I've been asked a couple of times for my opinions on this device.
Alan Brown has said much of it already. As someone involved with reviewing displays, and also doing film cleanup and DVD/BD video encoding and authoring work, I agree with what he's already said.
First of all: I haven't seen one in the flesh. What I'm discussing is the principle behind blanket "enhancement" in a video chain, and also my thoughts on the example images on the Darbeevision site.
The images that have been processed by the algorithm, as seen on the site gallery, are obviously distorted. In particular, the images of the butterfly and the bee don't "read" as well as they did before, to my eyes. The processing looks similar to an Unsharp Mask algorithm, and looks to be more strongly applied in the horizontal direction. And it appears to be frequency adaptive, so it'll leave out of focus areas alone. That's just my observations, I don't know the specifics of how it works.
I've been told though that the images on the site show the device running at full strength. Perhaps it'd be useful for adding gentle sharpening to SD sources?
Speaking purely in terms of HD:
The resolution of properly-done BD content doesn't need "enhancement". Resolution is one of the last things wrong with Blu-ray Disc. Right now I'm working on a 2K cleanup and color correction of a 1980s film, which was shot with fairly cheap lenses and equipment. Even it doesn't need this sort of selective sharpening (although myself and the director agreed that adding a bit of controlled linear high frequency sharpening was a good idea).
And frankly, the idea of people using a device like this on discs that have already been carefully mastered and signed off on, well - it makes me more than a little uncomfortable. What goes onto the disc, what the director approved, doesn't need "enhanced" with selective contrast manipulation.
For film transfers that we know are compromised, perhaps the device would have some use. But the idea of plugging it in to the chain and using it indiscriminately is scary and it goes against any ideas of accurate imaging.
My last point is, that if the algorithm is a good one, why is it being sold to consumers? By that, I mean, why not sell it to professionals (either in hardware or software form) who could use it in a studio environment where DoPs and colorists could use it - carefully - as an artistic tool? That way its usage could be controlled, and its strengths could be employed where necessary.