I find that *some* images taken with the SD9/SD10 have more "presence" than the same image from a typical mosaic sensor. It's hard to describe; they just look more real. I think the issue that you run into sometimes with bayer sensors is that they do not record the same level of detail for all colors. I've been more than a little aggravated more than once with the bayer design when taking photos of bright red flowers that have dark details/veins in the petals. The green leaves around the flower petals often look sharper than the red petals and the veins in the red petals often look smudged over because green is captured at twice the resolution of red. This can make your main subject (the red petals) look out of focus while emphasizing the wrong details (green leaves). This effect *is* troublesome to me in shooting nature shots as the inconsistency in sharpness/resolution across the image may be hard to verbalize but it is very noticeable in some cases.

You'll hear arguments where people claim that even on deep/pure reds, there is still *some* information present in the green/blue channel but the truth of the matter is that the small amount of luminance detail still present in the other channels simply does not (cannot) contribute much to image detail. Bayer algorithms often smooth over details in colored areas more than B/W areas as a result. Here's an example:

http://www.ddisoftware.com/testpics/sd10-10d-2.jpg

Notice that the gray pinstripe pointed to by the "1" appears sharp in the bayer (10D) photo, even sharper than the SD10 photo. If image detail were truly consistent in the 10D photo, it would follow that the red/black detail inside the pinstripe area would also be sharper in the 10D photo. Notice, however, that the 10D has smudged over all of the red/black transitions and the SD10 shows much more detail for red/black. The 10D really has no choice. For "near pure" reds, the 10D sensor is only good for 1536 x 1024 resolution and its antialias filter drops it below even that level. Notice at "2" where the 10D has smudged over the vertical black space between the two red panels, making them look almost like they are connected. The SD10 shows the detail sharp, as it should be, even though the output image on the SD10 is only about half the resolution of the 10D image.

The above is what I like about the X3 sensor technology. So far, I'm having a bit of a difficult time getting consistent color from the SD9/SD10 however. Color response appears to be more non-linear and may suffer from reciprocity failure due to the way the sensor is designed. I believe that this issue can be fully addressed in software but so far, I don't believe the Sigma software is there yet. I'm getting a little better results after developing my ICC profiles but I'm still having trouble with inconsistent color, like diimly lit greens turning a bluish gray.

I'm trying to give the SD10 a chance and I am warming up to it slowly, but I can still drop the 10D on the floor and have it accidentally take a shot that looks better (color wise) than a deliberately composed shot from the SD10. Since I'm getting better at it (the SD10), it appears that I may be able to eventually get better shots with the SD10 once I learn all its idiosyncracies(sp). So for now... I march on. More later...

--
Mike
http://www.ddisoftware.com