Re: 8 verses 16 bit

[Kennedy McEwen <rkm@xxxxxxxxxxxxxxxxxxxx>]

In article <OxYCoLLi7Ah8Ewii@kennedym.demon.co.uk>, Kennedy McEwen 
<rkm@kennedym.demon.co.uk> writes

>I asked yesterday but you haven't produced any info.

That could be because my initial response to your suggestion seems to 
have exceeded the list's size restrictions!  :-)

Here it is in two parts:

Part 1.
In article <Pine.GSO.4.44.0203022005350.28181-100000@e4e.oac.uci.edu>, 
David Chien <chiendh@uci.edu> writes
>Basically, the more bits you have, the more colors you can reproduce or
>use within your files.
>
>Because bits refer to the binary bits used in a computer, you can
>calculate the number of colors you have total under each system.
>
>eg. 8 bits = 2^8 (2 raised to the power of 8) = 256 shades per color
>R, G, B (red, green, blue, the three colors used in emmissive display
>systems to create all other colors).
>
>256*256*256 (rgb) = 16 million colors
>
>16 bits = 2^16 = 65536 shades per R, G, B color
>
>65536^3 (rgb) = 281 trillion colors
>
True, so far.
>
>Human color vision studies easily show that the human eye can easily
>see more colors than what the 8-bit color range can support.
>
Please present some references for such human vision studies.

Every study I have seen concludes that, under a fixed level of 
illumination (ie. background light level) human vision is unable to 
discern more than around 100-150 individual tones per colour, which is 
certainly within the 8-bit per colour range.  Whilst the eye certainly 
has a wider range of light accommodation, this simply raises or lowers 
the range of discernible tones - the eye's pupil achieves most of the 
accommodation by opening and closing through many stops.  Photographic 
reproduction, whether emissive or reflective, operates in the former 
region of a fixed background light level, not the latter.  You can 
achieve the additional range of tonal discrimination simply by turning 
the light up or down that you view the image in - within any individual 
level of illumination, 8 bits per colour is MORE than is necessary, and 
is only implemented because:
1.  It is convenient in a "digital" format.
2.  It provides SOME accommodation of monitors which are inadequately 
set up in the first place in terms of brightness, contrast and gamma.

>In fact, simply looking at just one color, R, G, or B, alone will
>make it easier for most people to understand why.
>
>If you only have a 8 bit color system, Red, let's say, can only
>have 256 shades total!
>
Which is more than anyone can discern.

Otto Schade's early study into photographic noise levels indicated that 
the human eye can only differentiate around 100 discrete shades of grey 
printed onto a page - any more than this is indistinguishable from 
continuous tone imagery.  Huang's work in 1962 again confirmed similar 
(though slightly larger) values - all for pure grey images.  Are you 
proposing that a stimulation  of a SINGLE colour is more discernible 
than all three?  I might accept that stimulation of the green channel 
might be almost as good as the white/grey/black tests (though the red 
and blue differential response of the eye are significantly less than 
the green), but won't accept that more is necessary without documented 
evidence.  So far I have not seen any documented evidence of this - or 
found any indication of it in my own experimentation.

>As you well know perhaps from working with a bit 64 color Crayola box,
>there are far more shades of red than 256.
>
A VERY bad analogy, David!!

Since when have Crayola produced crayons on a photometric scale?  Few, 
if any, of their reds are pure - they mix different primary (or 
secondary) colours to achieve the desired tone and hue, they are not 
just changing the intensity of the red dyestuff itself!

The fact that you can see more reds than a 64 colour box of crayons 
simply indicates that 64 levels is inadequate to represent ALL colours, 
not just primary red!  That is a completely different argument from the 
number of tones each primary colour must be represented by, which is 
what you are applying it to.  This isn't even on the same scale as an 
apples and oranges comparison!

Go into Photoshop, create a black canvas and select the paintbrush tool. 
Now double click on the foreground palette and set green and blue to 
zero in the palette, but choose any number for the red colour - say 200. 
Now do the same on the background, but set a number on the red colour 
that differs only by one, say 199.  Now draw some foreground on your 
image with the paintbrush tool, switch between background and foreground 
and draw some more.  Can you see any difference between the two patterns 
that you drew?

"Do not adjust your set" to do so, since that means that not all of the 
full range of intensities will be reproducible on screen - you CAN pull 
out a difference at the expense of full colour reproduction, but that is 
simply causing the 8 bit range to saturate.

Try printing the test image - that should be even more accurate than 
your monitor, since the image on screen is produced by an 8-bit DAC on 
your video card, whilst the printer can dither individual dot densities 
to achieve exactly the tone desired for each drawn section. View at a 
suitable distance such that individual ink dots are completely dithered 
but close enough to see the patterns you have drawn.  Can you resolve 
them any better or any worse than you did on screen?

Without spoiling anyone's fun, I expect negative results - 8 bits is 
more than adequate to reproduce all of the colours discernible to the 
human eye under a fixed background level of illumination.

-- 
Kennedy
Yes, Socrates himself is particularly missed;
A lovely little thinker, but a bugger when he's pissed.
Python Philosophers
-
Turn off HTML mail features. Keep quoted material short. Use accurate
subject lines. http://www.leben.com/lists for list instructions.