Previous page Next page

ISO Accuracy

The actual sensitivity of each indicated ISO is measured using the same shots as are used to measure ISO noise levels, we simply compare the exposure for each shot to the metered light level (using a calibrated Sekonic L-358), middle gray matched. We estimate the accuracy of these results to be +/- 1/6 EV (the margin of error given in the ISO specifications). Measured ISO from all four cameras compared in these tests is within +/- 1/6 EV of indicated ISO, which we would expect from professional level equipment. The D3S gives almost exactly the same luminance readings from middle grey at ISO 102,400 as it does at ISO 200, which represents excellent consistency.

ISO Sensitivity / Noise levels

ISO equivalence on a digital camera is the ability to increase the sensitivity of the sensor. In very simple terms, this works by turning up the "volume" (gain) on the sensor's signal amplifiers (remember the sensor is an analogue device). By amplifying the signal, you inevitably amplify the noise too, which becomes more of a factor in images the higher up the ISO scale you go. Adding higher and higher ISO settings in DSLRs is easy - essentially you just apply more gain - but making these settings useable is much harder.

The ISO range of the D3S was briefly, the widest of any camera in the world, until the announcement of the Canon EOS 1D Mark IV. The ability to shoot at ISO sensitivity settings spanning 100-102,400 (equivalent) is certainly one of the main selling points of the D3S, and a lot of potential buyers will no doubt be keen to learn how well the camera performs, in terms of noise and detail capture at the ultra-high settings.

To measure noise levels we take a sequence of images of a GretagMacBeth ColorChecker chart (controlled artificial daylight lighting). The exposure is matched to the ISO (i.e. ISO 200, 1/200 sec for consistency of exposure between cameras). The image sequence is run through our own proprietary noise measurement tool (version 1.6 in this review). Click here for more information. Room temperature is approximately 22°C (~72°F), simulated daylight lighting.

Nikon D3S, Canon EOS 1D Mark IV, Nikon D3, Nikon D300S

  • Nikon D3S: Nikkor 85 mm F1.8 D lens, Manual Exposure, Manual WB,
    Default Parameters (Normal), High ISO NR (Normal), JPEG Large / Fine
     
  • Canon EOS 1D Mark IV: Canon 50 mm F1.4 lens, Aperture Priority, Manual WB,
    Default Parameters (Standard PS), High ISO NR Default (Standard), JPEG Large / Fine
     
  • Nikon D3: Nikkor 85mm F1.8 lens, Manual Exposure, Manual WB,
    Default Parameters (Normal), High ISO NR (Normal), JPEG Large / Fine
     
  • Nikon D300S: Nikon 50 mm F1.4 lens, Aperture Priority, Manual WB,
    Default Parameters (Normal), High ISO NR (Normal), JPEG Large / Fine

  Nikon D3S Canon EOS 1D Mark IV Nikon D3 Nikon D300S
ISO 100
ISO 200
ISO 400
ISO 800
ISO 1600
ISO 3200
ISO 6400
ISO 12800  
ISO 25600 (Hi 1)  
ISO 51200 (Hi 2)    
ISO 102400 (Hi 3)    

It is clear that at their respective base JPEG settings, Canon's approach to JPEG rendering in the EOS 1D Mark IV is slightly different to Nikon's in its D3S, D3 and D300S. Images from the EOS 1D Mark IV show a higher default level of contrast, coupled with slightly higher sharpening compared to those from the Nikon cameras. The higher resolution of the 1D Mark IV's CMOS sensor pays off in better detail capture too, which (unsurprisingly) is especially evident at the lower ISO settings you can see here.

Despite the near-parity in their ISO sensitivity span, the D3S unequivocally betters the Canon EOS 1D Mark IV at ISO settings above ISO 6400. At ISO 3200 and 6400 images from the D3S show slightly lower levels of noise than the 1D Mark IV, but above ISO 6400 the performance gap becomes very clear. However, the EOS 1D Mark IV holds its own against the older D3 very well, and offers extremely similar image quality, right up to the D3's maximum ISO sensitivity of 25,600. Given that the D3 represented the high watermark for high ISO image quality for more than two years, this will no doubt please Canon users very much indeed, and makes the EOS 1D Mark IV a much more useful tool in poor light than its predecessors.

But it's the Nikon D3S we're concentrating on in this review and it is plain to see that in this camera we have a new, and even higher watermark. Although visual examination of noise at pixel level and (ideally) print size is more easily digestible measure of a camera's image quality, the graphs below show that noise levels from its revamped sensor are impressively low.

