ISO Sensitivity / Noise levels

ISO equivalence on a digital camera is the ability to increase the sensitivity of the sensor to enable faster shutter speeds and/or better performance in low light. The way this works in a digital camera is by "turning up the volume" on the CCD's signal amplifiers, nothing is without its price however and doing so also amplifies any noise that may be present and often affects colour saturation.

Just like the EOS-D30 the EOS-D60 provides four selectable ISO sensitivity equivalence's from ISO 100. Unlike the EOS-D30 however the D60 doesn't have ISO 1600, instead it has ISO 1000.

I decided to use the same system created for our EOS-1D review. This involves shooting a colour patch chart (a GretagMacBeth ColorChecker) at the full range of ISO sensitivities. To expand this test a little the patches were shot at both normal daylight levels (10.6 EV) as well as low light levels (6.3 EV).

Each of the comparison crops below are made up of a 90 x 80 crop of patches 19, 22 and 24 of the ColorChcker chart (as indicated above). This helps to give a better impression of noise at different light levels within an image. Directly below this you will find the average standard deviation of these three patches, this is a quantifiable measurement of noise.

Test Notes

• We are now using the GretagMacBeth ColorChecker instead of Kodak Colour patches for a couple of reasons: (a) The grey patches are larger and are therefore easier to crop / measure (b) The patches are made from a matte non-reflective material and are therefore easier to shoot without accidental reflection.
• Measurements were taken at normal room temperature of 21°C (~70°F), you should expect noise to reduce in lower temperatures and increase in higher temperatures.
• Performance comparison carried out in RAW mode to remove the effects of JPEG artifacts skewing the higher ISO results. RAW images were converted to 8-bit TIFF's using Canon RAW Image Converter v2.0 (default settings).
• Camera settings were defaulted before tests carried out (thus default parameters, sharpening etc.). The visibility of noise will be reduced by decreasing sharpening.
• The 'standard deviation' was measured using the histogram feature in Photoshop on 8-bit TIFF or JPEG files (later comparison). Lower figures mean there is less 'deviation' from the average luminance level, and thus less noise.

Performance of EOS-D60 vs. EOS-D30 in normal and low light

Settings: Parameters: Normal, EF 50 mm F1.4 @ F9.0

	Normal light (10.6 EV)			Low light (6.3 EV)
Camera	Canon EOS-D60	Canon EOS-D30		Canon EOS-D60	Canon EOS-D30
Sensitivity	ISO 100			ISO 100
Exposure	1/10 s, F9.0			1.0 s, F9.0
Patch crops  ISO 100  RAW
Standard dev. avg.	0.64	0.93		0.68	0.98
Camera	Canon EOS-D60	Canon EOS-D30		Canon EOS-D60	Canon EOS-D30
Sensitivity	ISO 200			ISO 200
Exposure	1/20 s, F9.0			0.5 s, F9.0
Patch crops  ISO 200  RAW
Standard dev. avg.	0.94	1.20		0.95	1.20
Camera	Canon EOS-D60	Canon EOS-D30		Canon EOS-D60	Canon EOS-D30
Sensitivity	ISO 400			ISO 400
Exposure	1/40 s, F9.0			1/4 s, F9.0
Patch crops  ISO 400  RAW
Standard dev. avg.	1.37	1.74		1.36	1.78
Camera	Canon EOS-D60	Canon EOS-D30		Canon EOS-D60	Canon EOS-D30
Sensitivity	ISO 800			ISO 800
Exposure	1/80 s, F9.0			1/8 s, F9.0
Patch crops  ISO 800  RAW
Standard dev. avg.	2.30	2.74		2.31	2.76
Camera	Canon EOS-D60	Canon EOS-D30		Canon EOS-D60	Canon EOS-D30
Sensitivity	ISO 1000	n/a		ISO 1000	n/a
Exposure	1/100 s, F9.0	n/a		1/10 s, F9.0	n/a
Patch crops  ISO 1000  RAW
Standard dev. avg.	2.84	n/a		2.86	n/a
Camera	Canon EOS-D60	Canon EOS-D30		Canon EOS-D60	Canon EOS-D30
Sensitivity	n/a	ISO 1600		n/a	ISO 1600
Exposure	n/a	1/160 s, F9.0		n/a	1/15 s, F9.0
Patch crops  ISO 1600  RAW
Standard dev. avg.	n/a	4.38		n/a	4.44

Having now removed the element of JPEG artifacts from the test we can see that both the D30 and D60 manage to maintain relatively low noise levels all the way through the available ISO sensitivities. Our measurements reveal the D60 to have lower noise levels compared to the D30, however a visual comparison shows that in effect the differences are hardly noticeable. The fact that Canon's engineers have managed to slightly reduce noise levels while effectively halving the photosite size is an impressive achievement. One thing I'm still curious about however is why the D60 doesn't support ISO 1600, it's clear that noise levels wouldn't have been too bad at that setting. Below is a summary graph of these results.

The other thing this comparison shows is that there is also very little difference in noise between fast and slow shutter speeds. For example, at ISO 800 1/80 sec the D60 measures 2.30 std. dev. average and almost identically 2.31 at a much slower 1/8 sec. There isn't much point measuring noise levels at shutter speeds beyond 1 second because here the camera's inbuilt long exposure noise reduction systems take over and thus results will be skewed.

Performance of RAW vs. JPEG in normal light (EOS-D60)

Finally to confirm what we know of JPEG compression. Below is a graph of noise levels from JPEG files versus noise levels from the RAW files above. As you can see the JPEG compression algorithm introduces its own artifacts which can in effect amplify noise at higher sensitivities. Having said that the trade off between a little more noise (a hardly a visible difference) and the extra storage / write time expense of a 9 MB RAW file depends totally on the priorities of the photographer.