Computer Graphics with Photoshop - Profiles and gamuts

Author

Profiles and gamuts

Golden

2005-08-23, 7:14 pm

In order to preserve as much color data as possible for editing in
Photoshop should I use the widest gamut profile available on my
scanner?

Are there any pitfalls to using a gamut which is too wide (unwanted
dynamic range compression, for example)?

Specifically, I'm thinking of Wide RGB and a Nikon V.

Thanks!

Golden

Bill Hilton

2005-08-23, 11:14 pm

> Golden writes ...
>
>In order to preserve as much color data as possible for editing in
>Photoshop should I use the widest gamut profile available on my
>scanner?

You can do this but there are problems editing the files because the
gamuts can be much wider than what your monitor can display.

>Are there any pitfalls to using a gamut which is too wide (unwanted
>dynamic range compression, for example)?

The main problem is the saturated colors outside the gamut of your
monitor are displayed clipped (the monitor profiles are relative
colormetric, not perceptual) so you can be changing these colors
radically yet not even know it until later when it's time to print.

Here's a link to a good article that discusses the good and bad sides
of using wide gamut spaces ...
http://www.creativepro.com/story/fe...ml?origin=story ...
here's a quote from the article illustrating the dangers of shifting
colors your monitor can't display ...
"(using) Kodak's ProPhotoRGB, you can reduce saturation by about 20 to
25 points, and shift hue by about 7 to 8 degrees, without seeing any
visible change in the display of strongly saturated colors." Except
you WILL see the changes when you convert to a smaller space for output
and usually you'll be surprised by the results.

ProPhotoRGB is wider than the WideRGB you mention but it's available in
the Adobe RAW converter as a target space and digital camera users make
use of it sometimes for saturated images, but one suggested workflow is
to convert (or in your case, assign after scanning) to the wide space
to preserve the subtle tonalities (using 16 bit/channel mode) and then
convert again (using perceptual rendering) to something easier to work
in, say AdobeRGB, so you are working in a space that better matches
what the monitor can display.

This is a controversial topic so I'm expecting some to disagree with my
post :) What I do is use the wider spaces (often Ektaspace since it
matches the gamut of slide films) with saturated films like Velvia or
for digital RAW images with saturated colors, especially reds. But for
most RAW conversions I just use AdobeRGB unless I'm worried about
saturation issues. I personally stay away from ProPhoto since it's
much wider than Ektaspace and I'm used to Ektaspace from my film scan
days (the gamut of ProPhoto is, I think, "all visible light") but
others seem to like it.

At any rate, run a few scans and convert to different profiles for
different types of images (ie, highly saturated vs not so saturated)
and see what works for you ... by the time you get to the printing
stage you might be hard pressed to find the differences for most
images, especially if they are printed small or don't have highly
saturated colors to begin with.

Bill

Bill Hilton

2005-08-23, 11:14 pm

Hecate

2005-08-23, 11:14 pm

On 23 Aug 2005 15:10:13 -0700, "Golden" <goldenlasky@hotmail.com>
wrote:

>In order to preserve as much color data as possible for editing in
>Photoshop should I use the widest gamut profile available on my
>scanner?
>
>Are there any pitfalls to using a gamut which is too wide (unwanted
>dynamic range compression, for example)?
>
>Specifically, I'm thinking of Wide RGB and a Nikon V.
>
It depends on what you're starting with and what you intend to do with
it. More info will result in a more useful reply.

--

Hecate - The Real One
Hecate@newsguy.com
Fashion: Buying things you don't need, with money
you don't have, to impress people you don't like...

Waldo

2005-08-24, 7:18 am

Golden wrote:
> In order to preserve as much color data as possible for editing in
> Photoshop should I use the widest gamut profile available on my
> scanner?
>
> Are there any pitfalls to using a gamut which is too wide (unwanted
> dynamic range compression, for example)?
>
> Specifically, I'm thinking of Wide RGB and a Nikon V.
>
> Thanks!
>
> Golden
>

If you use the widest gamut that is available, converting back into
sRGB/AdobeRGB may look very ugly.

Waldo

Bart van der Wolf

2005-08-24, 7:25 pm

"Golden" <goldenlasky@hotmail.com> wrote in message
news:1124835013.237769.165710@o13g2000cwo.googlegroups.com...
> In order to preserve as much color data as possible for editing
> in Photoshop should I use the widest gamut profile available
> on my scanner?

You should use the narrowest gamut that still accomodates your capture
device's output.

> Are there any pitfalls to using a gamut which is too wide (unwanted
> dynamic range compression, for example)?

Don't play safe by using the widest you can find, because that will
compromise accuracy for most colors, instead of for just a few
outliers.

> Specifically, I'm thinking of Wide RGB and a Nikon V.

Depending on the material you scan, the Ektaspace profile may be a
good choice. A better choice is a profile you specifically make for
the scanner + film-dye family in question.

