Well, I've been comparing full-frame sensors against cropped sensors in this thread so why not go the extra mile and look at how crazy the cropped sensors in Compact cameras are? Sorry guys and gals - I love you really.
As a typical example, take the Panasonic Lumix DMC-TZ5, which Gordon reviewed here
. From the specification page
we see that it can record, in 3:2 aspect ratio, 3552x2368 pixels and that its lens has a focal distance range of 4.7-47mm (35mm Equiv.: 28-280mm). This gives a crop factor of about 6x (280÷47) so implying that the height of the sensor is about 4mm (the 24mm height of a full-frame sensor divided by 6). As a gross error check my post here
shows we are in the right ballpark for a "1/2.33 inch" sensor. Remembering that we have a vertical pixel count of 2368 then each of those pixels must be about 1.7μm high.
Now 1.7μm equals 1700nm (nm stands for nanometres, of which 1,000 fit into one μm and 1,000,000 fit into one millimetre). "Why do I care?" I hear you yawn. Well, the light we capture with our cameras has wavelengths ranging from roughly 475nm at the blue end up to 650nm at the red end. It's bad enough, with this type of sensor, that to avoid image softening from diffraction the lens needs to work at roughly f/4 or faster (see this page
for more on that) but it seems that the lens has to focus each photon onto an area only three times wider than the wavelengths typical in a sunset.
Frankly, I'm absolutely gob-smacked that the things work at all, though there is plenty of evidence in various images posted here on the forum that they do. It certainly seems as though the pixel count can only be significantly increased from this point on by increasing the size of the sensor, not just from the well worn arguments about high ISO picture noise, but also because we already have pixel sizes not much bigger than the wavelength of the light they are meant to capture. Forget micro-lenses - they'll have to fit waveguides next.