The D300S turns in excellent performance for its class, but it is clear that it can't hold a candle to the full-frame and APS-H competition against which we're comparing it here. The D300S is useable at ISO 6400, but only just, and preferably in raw capture mode where careful post-capture noise reduction gives best results. In everyday use of course, ISO 3200 and 6400 are at least as high or higher than most people will need to shoot at.

Noise graphs

To keep the separation of ISO settings usefully well-defined, we have split these graphs into two sections, spanning ISO 100-1600 and 3200-102,400 respectively. The Canon EOS 1D Mark IV is the only camera of this set to have an extension setting equivalent to ISO 50 - the other three cameras start at ISO 100 (equivalent). In all four cameras, their lowest ISO setting is a 'pull' setting, achieved by underexposure at the 'true' base ISO (100 in the Canon, and 200 in the Nikons) followed by in-camera brightness adjustment.

ISO 50-1600

  Nikon D3S

Chroma

Gray
Black

ISO 3200-102,400

  Nikon D3S

Chroma

Gray
Black

Indicated ISO sensitivity is on the horizontal axis of this graph, standard deviation of luminosity on the vertical axis.

The data for these graphs is taken from JPEG files with default noise reduction applied. This explains the sudden 'dip' in measured noise between ISO 400 and 800 in the results from the Nikon D300S, and the temporary 'plateau' in measured noise from the D3S between ISO 6400 and 12,800. These temporary deviations from the upwards trend are a result of how the cameras' noise reduction systems work in their respective default configurations.

In the D300s, by default, high ISO noise reduction begins to be applied at ISO 800, and in the D3S, this occurs at ISO 6400. The effect of this noise reduction may be judged from the images shown above. The D3S offers remarkably clean images up to and including ISO 6400, and only at ISO settings higher than 12,800 does noise become problematic in terms of its impact on medium to high contrast detail. Image quality at ISO 102,400 is nothing much to shout about, but at least you can tell what you're looking at, unlike this setting on the Canon EOS 1D Mark IV, which as you can see from the images further up this page, has delivered little more than an intriguingly textured mush.

RAW noise

Finally let's take a look a the D3S's RAW output alongside to the competition. Removing any in-camera noise reduction and processing the images using Adobe Camera Raw (V5.6 in this case, at default sharpening/contrast settings but all NR set to 0) gives us the nearest thing to a 'level playing field' that is practically possible for assessing the relative noise levels of the four cameras' sensors.

  Nikon D3S RAW Canon EOS 1D IV RAW Nikon D3 RAW Nikon D300S RAW
ISO 100
ISO 200
ISO 400
ISO 800
ISO 1600
ISO 3200
ISO 6400
ISO 12800  
ISO 25600 (Hi 1)  
ISO 51200 (Hi 2)    
ISO 102400 (Hi 3)    

Raw Noise graphs

Using a common raw converter makes clear that much of the difference between the cameras' performances in JPEG are, unsurprisingly, down to decisions made by the manufacturers about noise reduction levels. As we would expect, in RAW mode, when the images are processed with no post facto noise reduction, we have measured a lot more noise than we do from the JPEG files. In order to fit all of the data into our graph format in as meaningful way as possible, we've again split these graphs in two, and show ISO 100-1600 and 3200-102400 on separate scales. Note also that because the noise readings are so much higher here than from the JPEG output, we have changed the vertical scales of these graphs.

iso 50-1600

  Nikon D3S

Chroma

Gray
Black

ISO 3200-102,400

  Nikon D3S

Chroma

Gray
Black

Indicated ISO sensitivity is on the horizontal axis of this graph, standard deviation of luminosity on the vertical axis.

Chroma noise gives us the highest measured figures here, and it is no surprise that towards the top of the extended ISO scale of the Nikon D3S and Canon EOS 1D Mark IV the figures are very high. What is significant is that even when noise reduction is effectively removed as a factor in image quality, measured noise figures from the Nikon D3S are significantly lower than both the Canon EOS 1D Mark IV and its predecessor the D3.

Expressing image quality in graphical form has disadvantages, not least that if viewed in isolation it can be frustratingly abstract. However, it does provide a convenient platform for comparisons, and looking at these graphs it is interesting to note that the D3S gives roughly the same chroma noise measurements at ISO 12,800 as the Canon EOS 1D Mark IV does at ISO 3200. Naturally, it is unlikely that any photographer would shoot at these ISO settings without applying some sort of noise reduction, but the comparison is revealing, nonetheless.

Previous page Next page

Comments