http://www.ddisoftware.com/sd14-5d/Now take a look at the color wheels and resolution charts about 15-20% the way down the page. The three paragraphs below it describe what's happening far better than I can here.
Halfway down you see the pictures of the flowers side by side. Here's the excerpt from the paragraph below that I feel is important:
"Where the SD14 holds consistent sharpness across the frame, the 5D has smudged over a bit of detail in areas notoriously problematic for Bayer sensors such as the red carnation and even the white flower where edge detail is being lost to the AA filter. Looking at the 5D shot, you'd be tempted to believe that the red flower is just a little out of focus because it's in front of (or behind) the other flower due to it not looking as sharp. In reality, all the flowers in the above crop are in the same plane relative to the lens. "
Parts of the flowers are in in focus and parts aren't, yet the focus of the lens isn't a factor - the way the sensor, AA filter, and camera are processing and blurring the data is what's causing this effect. I find it to be very annoying, to be honest. Parts of the image look sharp and parts look blurry, despite being right next to each other.
I don't see any of the review sites testing for this sort of thing, though. It honestly bothers me far more than most of the other issues, but perhaps that's because I shoot mostly black and white film where sharp and highly defined results are the norm. Most Bayer sensors after being put through a consumer grade lens and an overly aggressive AA filter look almost like someone hit the blur feature in Photoshop on purpose on bits and parts of the image.
It's why I like the look of the Sigma cameras a bit more - it looks more like film in how it's consistent from one place to another. Of course, it's also like dropping Velvia 50 in a half-frame camera. Beautiful *little* pictures. Sigh.