The Processing
In previous H2H shootouts we’ve done, we compared the in-camera JPEGs with files processed by the manufacturer's software and Adobe Camera RAW. In this article, we can’t compare the JPEGs, since the Hasselblad doesn’t shoot in JPEG mode, but we can compare the manufacturer’s software -- Nikon NX2, Canon DPP, and Hasselblad Phocus 2 -- with Adobe Camera RAW, since its supports RAW files from all three cameras. We processed the files at default settings. We could have, as some have noted, used many devices and techniques to get the best from the files, but that’s more a test of the guy doing the processing than the manufacturer’s files. The point here is that we’re better off if we get a better start. If we can push the file through the system with minimal effort and get close to what we want, that saves us time and effort. Plus, most of the time, it gives us ultimately a better-looking file.
Manufacturer's Software
Let’s start with the files from the manufacturers' software. Here are the figure shots with strobes. Left to right: the Canon 5D Mark II, the Nikon D3s, and the Hasselblad H3DII-31:
 |
 |
 |
Canon 5D Mark II - DPP
|
Nikon D3s - NX2
|
Hasselblad H3DII-31 - Phocus 2
|
This is what we’re going to see in the coming samples as a trend. The Canon files have a slight cast, leaning towards a blue-magenta. As such, the files feel less full, since we’re losing a lot of the yellow and orange tones that give a shot like this a lot of richness. The Nikon files are better “centered”, and have a nice, rich, but accurate feel. Although the skin tones are full and warm, the background holds a neutral, even a bit of a cool tone. The Hasselblad files processed in Phocus 2 are looking almost like a consumer camera. They’re a bit warm, and the skin tones are flattering. The reds pop, and all this is pointing to the Hasselblad “Natural Color Solution (HNCS) for outstanding and faithful colors”. In the past, with software like Flexframe, accuracy was paramount. It seems that Hasselblad is leaning towards “pretty” more than “accurate” now, trying to offer a file that satisfies portrait, wedding and fashion photographers more than prepress guys. It’s an interesting decision, and one that high-quality audio makers face: do you want to make it pretty, or correct? Big, booming base sells speakers, snappy colors sell cameras.
Here are the three shots zoomed to 100%. Keep in mind a few caveats: on the DSLR cameras we’re using high-quality zooms, the Hasselblad has the 80mm prime. The three cameras have very different pixel pitch, so the magnification at 100% for a similar frame and crop is going to be different. Finally, these are without any final output (Unsharp Masking) sharpening.
Again, Canon, Nikon and Hasselblad:
 |
| Canon 5D Mark II - Digital Photo Professional (unsharpened) |
 |
| Nikon D3s - Capture NX2 (unsharpened) |
 |
Hasselblad H3DII-31 - Phocus 2 (unsharpened)
|
They all look fairly similar, and they all could stand some application of Unsharp Mask. Before we look at that, though, take a look at the zipper in the Hasselblad shot. See that purple fringing? That, my friends, is aliasing, and it’s the downside of big pixels and no anti-aliasing filters. You get a sharper file, in most cases, but you’re going to battle aliasing. Here’s a closeup of what that looks like:
 |
| Purple fringing apparent in the Hasselblad H3DII-31 image |
Here’s some copy related to aliasing from Hasselblad's Phocus 2 product pages:
“Leading Edge Moiré Removal Technology... With Phocus, the moiré that can occur on even extremely high resolution images is effectively removed from your images. This is automatically performed directly on the raw data, leaving image quality intact and eliminating the need to carry out special masking selections or other manual procedures, saving hours of tedious post-production work.”
Apparently it can’t quite stand up to a standard DSLR with an AA filter.
Here are the same three files with an equal amount of USM applied. Again, these images are viewed at 100%.
 |
| Canon 5D Mark II - Digital Photo Professional (USM applied) |
 |
| Nikon D3s - Capture NX2 (USM applied) |
 |
| Hasselblad H3DII-31 - Phocus 2 (USM applied) |
Of the three, the last Hasselblad file responds best, sharpening not only the hair, but the skin and eye details. The 5DM2, curiously, responds the least. Where we’d expect to see sharpening around the wisps of hair at the very least, we’re getting a minimal effect. The Nikon file sharpens softly overall, and most pronounced in the highlights of the jewelry. In the Hasselblad file you can also see the negative effects of sharpening on the aliasing problem. It gets pretty ugly fast.
Most of this is due to the significantly larger sensor and significantly larger pixels, and the absence of an anti-aliasing filter in the Medium Format Digital Back (MFDB) file. Small pixels packed closely together with a filter that, essentially, blurs and blends highlights to eliminate single-pixel point illumination results in soft edges. Soft edges don’t sharpen.
Adobe Camera RAW
Here are the files processed in Adobe Camera RAW.
 |
 |
 |
Canon 5D Mark II - ACR
|
Nikon D3s - ACR
|
Hasselblad H3DII-31 - ACR
|
The Canon file still processes out as slightly less full, almost “garish”. The Nikon strikes the best balance, it seems to hit the middle point here as well as in it’s own software. The Hasselblad is clearly not happy in Adobe Camera RAW, as you can see by the oddly bright and magenta-hued reds of the dress. Hasselblad uses “secret sauce” data both from the sensor as well as the lens to process the files, and clearly, Adobe Camera RAW isn’t getting the complete story.
We’re getting a similar result with all three, although they all seem a slight bit crisper. The aliasing in the zipper is slightly less pronounced, by the way, which brings up an interesting point. All RAW files have aliasing, to some extent, and an important part of the processing is to blend and remove it. Technically the term is moiré, and there are several ways it manifests, but all of it is a result of the basic grid and color pattern of the pixels on the sensor. Adobe seems to handle Hasselblad’s aliasing better than Hasselblad does.
Now, let’s look at the files with Unsharp Mask applied.
 |
| Canon 5D Mark II - Adobe Camera RAW (USM applied) |
 |
| Nikon D3s - Adobe Camera RAW (USM applied) |
 |
| Hasselblad H3DII-31 - Adobe Camera RAW (USM applied) |
The Hasselblad file is still clearly the winner, but the two DSLR files look a whole lot better than they did in their own software. This really highlights the difference between the sharpening that is applied, actually, required, in the RAW processing compared with the application of final output sharpening, Unsharp Mask, once the image is sized and prepped for printing. You simply have to apply sharpening in the RAW process, and how you apply it dictates how Unsharp Mask will work on the file. Without a good “tooth”, as I like to describe it, USM has nothing to work with.
Before we move on, look again at the jewelry in the Hasselblad shot. Not bad aliasing at all, by Adobe.
The conclusions here are that the AA array on the sensor certainly has a profound effect on the final file, but also that the processing can make or break that. Take a heavily filtered file and process it poorly, as in the 5D Mark II with Canon’s own DPP, and you have a mushy file.