G3 - Image quality (theory)

The G3 has a sensor measuring 17.3mm x 13mm, compared to the 6.17mm x 4.55mm of the SX10. This meant the G3 sensor had eight times the surface area of the SX10 sensor and so would collect eight times more light for a given exposure (e.g. using the same 1/100 sec at f/5.6 with the SX10 and the G3, the G3 would collect eight times the amount of light). This would give a less noisy G3 image with a higher signal to noise ratio, which would in turn allow more detail to be extracted from the image than with the noisier SX10.

However it turned out to be not so straightforward. This was my first introduction to Equivalence, which is about how images vary (or not), for example in terms of noise and depth of field, when captured with cameras with different sized sensors. This is a surprisingly controversial subject, at least on the dpreview forums, and especially the Micro Four Thirds Talk forum, which I started frequenting around the time I got the G3. For years there have been arguments about equivalence on that forum, sometimes so acriminonious that threads have been locked by the moderators, and this on a site which is pretty permissive about what is allowed. One school of thought has it that equivalence is a well-defined idea which is useful in practice. On the other side are those who think, variously, that equivalence is wrong, irrelevant to real world concerns and/or part of a conspiracy on the part of advocates of full frame cameras to do down micro four thirds cameras and/or the people who use them.

From my point of view equivalence came down to a few simple, practical issues. The key issue, which drove the other considerations for me, was that for invertebrates I was in the habit of using the SX10's minimum aperture of f/8 to give me as much depth of field as possible. To get the same depth of field with the G3 I had to use f/22, the minimum aperture of the 45-200, on which I mounted my close-up lenses. 

Using f/22 on the G3 meant that in order to get the same shutter speed as on the SX10 using f/8 I had to use ISO 800 on the G3 rather than ISO 100 on the SX10. Alternatively I could use the same ISO on both cameras but use an 8 times longer exposure on the G3. Or I could use some other combination of higher ISO and longer exposure on the G3. Given that I was using available light much of the time and it was often breezy, an exposure 8 times longer was often impractical, and it turned out that I used the G3 at ISO 800 most of the time when using natural light so as to get the same shutter speeds as I would have been using with the SX10. This meant that I was not collecting 8 times the amount of light with the G3 as with the SX10, but the same amount, and so I didn't get the advantage of less noise with the G3 or the benefit of being able to extract more detail because of having less noise. In practice it turned out (as equivalence theory predicted) that the noise I got at ISO 800 with the G3 was very similar to the noise I got at ISO 100 with the SX10.

Of course I could add light to the scene using flash. That way I could use base ISO with the G3 and get better, less noisy results from its larger sensor. However, that turned out to be problematic. I had bought a quite powerful flash, but after adding diffusion to it I had to use it at full power when using f/22 on the G3, and sometimes even that led to under-exposure. The recycle time at full power was too long for my purposes and I settled on turning the ISO up to 800 which then let me use the same flash output and the same, acceptable to me, recycle times as with f/8 on the SX10. So in practice I was getting rather similar noise from the G3 as from the SX10 whether I use flash or available light.

It also turned out that the sharpness/detail I got with the G3 at f/22 was very similar to what I got with f/8 on the SX10. This has to do with diffraction, which is related to the physical size of the aperture - smaller aperture, more diffraction; same size aperture, same amount of diffraction. It turns out that the 45-200 at f/22 on the G3 (or any other micro four thirds camera) has almost the same diameter aperture as f/8 on the SX10 (or any other 1/2.3" sensor camera) when focal lengths are used which give the same angle of view on both cameras. With the aperture being the same physical size, the amount of diffraction and consequent loss of detail is the same in both cases. Because the effects of diffraction dominate the image quality at such small apertures the image quality is very similar in terms of detail captured.

Aside: How come the aperture is the same with one at f/22 and the other at f/8? Here is an example. If the 45-200 is at 200mm then the SX10 has to be at focal length around 71mm to give the same angle of view. The diameter of the aperture is the focal length divided by the f-number. For the 45-200 at 200mm at f/22 this is 200mm/22, which is 9mm. For the SX10 at f/8 it is 71mm/8 which is 8.9mm.

The situation was different for my botanical subjects. As outlined at the end of the post G3 - No macro lens there was a potential image quality advantage in using a larger sensor.

Another potential advantage of the G3 was that it could capture raw, thus giving me the benefit of using the extra information available from a raw file, for example for the better recovery of highlights and shadows. Conveniently the G3 came with a free copy of Lightroom which I could use to handle the raw files. However, for reasons I'm not clear about, I did not start using raw or Lightroom until over a year later, in October 2012. So for the first year and a bit I was still shooting JPEG and using Photoshop CS2 for post processing. After that I started shooting raw and using Lightroom in combination with CS2 (which I think I used for sharpening and any cloning that needed doing).