range” in that it increases the “sweet spot” for correct exposure. In essence, shooting in 14-bits is the equivalent of having bracketed a 12-bit exposure at +2/-2 EV (except for clipping). Shadow and highlight detail which would be blown in a 12-bit image will still be there in a 14-bit image.

    In Photoshop terms, this means that if you take an over- or under- exposed image you can move the white-point or black-point the equivalent of two full stops and still have as much color detail as a correctly exposed 12-bit image.

    This isn’t dynamic range in the sense of increased absolute latitude, but it is dynamic range in the sense of increased useful latitude.

  • Giorgos

    According to this site: http://www.nx101.com/12vs14.html there is a difference between the 12 and 14 bit mode in D300’s raw files.
    Again, this difference is in the tonal range(smoothness of gradiations),not in the dynamic range and it can only be seen after a lot of post processing.
    Still,there is a difference and it proves that 14bit raw files are not just a marketing gimmick.
    Personally, i don’t think that nikon used 12bits just to cripple the D90 in favour of the more expensive D300. Officially, they don’t share the same sensor, and they don’t have the same megapixel count(not the effective one,as a whole). It is worth noting also that a 14bit A/D converter costs more than a 12bit one. Maybe some parts are common and some others not, but these sensors should not be identical.
    Anyway, it’s picture quality is amazing an i will buy it when i have all the money-i’m stuck at the 70% of its price for a couple of weeks now,damn…

  • My suggestion that there was no difference between 14- and 12- bit RAW files was facetious. Obviously there is one, and obviously Nikon hopes to make people consider the D300 over the D90 based on image quality alone (and not just build quality, autofocus accuracy, etc.).

    Now, going to the tonal- versus dynamic- range point. Yes, technically 14-bit is all about tonal range, but it also determines how much of the dynamic range is actually usable.

    In non-technical terms — dynamic range = latitude (how much can I over- or under- expose and still get decent results). That’s how I think of it, and in those terms there’s a good deal of overlap between tonal- and dynamic- range.

    If you take a given photograph and adjust it to bring out shadow detail, for example, you’ll quickly see that there’s very tonal little detail in deep shadows. 2-bits = 4x as much detail. Generally I can push a RAW exposure by about 2 stops in either direction without producing something that looks like crap. 14-bit should give me an extra stop of two.

    This is not a “lot of post processing” — just one curves or levels adjustment.