What about a physics simulation to back into those parameters? Given a front and side image without wearing a bra are there really so many different archetypes that can produce the resulting shape?
- Tall vs. short vs. average height breast root
- Wide vs. narrow vs. average width breast root
- Breast root separation (also wide, narrow, average)
- Tissue density and malleability (some women have tissue that can work with a variety of bras whereas some [like me] have tissue that is pretty dense and will resist any shaping efforts by the bra)
- Tissue distribution/Tissue fullness (upper fullness vs. lower fullness for a total of four quadrants): Full upper and shallow bottom [very rare outside of situations like lumpectomies], shallow upper and shallow bottom, shallow upper and fuller bottom, full upper and full bottom.
- Ribcage shape (some women have barrel shaped ribcages or pectus excavatum plus those who have less padding/are on the smaller band size range and therefore no/less fat to smooth out the bumps can have this cause fit issues)
The primary difficulties are the UX/end user complexity problem AND the fact that those different archetypes can produce the same visual or physical results.
For instance, being able to tell if a cup is too big or too tall or whether a cup is too closed on top or too small can be really difficult even for humans. Does the gore not tack because the breasts are too close-set or because the cup is too small? Etc.
And yeah, images not in a bra would be useless because there are literally physics forces acting on the tissue when it's in a bra versus not that change things. For example, pendulous breasts would not scan as having any upper fullness at all, but when the root is supported, upper fullness is possible and therefore needs to be considered in fit. There's information that's necessary to a proper fit which literally cannot be gathered from a nude look/scan.
From a UX/customer service perspective, most women don't understand the sizing chart. Adding extra variables isn't going to work well, especially without hard and fast rules to follow. Even asking women to wear a well fitting bra at home won't solve basic user error problems such as women forgetting to scoop and swoop all their tissue in.
So a physics simulation would need to include the shape and density distribution of the breasts. Hypothetically, it would then be able to simulate dozens of bras with reference sizes and how this would look (i.e. an image output). At this point an expert system trained on images of good vs bad fit could probably visually recognize common problems and return a few of the best bras. This is probably oversimplifying, and it's already too complicated to actually be done, so I think it's not the right approach.
Do you see any aspect of bra fitting which _would_ benefit from technology? i.e. some kind of specialized fitting tool
Bra production absolutely. There are certain markets of women who are almost completely neglected when it comes to well fitting bras. Large band small bust (40+ A-D) are completely screwed by the good bra brands. Same for women who are under a 28 or 30 band. The economics just don't make sense for the brands to produce those sizes. Also bras for elderly women or the disabled that have more hooks, rely less on tension, etc. and basically accommodate for mobility or sight limitations.
Where I think tech could be very helpful is marrying JIT manufacturing with CNC fabric cutting to produce single bras or small batches for this economic 'long tail'. Think 1 hour glasses stores where you could go in, get fit, and just have one made if you happen to be one of the neglected demographics.