Re: 1 billion colors
Re: 1 billion colors
- Subject: Re: 1 billion colors
- From: Andrew Rodney via colorsync-users <email@hidden>
- Date: Mon, 6 Jan 2020 13:06:27 -0700
> On Jan 6, 2020, at 12:32 PM, Wire ~ via colorsync-users
> <email@hidden> wrote:
>
> On Mon, Jan 6, 2020 at 06:52 Andrew Rodney via colorsync-users <
> email@hidden> wrote:
>
>> I already provided an example in 24-bit encoding within sRGB where two
>> sets of device values are the same color (deltaE of 0.01). Two numbers, ONE
>> color.
>>
>> It's just math to divide up numbers. Analogy. I have a 3 pound apple pie.
>> I can make 16 slices to serve to company. I can make 8 bigger slices or 32
>> much smaller slices, it's still an 8 pound pie.
>
>
> Hi Andrew, would you be so kind as to contextualize your observations?
http://digitaldog.net/files/ColorNumbersColorGamut.pdf
> Why is it significant that there might be indistinguishable colors in an 8
> bpc format vs an 10 bpc format?
The significance is that colors have to be observed as outlined by Fairchild
and others. If you can't see it, it's not a color.
There are numbers that don't define colors. I provided an example in ProPhoto
RGB.
There are differing sets of numbers that are the same color. I provided an
example in sRGB using 24-bit encoding.
> Aren't Indistinguishable colors a trivial concern? Just turn off your
> screen and bamm! :)
Colors are always observable. Some triplets of color numbers are not. Some
close triplets are the same observed color.
We can define 16.7 million numbers but we can't see them all as colors. Because
depending on the color space, the numbers fall outside human vision (and as
such can't be called colors.
There are groups of numbers in that 16.7 million sets that appear as the same
color.
Depending on the color space, there are groups of numbers that in 16.7 million
sets are invisible. They are not colors.
There are numbers used in all kinds of industries and in science that define
things we can't see. Colors are seen, not all color numbers (device values) are
visible or indistinguishable.
I know of no color expert or resource that states the standard observer model
provides the visibility of 16.7 million colors. While I know of no total
agreement, I've heard figures of 7-12 million. 16.7 is simply a value that's
new and 'special' due to the encoding of 8-bits per color NUMBERS on computer
systems. That doesn't make them colors. That you can divide up numbers into
billions of values doesn't make them colors and certainly doesn't make them
when visisble., indistinguishable; my 24 bit sRGB example illustrates that.
> The marketeers have "counted" the "colors". It's as simple as this. And the
> claim is no lie🎱
No, they have not! They have counted the numbers of encoding values. Just like
in my Pie analogy, no matter how you slice up a 3lb pie into tinier pieces,
it's still 3 pounds.
This isn't to say high bit encoding isn't useful; it absolutely is! And I've
gone on record discussing where it is useful:
http://digitaldog.net/files/TheHighBitdepthDebate.pdf
> If you're gonna trudge on about invisible colors, we should keep in mind
> that as well as color blindness, which is quite common 1/13 men I think,
> there are some people—very rare and typically women—who can see well beyond
> the spectrum locus, like they can see a fourth primary. Also, from an
> anthropological perspective, local populations have been found that can
> distinguish within certain ranges, say green, much better than average.
There really is no such thing as an invisible color. If you can't see it, it's
not a color.
I can't speak to the 'Standard color blind Observer' as I don't know if any
such studies were ever done on this group.
I CAN speak of the Standard Observer based on perpetual experiments dating back
to the 1930's and the theoretical plot of what the Standard Observer can see as
colors.
And I can plot device values in some color spaces, numbers, that fall outside
what the Standard Observer can see as colors. Hence, if we are to believe
experts like Fairchild, they are NOT colors. They ARE numbers.
> Then there is the psychological adaptation: you can learn to observe.
You can't see R0/G255/B0 in ProPhoto RGB any more than you can see into the
infrared spectrum.
> And there's the matter of unconscious bias, as this whole topic of the
> physics of the qualia is being encountered under the even more strange and
> elusive regime of language.
You can't see R0/G255/B0 in ProPhoto RGB any more than you can see into the
infrared spectrum.
> It seems like what you are saying is the "billions of colors" claim is a
> spooky bit of both?
It's not spooky, it's simply wrong if we agree, as many experts like Fairchild
do, that color is a perceptual attribute of (in this case) humans.
> If you don't need 10 bpc that's fine.
Yes, it is and it's useful as outlined in the article above about editing high
bit data. That doesn't change two facts:
1. We can't see 16.7 million colors let alone billions.
2. We can define millions and billions of numbers.
3, There are numbers we can define that we cannot see and thus, they are not
colors. They are numbers.
> Per Florian's comment about cal headroom, and Roger's question about HW
> LUTs, putting the extra bits in the display can add life to good 'ol 24bit
> color systems.
I never stated that wasn't the case and totally agree that high bit data is
very useful. As you can see, I've provided articles on that topic. The
usefulness of high bit data isn't under debate, never was.
> If what you are saying is a high fidelity image can be built from a small
> palette, isn't that called half-toning? Aren't you DTP nerds?! omg2
That isn't what I'm saying and never did. I hope the text above will make my
position clear.
_______________________________________________
Do not post admin requests to the list. They will be ignored.
colorsync-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden