Hacker News new | ask | show | jobs
by midasuni 1097 days ago
A 32 bit float and only represent 2^32 numbers (maybe less), same as a 32 bit int.

What’s happening here is using a short (16 bit) rather than a byte (8 bit), although not quite to the same extent, for each channel. HDR allows you to represent 2^6/12/18 times as many colors as SDR.

1 comments

HDR10 is 10-bits per component and there are plenty of 16-bit images that aren't HDR, for example: http://i.imgur.com/Wm2kSxd.png -- the main thing about using HDR is to increase the range of values, i.e. what the biggest and smallest values are, not merely how many distinct values you can represent.
Well that’s just the colour space, you’ve also got the gamma curve to consider when you look at the perception of how something looks