They absolutely do! If you have watched even a single episode of What Not to Wear you'd have known that most women, especially ones with breast sizes a couple of sigmas away from the mean, have no easy way to buy a well-fitted bra. They either have to go to a fancy store to do personal fitting (may be expensive) or go to a place like VS and hope for the best.
A suitable algorithm, based on Kinect 3D models, say, would totally take this market by surprise and make a huge splash.
Well, maybe. The "wrong size" problem is only part of the story. Another is that many bras are generally poorly constructed, lose their shape after continued use, and sizing is not the only issue (fabric tolerance, flop etc).
But more than anything else, the problem is price. A "good" workhorse bra (workwear, not lingerie fantasy stuff) will set you back on average, oh, say about $65 ($100 is not uncommon). You probably need half a dozen. I have no idea what the economics are and why they are so expensive, but I can't imagine what price a "Kinect 3D" custom bra would come in at.
And the other aspect of "a good bra is hard to find" is that once you find one, there's a huge lock-in factor. Women will re-order the same model for years.
So yeah, while there is certainly a non-zero market for something like this, it's probably not huge. Witness the fact that it is quite easy and even not too pricey to get custom jeans made to measure, and yet the majority of the population doesn't.
Now there is a disruption possibility in that market if someone took the time to do actual biometric and customer search and some innovation in production methods.
I had discussed a Kinect idea with some fellow employees when I worked at VS and they scoffed at the idea.
I think the issue is that breasts are more complex than just external shape with internal density and natural placement being just the tip of the iceberg. All the factors involved rule out a cheap modeling method such as Kinect or webcams and in the end, make visiting a store, to be measured properly and try for fit, the more economic choice.
I don't think you can get by the trust issue, but a redboxish booth to be placed at stores might work. I know I would really like to order jeans / pants customized to fit. Not sure if I wound stand there in a booth in my skivvies to get scanned. Imagine being topless for a girl would require a fair bit of trust.
A suitable algorithm, based on Kinect 3D models, say, would totally take this market by surprise and make a huge splash.