4 ms·

> Arguably, even "modern" cameras are apparently not designed with noncaucasian skintones in mind, something interest groups do not tire to bring up. That does

• robotresearcher 4y ago
  Not so long ago, film processes and machines were optimized and calibrated for reference images of only white faces[1]. Darker-skinned people often found their skin tones were badly reproduced. The analogous thing happened again in digital processes, up to and including AI image processing trained on datasets where white skin is overrepresented. The industry's awareness of this issue is now pretty high. I don't know if the issue is solved in practice on current-gen cameras/phones/software or not. Anyone have latest info? [1] https://www.npr.org/2014/11/13/363517842/for-decades-kodak-s-shirley-cards-set-photography-s-skin-tone-standard https://www.npr.org/2014/11/13/363517842/for-decades-kodak-s...
  • Gordonjcp 4y ago
    That's interesting, I hadn't really thought of it that way. I wonder if that's why the first colour grading exercise in the training material for BMD DaVinci Resolve is an excerpt from a documentary about rhinos, where you need to balance a reasonably clean shot of a white woman, a gorgeous but fairly tricky shot of a black man (in profile, strong sunlight on his nose, eyes and cheek, rest in deep shadow), and a couple of mid and long shots of black gamekeepers walking around. There's a lot to get right and a lot to get wrong, and you can end up making everyone look a pretty sickly colour.