Hi Keith
Thanks for your comment; for those of us who write some of these articles, it's good to get feedback and to know that members are reading them and thinking about the issues.
I think your comment: "So the 300mm Olympus lens becomes 600mm in 35mm equivalence" is precisely the issue that Ian Wilson's paper is trying to address, with detailed evidence and calculations.
There are probably two key separate but related points in Ian's paper. First is the issue of 35mm equivalence; and then there is the issue of sensor noise performance.
35mm equivalence is a concept related to sensor size. A 300mm lens remains a 300mm lens regardless of which camera it is attached to. The focal length, and degree of magnification, does not change as you move from a cropped sensor to a full frame sensor. What does change is the amount of the scene that is being recorded. A smaller sensor records a smaller part of the image (ie has a smaller field of view or angle of view) - in other words, the pixels are placed on a smaller area. Superficially, this looks like you are using a longer telephoto lens than you actually are. However, crucially, the final image recorded will be a product of the number of pixels used. This is why the concept of 'pixels on the bird' is an important one.
My 'Choosing a camera and lens' in Our Articles paper has a diagram of this in an appendix.
If you take a picture from the same location of the same scene with a 20Mp full frame camera/300mm lens vs a 20 Mp micro 4/3 camera with a 300mm lens, the micro 4/3 camera will place those 20Mp over a smaller part of the scene. If this is a bird, more of the Mp will be on the bird so when you view the image on your screen, there will be less background and more bird in the image. The bird might be recorded by (say) 10 of the available 20 Mp whereas on the full frame camera the bird would be recorded by fewer Mp. To this extent, the '35mm equivalence' concept holds true, and to get the same number of pixels placed on the bird you would need a 600mm lens. (I'm not trying to pick on micro 4/3 here - the same concept would apply with any crop sensor camera.) But the pixels on the micro 4/3 are much smaller (see below).
However, if you increased the number of Mp on the full frame camera, you will reach a point where both systems will place the same number of Mp on the bird. The full frame camera will record more background as well, but if you cropped the image appropriately you would have an image which is approximately the same as the micro4/3 image. In this instance, the simplistic concept of '300mm looks like 600mm' breaks down - both cameras are recording the image of the bird with the same number of pixels. Viewed on a computer at 100% both images would look (roughly) the same. Ian's tables set out mathematically how this works.
In other words, if you shot a scene with a Canon 5Ds, and an Olympus, both using a 300 mm lens, there would be no advantage of micro4/3 in terms of 'getting close to the bird'. The Table on the second page of Ian's article illustrates this. This is the critical point - it's 'pixels on the subject' not 'focal length equivalence' that determines what the final image looks like.
The situation becomes more complex when you throw noise performance and dynamic range into the equation. Noise performance depends on the surface area of the pixel - smaller pixels will always show more photon shot noise. Generally speaking, full frame cameras have larger pixels (a 20Mp full frame camera will have much larger pixels than a 20Mp APSC or micro 4/3 camera). You may not notice that much difference at ISO 100, but you definitely will notice at speeds like ISO 3200. So, even if the full frame camera brings fewer pixels to bear on a bird, they will be 'better' pixels in terms of noise and dynamic range. Noise, in particular, ruins details - so a noisy sensor may well show a lower level of fine detail at higher ISOs even if it has more megapixels than a less noisy one. The benefits of full frame increase as the 'going gets tougher', so to speak. (Note, I am not a full frame user.)
All of this shows that a simple reference to '35mm equivalence' has its limits, and concepts like pixel pitch and 'the number of pixels on the subject' give a more accurate representation of how different cameras will 'view' a scene with different lenses. In other words, the point Ian is making at the bottom of page 2, under the table. I don't think any of this is intended to denigrate any brand of camera - its all about compromises and what is important to the photographer (eg size and weight vs low light performance, cost effectiveness, plus other features such as focusing performance, stablisation, viewfinder quality etc).
Regards
Simon
