and hd tv's that boast way more colours than 24bits 16777216 ... come now can the human eye really tell the difference between $FFFFFF and $FFFFFE ???
With TV's boasting crazy figures like 10,000:1 contrast ratios (or more, OLED displays are 1,000,000:1!) then extending 256 shades of any colour between that kind of range could potentially give rise to visible banding. 16 bits per pixel solves that problem. 16 bits per pixel also means that when the TV does post-processing on the picture (even just altering brightness or contrast) it doesn't 'squash' the colour space so much that you could perceive it. Same methodology applies to sound samples. 32 bit sound samples can be processed, filtered, mixed etc without causing a loss in quality when they are exported to 16-bit files.
ok i could look it up but im not that bothered tbh ;)
i was going to make a 24bit to 12bit (amiga ham 4096) jpeg converter to see what the diference would be
but in uncompressed bmp im still using the same ammount of video ram so i thought 'why bother'
(mind you a pure white .bmp would be the same as a multi coloured picture so my initial argument was moot its the compression that matters)
12-bit total gives 4 bits per pixel, or 16 distinct shades for R, G and B. It's pretty obvious that you could tell the difference between each shade ;)
edit:
i have started making a .pal file for psp that has the first line of 16 as $000000 the next as $111111 etc all the way up to $FFFFFF (if pal files gave hex values it would be so much quicker now i have to find out what AA is in dec etc)
then im going to make a 'c64ish' 64 colour pallet too
might post some random pix with each option done
afaik all i have to do is split to rgb creating a grey scale
load in new pallet 3 times and merge and bobs your uncle a total waste of my time
Colour conversions can be quite complex, involving octrees and 4D transforms. Best not to bother ;)