The D3X wins. Not only is it sharpest, it's the most natural, even shot as JPG.
The Canon 5D Mark II looks crappy by comparison as direct-from-camera JPGs. The 5D Mark II applies too much oversharpening and too much noise reduction to remove subtle textures.
The wild part is that if I go through the trouble of shooting CR2 with the 5D Mark II and open it in Lightroom 2.2 that it looks almost identical to the JPG direct from the D3X!
As processed for the web, but not visible to you as you view this page, is that each of the crop files as shown on this page varies in file size depending on its contrast and sharpness. What's freaky is that the D3X version and the 5D Mark II processed CR2 version not only look alike, their file sizes on this page are also the same: 373kB. The other files vary from 207kB (SD700) to 571kB (Leica M7).
The Canon 5D Mark II comes close, but as expected from other Canons, its image direct-from-camera looks more artificial with more artifacts of digital processing. The Canon image has more obvious edge sharpening, and it also smudges-out subtler textures. The Nikon D3X image just looks like an image.
If you're not seeing this, first look at the hard edges of the big white tree. Both the D3X and 5D Mark II look about the same, with the hard edges being even more pronounced on the 5D Mark II. Now look at the wispy-thin branches on the gray tree on the right. See how these softer gray lines seem to evaporate in the 5D Mark II's noise reduction? If you can't see this, then just look at the eucalyptus leaves behind the trees. You'll see that even though the stark white branches are pronounced in the 5D Mark II images, that the softer leaves are less pronounced, This weird differential effect is what's so bad about the 5D Mark II: different details are presented different, as they are in electronic video images. Yuck.
Since the Canon 5D Mark II has a design flaw where it applies too much noise reduction, even at ISO 100 as shot here, I gave it a "Special Olympics" style do-over and allowed it to try again with its noise reduction disabled. This helps, but the image still doesn't look as natural to my eyes as the one from the D3X.
The 5D Mark II looks harsher because its contrast is slightly higher, and it's adding more sharpening to edges while not sharpening the subtler textures equally.
Look in the shadows and look for subtle texture. The 5D Mark II is burying these with noise reduction. This is just one of the many reasons I love film: you get what you shot, not some anonymous firmware programmer's chewed-up version of your image.
The Canon looked worse when I tried it under softer light. This is because the Canon smoothes subtler textures. These examples under very harsh light show the Canon at its best because there aren't many subtle textures. For those of you who shoot under softer light as I showed earlier, the defects in the Canon firmware become even more egregious. Be careful.
Even the old Canon 5D doesn't look that bad at these crazy levels of enlargement. The old 5D certainly looks better than Nikon's D3, as I've shown before in less detail.
If you wanted to screw around with raw files, the Canon 5D Mark II looks just like the JPGs form the D3X. (I have no software which can open the D3X NEF files to compare.)
To show you the insane level of enlargement I've used, I've also included examples from a 6MP Nikon D40 SLR and 6MP Canon SD700 pocket camera. The D40 looks great printed at 20x30" (50x75cm), and looks horrible blown up twice as large here. Will you ever make 40 x 60" (100 x 150cm) prints, and if you do, would you ever look at them this close?
I doubt it. That's why 10 MP cameras are more than enough for all the guys I know who earn their livings selling images. Stand as far away from your computer screen as you would from a huge 40 x 60" (100 x 150cm) print. All the images look OK now, don't they?
If you're a hobbyist, and if you think a scrawny 13x19" (30x45cm) print is big, then even a 3MP camera looks great. These results will only be as significant as shown on this page if you blow them up 5 feet (1.5 meters) wide.
Bonus: I discovered some film tests shot the day before, and added them for fun. The colors are bolder because I shot bold film, not the dull defaults of the digital cameras. The film isn't being compared at its best here, because you're not seeing the film. All you're seeing are relatively crappy minilab scans of the film at too low a resolution.
When I put the film under my microscope (a Nikon Alphaphot YS stereo lab scope at 40x) so I could see what was really on the film, the Leica M7 with a 39-year old lens is about the same as the Nikon D3X.
What's intriguing about the two film scans is how similar a great 35mm lens, like the Leica, is compared to a mediocre medium format lens, the Pentax 75mm f/2.8, shot at a sub-optimal aperture. They're each about as sharp as scanned at the minilab at different resolutions. The biggest difference is the cleaner grain of the larger format.
