Originally posted by Byrd-2020 I quote Roger Clark from Clarkvision.com: "The number of photons a digital camera collects in each pixel is directly related to the size (area that converts photons into electrons) of the pixel and the lens feeding light to those pixels. The more photons collected, the better the signal-to-noise ratio in the image, thus the larger pixel sizes using larger lenses do better in this regard. Larger pixel cameras have better signal-to-noise ratios at all levels, but this becomes more obvious especially at low signal levels compared to cameras with smaller sensors which use correspondingly smaller lenses. In the extremes of current digital cameras with small cameras having pixel sizes near 2-microns, and large pixel cameras (currently found in DSLRs), there is a factor of about 12 to 16 in photons collected. That means the large pixel camera performs at ISO 1200 to 1600 with similar noise and dynamic range performance of a small pixel camera operating at ISO 100." (part 2 of "Digital Cameras: Does Pixel Size Matter")
Clark's conclusions are contentious. I can think of 3 scientists/engineers who disagree with him. Emil Martinec's take on this topic is here:
Noise, Dynamic Range and Bit Depth in Digital SLRs -- page 3 Quoting:
Bottom line: Among the important measures of image quality are signal-to-noise ratio of the capture process, and resolution. It was shown that for fixed sensor format, the light collection efficiency per unit area is essentially independent of pixel size, over a huge range of pixel sizes from 2 microns to over 8 microns, and is therefore independent of the number of megapixels. Noise performance per unit area was seen to be only weakly dependent on pixel size. The S/N ratio per unit area is much the same over a wide range of pixel sizes. There is an advantage to big pixels in low light (high ISO) applications, where read noise is an important detractor from image quality, and big pixels currently have lower read noise than aggregations of small pixels of equal area. For low ISO applications, the situation is reversed in current implementations -- if anything, smaller pixels perform somewhat better in terms of S/N ratio (while offering more resolution). A further exploration of these issues can be found on the supplemental page. Rather than having strong dependence on the pixel size, the noise performance instead depends quite strongly on sensor size -- bigger sensors yield higher quality images, by capturing more signal (photons).
The other main measure of image quality is the resolution in line pairs/picture height; it is by definition independent of the sensor size, and depends only on the megapixel count. The more megapixels, the more resolution, up to the limits imposed by the system's optics.
Martinec's article mainly dates from 2008 and uses the 20D as the example. Since then, fill-factors have doubled and ADC noise has been reduced greatly in some brands of cameras. So his conclusion would be even more affirmed today. Check it out for yourself:
Side by side - DxOMark
Notice how the DR is similar for the D7100 (3.9µm sensel pitch) vs the K5IIs (4.75µm sensel pitch - both cameras are APS-C) when you look at similar sized final output sizes (the "Print" tab). If you "pixel peep" (the "Screen" tab) instead (look at individual sensel performance) you'll see a difference in DR. The reason the "Print" performance is so good these days is that the fill-factor is now very high. So the
overall result is similar, because little light is being lost any more in the "cracks" between sensels.
Let's compare against some FF cameras now: K5IIs (4.75µm, APS-C) vs D800 (4.75µm, FF) vs D600 (5.9µm, FF):
Side by side - DxOMark
First off, look at the SNR (18%) ("Print" tab). This is a proxy for Quantum Efficiency & Fill Factor. As you can see, both FF cameras are about 4db (1.3 stops) better than the APS-C when the
overall sensor output is considered. This is due the the 2.34x larger FF sensor compared to APS-C. If we switch to the "Screen" tab, where the effect of sensel size is considered, rather than the total output performance, we see that both 4.75µm sensels perform much the same, while the bigger 5.9µm performs better.
Now, consider the DR. With the "Print" tab (the overall output when considered at the same display/printout size), both FF cameras are similar, with the APS-C sensor, about 0.6-0.7 stops worse. When comparing sensel performance ("Screen" tab), we again see the smaller sensels grouping together, with the bigger sensel performing better.
What about higher ISO/low light performance? I think the best indicator of this is the DxOMark Sports score. This is the highest ISO setting for a camera that allows it to achieve an SNR of 30dB while keeping a good dynamic range of at least 9 EVs and a color depth of at least 18bits. While any of the three (SNR, DR, Colour Depth) can be the criterion performance limit which is reached first and which thus determines the score, usually it's the SNR. Let's compare 3 recent high-performing MFT vs APS-C v FF cameras. I've used the K-3 as the reference:
Side by side - DxOMark Camera | Sensel Pitch | ISO/Score | Score Ratio | Sensor Area Ratio |
D800 | 4.7µm | 2853 | 2.34 | 2.34 |
K-3 | 3.9µm | 1216 | 1 | 1 |
OM-DE-M5 | 3.7µm | 826 | 0.61 | 0.68 |
As you can see, the ratio of the sensor areas is a good indicator of the LL performance. It's not the sensel size that's making a significant performance difference here, since the overall sensor performance is being considered.
So, I consider it a waste of time fretting that the trend to smaller sensels is ruining sensor performance. (It's true that more sensels/mm2 increase storage & buffer size and need more processing power for the same speed.) It's better to look at the overall sensor performance, rather than looking at the size of the shrinking sensels. I see the role of more sensels per mm2 being not to enable larger printouts, but to provide more resolution (a significant factor in perceived quality), finer grained shot noise (intrinsic to light) and less visible demosaicing artifacts, at the same printout/display size.
Dan.
Last edited by dosdan; 02-05-2014 at 02:11 AM.