[magnusbaringer]

Thank you very much for your feedback and time! We know that we do not cover aspects like skin tone - and we would love to somehow implement that. The problem is that to get the best statistical results, the "decision-influencing" differences between images ideally have to be reduced to just the exact condition (e.g. different breast size). Every other additional variation (e.g. skin tone) would "blur" the data. As we need 88 picture pairs to cover all required breast conditions anyways, it is practically impossible to take any other aspects into the survey (e.g. skin tone) without completely exploding the number of pictures and blurring the data. So we apologize for that "inconvenience" - but we haven't found a better solution. Thank you very much for your support and feedback!

[waterhouse]

I wouldn't even call it white skin, actually. It's not really a flesh color; breaking out the Digital Color Meter, it's slightly warmer than a grey.

Then there's the way light reflects off of it, some of it looking rough or dirty in a way skin usually doesn't, with some parts brightly lit as if by one very bright source. (I think glare explains some of it.) The breasts themselves often look brighter, and sometimes cleaner, than everything else; occasionally they do match the surrounding skin, which naively looks a lot better until I realize I have to abstract out the lighting.

Even if you can't vary the skin tones, I think that using more uniform lighting would help. And maybe something dumb like spraypainting the main body a mono-color would work, although that might make it more complex to show scars.

[derbOac]

Another option is to have a 2-4 skin tones and just randomly select one for each respondent. It's introducing another variable but presumably you could collapse over it.

[shigawire]

Maybe that could be a selection criteria at the start? If you add 5 skin tones doesn't that only x5 the number of images?