These are crops from the center of each image. The full image at this magnification would print 60 inches (1.5 meters) wide.
Each camera was at its defaults, although I cheated and re-ran the 5D Mark II with its noise reduction turned off. I also usually shoot my Nikon D3 with its sharpening set a couple of notches higher than default, but left it turned down to its softer default for comparison here.
I use specially grown reference trees at a regional arboretum (tree museum) instead of man-made objects like star targets, classified ads or buildings precisely because an educated eye can see far more with natural targets. Trees have a fractal nature, meaning that there is detail at every level, every resolution, every amplitude and in every direction. Artificial target edges only have detail at odd-harmonic series, at discrete amplitudes and only in some directions. Test targets miss a lot of things that become obvious with appropriate natural targets. There was no wind and no heat shimmer.
You'd never see the way the 5D Mark II smudges subtle textures if all you shot were hard black-and-white printed resolution test targets.
Everything else like NR and sharpening was left at each camera's defaults.
You could go blind trying every possible combination of settings.
I cheated by letting the 5D Mark II have a second try with its high-ISO NR turned off.
The D3X was at firmware version 1.00/1.00
It was shot at ISO 100, 1/160 at f/8, Nikon 50mm f/1.4 AF-S, Sunny WB, Standard Picture Control.
5D Mark II
The 5D Mark II was at firmware v. 1.0.7.
ISO 100, 1/160 at f/8, Canon 50mm f/1.4 USM, Sunny WB, default Picture Style (3, 0, 0, 0).
For the CR2 file, I had to open it in Lightroom 2.2 and export it at the same image size as the D3X. The result was the same as resizing in Photoshop. I made no tweaks in Lightroom.
ISO 100, f/8 at 1/160, Canon 50mm f/1.4 USM, Sunny WB, default Picture Style (3, 0, 0, 0).
The D3 was at firmware 2.00/2.00.
ISO 200, 1/250 at f/8, Nikon 50mm f/1.4 AF-S, sunny WB.
ISO 200, 1/500 at f/5.6, 18-55mm kit lens at 31mm to give the same angle of view on its smaller DX sensor.
ISO 80, f/3.2 at 1/640, zoom set to 8.6mm to give same angle of view on its sub-pea sized sensor.
Each image came directly from each camera. I did not use any external raw processing software, as each and every piece of software (Adobe Camera Raw, Phase One, Capture NX, etc.) uses different processing for different cameras.
If I shot raw, we'd be comparing the variations in how any given piece of software processed images from different cameras instead of getting a clear view of what each camera actually does when processing an entire image to completion as a JPG.
In-camera JPG processing is also usually more advanced than third-party tools, as designers behind one of the major brands of raw software have told me in person. For instance, while Nikon automatically corrects lateral color fringes in-camera, that company's raw software didn't. Nikon's in-camera image correction and ADR is smarter than most raw software.
If I wanted to twiddle with raw software, I undoubtedly would get different results.
The Nikon looks great shot as JPG. The Canon looks worse, and might benefit greatly if you wanted to try raw. I have better things to do, like shoot.
Does anyone who shoots really care? I know many people sit bored at their office jobs researching purchases they'll never make, but is anyone really wondering whether to buy this Canon versus this Nikon, realizing that along with it you have to buy all the lenses, batteries and everything that goes along with a system? Are you just curious (as I am, too), or do you have a real need to know?
Another huge benefit of the D3X is handling. It just goes, it feels great, and great pictures come out. The ergonomics of the Canon are awful. It hurts to use since the buttons are made of cheap materials and put in the wrong places, as I explained in its review. The D3X is a serious pro camera, while the 5D Mark II is just another consumer electronics product. It's not like the two cameras are identical in all other aspects, in fact, the only thing they share in common is pixel count and sensor size.
The D3X easily wins, but I wouldn't buy one unless you have money to earn with it right now.
The D3X price will undoubtedly come down as a result of the market's boycott, and even if you do break the boycott and buy one, when the D700x comes out, you're going to want one instead for its better handing, design, size and weight.
You people know who you are, but almost no one needs this many pixels. My pals who earn their livings through photography know that 10MP is more than enough for almost any paying job they get, and they all shoot JPG, too.
If you've got eight grand today and don't mind the stigma of breaking the boycott and screwing all your buddies out of the price drop we're earning for you, go right ahead.
See also Is It Worth It.
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!