In my comparison of the D3x to medium format, I made clear that I was using a Super Coolscan 9000 to scan the Velvia (I was also using Velvia 100, not 50). A megabuck drum scanner might extract more detail from the film (although the Super Coolscan is already scanning grain), and would certainly bring the film closer to the D3x's dynamic range. Remember that a drum scanner costs substantially more than a D3x, and is a royal pain to use (often run by an operator who has received manufacturer training). Of course, you could have someone else make your drum scans, at 50-100 dollars per scan. With comparable equipment investments, I stand by my statement that the D3x is roughly equivalent to 6x9 cm Velvia 100 scanned on a Super Coolscan 9000. A D3x, a Nikkor 24-70 f2.8 and a couple of flash cards cost about $9000.
I was going to stay out of this fight. But there's just too much misinformation about drum scanning here. I do a lot of drum scanning for people, I hope I've got a clue about my market.
First, anyone who would pay $9000 USD for a drum scanner these days is a nut case; please introduce me. I would gladly
part with one of my scanners for $9000 USD. There are excellent
used scanners on the market, full turn-key systems including mounting stations, computers, and software for way less money. As in less than $2000 USD. Used working Howtek 4000/4500s routinely go in this range. I've seen better scanners like Optronics ColorGetters go for less than $1000 USD. I'd explain why but I doubt anyone actually cares.
The only people who say that drum scanners are "a royal pain to use" are people who have never run one. Drum scanning has it's own set of learning curves, just like both film and digital photography have theirs. It's not insurmountably difficult. It does not require factory training. If I can do it anyone can. What it takes is care, precision, and patience. Anyone who's tried to get the most out of their cameras knows about this; it's not unlike doing the tripod, mirror lockup, precision focusing, picking the right taking aperture to maximize sharpness level of care, precision, and patience that most of the participants here know all too well.
While I'm here I might as well dispel another myth about drum scanners that's bound to come up if it hasn't already. That myth is about how wonderful a Tango is. Heidelberg designed the Tangos for advertising work and optimized the hell out of them for big pre-press houses. At this they excel. But unless you only shoot trannies and expect they'll never be printed larger than a magazine page using an offset press, you'll likely do better with another scanner.
There are two main problems that keep the Tangos out of the running for excellent fine art scanners. First, they have large minimum apertures (10 or 11 micros depending on who you talk to). This limits the optical resolution you can obtain with it. The theoretical maximum you'll get from about 10 microns is around 2500 spi. That's all. The "lowly" Howtek 4000 has a smaller aperture (around 6 microns, or around 4000 spi) and is noticeably sharper than a Tango. The Howtek 8000 and Premiers have a minimum aperture of around 3 microns, as do the later model ICGs, all of which can give you a real optical resolution in the range of 7000 spi, and crush Tangos for real optical resolution. Not that you'll produce a lot of film that has that much image information on it.
The second problem is that Heidelberg optimized the hell out of their scanners for tranny work. Operators have told me that working with negatives (color or B&W) is a royal PITA because the Linocolor/Newcolor software makes working with negatives operator-hostile. Heidelberg did this for good reason -- their market was nearly 100% trannies. Think art directors. Think time-is-money, and WYSIWYG, art director making decisions on the light table. Has to be trannies.
I'm not saying that Tangos are bad; they are excellent at what they were designed for. But there are considerably better scanners for fine art work. Especially negatives. So if you are comparing digital capture with Tango scans, don't be surprised when the digital capture comes up with more sharpness. If you want to know what film will really do, go up against an Aztek Premier with DPL software. That's probably the current state of the art in scanning any film. And a Premier almost certainly will
cost you more than $9000 USD.