Originally posted by Alan 2 if a photo taken at F13 with a 28 MP FF camera looks no better (and possible worse) then a photo taken at F9 with a 12 MP 1.5x camera there may be some very disappointed landscape photographers.
[...]
Interestingly, as far as I know film doesn’t have a diffraction limit. The only limit is the diffraction limit of the lens.
Alan, you are jumping to conclusion here and miss the point.
The limits we have been talking about apply to perfect optics which is diffraction-limited at all apertures. However, such glass isn't currently a product.
Current good glass typically has its sweet spot at f/4 and isn't diffraction-limited there. And doesn't become so if you stop down to f/5.6 for DoF on FF.
And of course, it is
always the optical lens which is diffraction-limited. Neither film nor sensors are.
Originally posted by jeffkrol Jeff, the post is is nearly impossible to understand
The key is: you must compare the small image in the left hand corner to the large image on the right side (as those are the corresponding 100% crops).
The test could be useful because it compares the same amount of light falling onto the same size sensor surface (hosting a different numbers of pixels).
However, it isn't: The boost of exposure by a factor 135 means that the entire image was contained in 8-Bit RGB values 0 - 2. So, what we see, is a demonstration of different read-out noise (which is higher with the P&S but more than compensated by their higher number).
So, the lesson is that smaller pixels may help with the read-out noise. As will many other measures as well. Interesting for engineers. But not us.