Low distortion Class D amplifiers? What to expect?

Are we just talking about the technical techniques or the practical implementation?
Because I find Bruno's nor Hypex practical approach very far from ideal.

I would rather see a nice small ASIC, like TI TPA chips or like IRS2092 and the like.
These solutions will give a much better board real estate, the magic is just in the feedback loop.
 
I'd venture to say that once THD+N, IMD and TIMD measurements—and I think you can trust Bryston's own specs—drop below 0.01%, the user's only recourse for comparing quality of reproduction has to focus on listening to the amp(s) perform in a given system.