Photo samples: Canon EOS 40D

This animated GIF simulates the burst speed of the 40D. (As it's in GIF format, it in no way represents the image quality, and it runs at about seven frames per second, slightly faster than the 40D's tested 6.3fps.) To start the loop again, hit Shift-Refresh in your browser.

Preliminary tests show that there's practically no visible noise in 40D photos shot at ISO 400.

The 40D does a very good job of preserving both highlight and shadow detail. For instance, here you can see the detail in the dog's white fur as well as make out the pupil in the dark, shadowed eyes.

In casual testing, the EOS 40D's photos display little noise in shots at ISO 1600. Note that this image has some blur--I shot it handheld at 1/8 second--which can reduce apparent noise.

Underexposed shots, like the one on the left, seem to have plenty of latitude for enhancement. This was shot at ISO 800 and given an exposure boost in software (please pardon the quick-and-dirty white balance correction). As you can see, some noise appears in the shadow areas, but overall it remains under control.

The EOS 40D delivers very nice color saturation and, in casual analysis, good automatic white balance.

(Note: this visual is a very rough approximation of relative sizes of spot-meter spots for the purpose of illustrating my point. And, in fact, the spots are circles, not rectangles. But I couldn't spend all day on this.)
Canon is infamous for its overlarge spot-meter circle sizes, and the 40D's has increased over the 30D--up to 3.8 percent of the viewfinder, the same as the 1D Mark III. In contrast, the Nikon D80 uses a 2.5 percent spot size.
The spot circle size matters for photos like this, where the area you want to meter off is small and proximate to an area that would deliver a different result. I was trying to meter off the bird's head, but think the metering circle extended to its more brightly lit shoulder, which caused the underexposure.

The 40D's sharpness setting defaults to None, and we use the default settings for our lab tests--hence, the apparently soft samples you see here. However, you can still make out the relative differences across the high ISO shooting noise--or, more to the point, how little degradation there is.