To me, the BMD scanner is just a show-piece to have on the wall of the colourist suite, I don't think it will do very well to be honest.
The Lasergraphics Scanstation Personal (http://www.lasergraphics.com/scanstation-personal-features.html) gives much better results and is in the same ball-park pricewise and has many, far more useful features as well as better quality output.
There are also the Muller scanners in the 'lower' price bracket (i.e. the under $50K bracket), and moviestuff has a the sub $5000 market covered with their Retro-line for small format films (8mm to 16mm).
You can also buy a used Imagica XE for under twenty grand if you want 35mm scanning, and they are a truly excellent machine.
To me, there are three main parts of the market that are not currently addressed.
A Sub $1000 film scanner for small format
A sub $5000 scanner for 35mm
A realtime (e.g. 24fps) scanner in any format that is under say $5000, preferably under $2000
I think the dream of covering all formats on a single scanner will likely lead to failure, or compromises that make the system unwieldy. It may be better to come up with a large format, and small format model, that can share the same sensor (and maybe light source) in a quick-swap method, as the sensor and light source are really what dictate the quality, and where a lot of the cost is.
For anyone with a medium to large amount of film to scan, speed is really the key, plus realtim capture means you can also capture any audio at the same time, another big time-saver.
Many argue (that haven't had to do it) that slower capture speeds are fine.
That 2fps just means it takes 12 times longer to capture than 24fps, so what would have taken you 8 hours on the 24fps unit takes 96 hours on the slow unit - so what if time is plentiful for you?
Well, It isn't just the faster capture speed that saves time. In the above example you can get through the 8 hours of film in a single 'work' day. The 96 hours however means 12 work days, and if that is a 5 day week, then it is 2 and a half weeks now, vs one day turn around.
Plus, life gets in the way, so some days you might only get a few hours in, some not at all, so it ends up taking a month, or two months, and often it means you have to pack it all up and get it all out again between sessions, costing you more time.
In a single 5 day working week on the fast system you might get through what would take 4-6 months to get through on the 2fps system, with the extra delays that slower running tends to introduce. In 3 weeks you could get through what could end up taking a year on the 2fps system.
Even for those with a lot of time, when doing it day after day, you do end up putting off doing it, or worse, decide you can leave it unattended while you go do something else for just a few minutes, and that is when it will decide to eat the film, or jam, or spool out all over the floor...
In my experience too, the longer it takes, the more problems you get, the unexpected windows update that tries to download in the middle of the capture causing dropped frames, other glitches, they add up more when you are taking many hours to do task vs doing it in 1/10th or 1/5th the time.
Anyway, you get the idea, a realtime system is far more feasible in the real-world, and has big benefits even for those with small amounts of film to capture.
This is easily acheivable now with harvesting projectors for the task, but more difficult when building from scratch.
I don't think an intermittent system is feasible for 24fps on a budget with a scratch built machine, but a continous flash-scan system is relatively easy and could be done inexpensively.
Other quick points, resolution is great, but low noise and wide dynamic range are more important, the price needs to be well under $5K or there are other proven solutions on the market already, realtime is a big feature, and allows capture at the same time.