Here is my test scenario:
* 5D Mark III and the 5D Mark II side by side on tripods
* Focal length set to 200mm (100-400 on the Mark III and 70-200 on the Mark II which could effect sharpness)
* Focused on the same object (a signed football helmet)
* Aperture of f5.6
* Set timer mode to 2 sec (to avoid any shake)
* Used the new Lexar 1000x cards (to test the speed of the UDMA7 card slot)
* Turned off the lights in my home office
* Shot in RAW and imported into Photoshop with no altering of the files
* Zoomed and cropped to show a close-up of a dark and light area
Room setup with both cameras side-by-side (you can not tell from this image, but the 5D Mark III LCD is much brighter and sharper than the LCD on the 5D Mark II)
(Original image taken with the 5D Mark III at ISO 100 - scaled to 33pct)
And here are my results. Note: You can click on any of the following images to see a much larger version.
At ISO 6400, we see significant noise in the Mark II image but the Mark III image still looks very useable.
Now we are taking the Mark III into an area where the Mark II shouldn't go. So, I kept the Mark II at ISO 6400 to show you the comparison of the previous maximum ISO to the new higher ISO of the Mark III. You could argue that these two images, even with very different ISO settings, are about equal in quality.
More tests to come!