They're all pretty decent at this size, with the 5D looking the noisiest. What look like sensitivity differences aren't really, since I found that the highlights stayed reasonably the same as below. Everything was shot manually, with the same Nikon manual lens on every camera.

Now let's zoom in for the hackers.

These are below are equal-sized crops from the middle of much larger images. If each complete uncropped image was printed at this same high magnification, each would be 40 x 60" (100 x 150cm). You can click each image to get to each camera's full review.

Analysis      top

I don't see much difference at ISO 3,200. The worst is the Canon 5D, but not by much.

I don't see any significant noise difference at ISO 3,200 unless I'm looking too closely.

If you look at the blowups, the D3X is sharpest, but that's because it is the sharpest.

The 5D Mark II looks the smoothest, but that's because it slops on the noise reduction with reckless abandon. Notice how the noise is smoother, but so are the fine details!

Because of all the noise reduction of the 5D Mark II, the Nikon D3, with only half as many pixels, looks much sharper than the 5D Mark II at ISO 3,200 because it is sharper. Noise reduction isn't free: it costs you sharpness in exchange for nice smooth backgrounds.

The D3 is as clean and sharper than the 5D Mark II, and since the D700 performs the same as the D3 at ISO 3,200, that means the D700 is as clean but much sharper than the 5D Mark II at ISO 3,200.

Technik      top

These are crops from the center of each image. The full image at this magnification would print 60 inches (1.5 meters) wide.

Why this self-portrait? This is actually a very sneaky test: the fur on my face and eyes lets us see texture to look for the horrid effects of smudging from noise reduction, and the out-of focus wash behind my head lets us see noise at various levels on flat parts of the image. The out-of-focus transition behind my head is very deliberate.

I used the 105mm f/4 Micro-NIKKOR AI from 1980 because it's an ultra sharp lens, especially at the 1:8 reproduction ratio used here, and because it doesn't use floating elements or zoom. Its optical core is locked down solid, so nothing changes as put on different cameras with kludgy adapters. Newer macro lenses and zooms would all change their optical properties with floating elements as the flange focal distance varied slightly with the silly Nikon -> Canon adapter I used.

I left NR and sharpening at defaults. These are the actual images (JPGs) created by each camera. If you prefer to shoot raw, your results will vary all over the map depending on which software you use to anneal the raw files into actual images.

The 5D Mark II was running firmware version 1.0.7.

PLUG

I support my growing family through this website.

If you find this as helpful as a book you might have had to buy or a workshop you may have had to take, feel free to help me continue helping everyone.

If you've gotten your gear through one of my links or helped otherwise, you're family. It's great people like you who allow me to keep adding to this site full-time. Thanks!

If you haven't helped yet, please do, and consider helping me with a gift of $5.00.

The biggest help is to use these links to Adorama, Amazon, B&H, Ritz, and J&R when you get your goodies. It costs you nothing and is a huge help to me. These places have the best prices and service, which is why I've used them since before this website existed. I recommend them all personally.

Thanks for reading!