Don’t confuse total dynamic range with usable dynamic range. The total range is what is measured by test rigs as the range between lightest highlight and darkest points lost in noise. The useful range is rather how low you can go until the noise becomes objectionable, i.e. anything further is not useful. I think that imaging-resource is among the best testers of useful dynamic range, so let me borrow a paragraph from them:
A key parameter in a digital camera is its Dynamic Range, the range of brightness that can be faithfully recorded. At the upper end of the tonal scale, dynamic range is dictated by the point at which the RGB data “saturates” at values of 255, 255, 255. At the lower end of the tonal scale, dynamic range is determined by the point at which there ceases to be any useful difference between adjacent tonal steps. Note the use of the qualifier “useful” in there: While it’s tempting to evaluate dynamic range as the maximum number of tonal steps that can be discerned at all, that measure of dynamic range has very little relevance to real-world photography. What we care about as photographers is how much detail we can pull out of the shadows before image noise becomes too objectionable. […] What makes most sense then, is to specify useful dynamic range in terms of the point at which image noise reaches some agreed-upon threshold. (from http://www.imaging-resource.com/PRODS/K5/K5IMATEST.HTM)
In the same article, they go on to explain how to define usable range and how to test it. Thereafter comes the really interesting part. I have chosen an older test, that of the Pentax K-5, since it contains the Nikon D70 – the topic of this blog – in its reference tables.
The useful dynamic range table in that article is a REALLY interesting read. It explains why countless folklore stories of superior dynamic range of full-frame cameras never seem to materialize in real-world images and real-world pulling of shadows. The Nikon D3S and D3x have worse useful (real-world) dynamic range than the D7000. At least 1/2 stop and even up to a full stop worse. The D3 and D700 are even further behind. The latter ones are equal to the good old D90. Make sure you swing by the URL above and have a look at that revealing table, look in columns 0.25 and 0.1, no others (the table is sorted by the 0.1 f-stop column).
In the same table, you can see the D70 fare pretty badly. One unsatisfying thing about the table is that only the JPEG range is shown for the D70 while for many others, both JPEG and RAW are shown. RAW beats JPEG by a wide margin, and the most indicative figure for the D70’s RAW performance ought to be the D40 figure of 8.30 (beating the D3100 and being close to the D300). The table underlines the well-known truth that you should shoot RAW at least in scenes with higher dynamic range and that you can pull a lot more out of the sensor than JPEG engines can, especially a ten year old one. Let a modern RAW converter such as Capture NX-2 or Lightroom 5 do the work and like magic you have bypassed ten year old software in the D70 camera body. Another unsatisfying fact is that the table lists the camera bodies at base ISO, which for the D70 is ISO 200 making it look much worse than some of its rivals. It’s a full stop behind the full-frames D3S, D3x and D700, and two stops behind the D7000 which is the usability king in this respect (confirming real-life experiences). Nowadays, the site does not publish a table, only individual scores. While the D800 did only fairly well (above D3x but below D7000), the D600 scored the best so far by 1/3 stop. The D4 equalled the D7000 as did the D7100. This all underscores the truth that you should choose half- or full-frame firstly based on your desires regarding depth-of-field, and secondly on your interest in high-ISO shooting. Dynamic range is not a factor in this.
Speaking of the D90, even better figures are achieved by its slightly younger and much lighter sibling, the D5000. An overlooked gem that you can pick up for $150, often with little mileage. It is the modern counterpart of the Nikon FM, FM2 and FM3, and of course the FE and FE2. Small, light, excellent. Pair it with the smallest 18-55 kit lens and a couple of small primes and you have a fantastic real DSLR travel and take-anywhere kit. The prime selection is actually quite simple: Nikkor DX 35 f/1.8 G and Nikkor 85 f/1.8 G. Both truly outstanding in image quality and fairly small. If you want something wider, that’s a little bit problematic. I suggest you stay with the Nikkor 18-55 f/3.5-5.6 G VR. It’s good, especially at wider focal lenghts, but not outstanding. Or go with the Voigtlander 20 f/3.5 SL-II. Better image quality than the Nikkors, both autofocus and manual focus versions. Especially good on DX cameras. It has manual focus but the D5000 recognizes the chip in the lens and thus it’s able to meter automatically in the normal way. Very small and light, plus quite easy to focus manually since it is an ultrawide lens. Huge depth of field on a DX camera.
Another notable entry in the table is the D3100 which I own and write about in the sensor blog entry. Its old Toshiba CMOS fares worse than the CCDs of the D40 and D60. This is also consistent with how it works in practise, pulling shadows in some photo editor. A final note is that the table confirms the long held view that both the D300 and the D300s are inferior to the D90 and D5000 when it comes to useful dynamic range. The virtues of the D300 bodies rather lie in durability and handling, among other things.
The Canon bodies are behind the equivalent Sony equipped bodies (most Nikons, Sony, Pentax). The Toshiba sensor of the D3100 is behind similar Sony sensors. The old Fuji S3 Pro is impressively placed before all Canon bodies and not that much behind the best Nikon ones, even if the gap up to the D7000 is considerable.
It’s a surprisingly little discussed fact that DxOmark‘s dynamic range measurements are made using total dynamic range instead of useful range. I would love to see the imaging-resouce table plotted in a 2D graph for every ISO value. I am sure such a diagram would correspond more closely to how different cameras are percieved and felt during post processing. But as of lately, imaging-resource have given up their measurements because of their test instruments being surpassed by recent sensor developments. Unfortunately, they now use DxOlab’s figures and charts that only correspond to the least informative column of imaging-resouce’s tables, the leftmost one showing unusable (total) dynamic range.
Perhaps this rather long entry sheds some light on why my sensor blog entry is true even though a lot of people wish it wasn’t. Or perhaps it does not…in which case I have to try and rephrase my thoughts once more…