Bart

Golden

2005-08-24, 7:25 pm

Bill Hilton wrote:
>
> You can do this but there are problems editing the files because the
> gamuts can be much wider than what your monitor can display.

That's basically why I'm asking. I like to preserve as much color data
as possible but I can't see most of it because the gamut is shrunk
before it is displayed.

>
> The main problem is the saturated colors outside the gamut of your
> monitor are displayed clipped (the monitor profiles are relative
> colormetric, not perceptual) so you can be changing these colors
> radically yet not even know it until later when it's time to print.

That's it again. Making editing decisions based on a small subset I can
see doesn't seem like a very good idea if those changes are then
applied to data I do not see. But as I say I don't know much about this
so that's why I ask.

> Here's a link to a good article that discusses the good and bad sides
> of using wide gamut spaces ...
> http://www.creativepro.com/story/fe...ml?origin=story ...
> here's a quote from the article illustrating the dangers of shifting
> colors your monitor can't display ...

Thank you very much for that. It looks like I'll be going back to sRGB.

Golden

Golden

2005-08-24, 7:25 pm

Waldo wrote:
> Golden wrote:
>
> If you use the widest gamut that is available, converting back into
> sRGB/AdobeRGB may look very ugly.

I noticed! ;-) The histogram ain't too hot either with all types of
artefacts after the conversion. That's what prompted me to ask.

Golden

Golden

2005-08-24, 7:25 pm

Bart van der Wolf wrote:
> You should use the narrowest gamut that still accomodates your capture
> device's output.

I'm still knew at all this but am I correct in saying that "scanner
profile" is basically my capture device's (Nikon V) output? Is that the
pure data from the scanner? When I compare that to sRGB (with sRGB as
my working space) I see virtually no difference.

If all that is correct it would suggest that Nikon V output can be
fully expressed in sRGB color space. Is that correct?

> Depending on the material you scan, the Ektaspace profile may be a
> good choice. A better choice is a profile you specifically make for
> the scanner + film-dye family in question.

I'm not a pro so that's probably going too far for me. I have a wide
variety of materials from some very old B&W film to all sorts of
negatives (local film I bought in the country I was at) to mostly
Kodachrome slides but there are some no name slides there as well.

So basically it looks like I should just stick to sRGB.

Golden

Johan W. Elzenga

2005-08-24, 7:25 pm

Golden <goldenlasky@hotmail.com> wrote:

> I'm still knew at all this but am I correct in saying that "scanner
> profile" is basically my capture device's (Nikon V) output? Is that the
> pure data from the scanner? When I compare that to sRGB (with sRGB as
> my working space) I see virtually no difference.

That is because you are looking at the image on a monitor, so you are
not really looking at the original colors, but a representation of those
colors in the monitors color space (which is close to sRGB).

>
> If all that is correct it would suggest that Nikon V output can be
> fully expressed in sRGB color space. Is that correct?

No, see above. The scanner probably has a wider color space, at least
AdobeRGB. The problem is that you cannot see the difference on the
average monitor.

--
Johan W. Elzenga johan<<at>>johanfoto.nl
Editor / Photographer http://www.johanfoto.nl/

Timo Autiokari

2005-08-24, 7:25 pm

"Golden" <goldenlasky@hotmail.com> wrote:

>In order to preserve as much color data as possible for editing in
>Photoshop

You should, in the first place, acquire and edit in the 16-bit/c mode.

>should I use the widest gamut profile available on my
>scanner?

In order to absolutely preserve as much color data as possible you
would of course need to optimize the acquire space according to the
material you scan but doing that that is not an easy thing to do and
is gratuitous especially when you work in 16-bit/c. In this case using
the largest gamut that the acquire device provide is the very best and
the most safest approach.

>Are there any pitfalls to using a gamut which is too wide

RGB gamuts are 3-dimensional volumes, they are either large or small,
they are all equal in "height" (luminance) but their "width" (or the
form of their cross-section plane) depends on the luminance level that
the measurement is done and their "width" at a particual luminance
level is way different depending on between what two edge colors the
"width" is measured.

Larger gamuts are good in that they have more room for saturation of
the hues at the higher luminance levels (near the whitepoint) than
what the smaller gamuts provide.

>(unwanted dynamic range compression, for example)?

No, when you acquire in 16-bit/c.

There are other aspects to the selection of the RGB space gamut,
especially to the selection of the working-space gamut. By
experimenting you can easily notice that the exactly same editing
operation can give quite different results in different RGB
working-spaces. So ...then... this immediately raises the question:
What is the correct result, they can not all be correct, and hence
what is the correct gamut to choose for the RGB working-space.

Timo Autiokari
http://www.aim-dtp.net/

Hecate

2005-08-24, 11:17 pm

On Wed, 24 Aug 2005 11:44:09 +0200, Waldo <nobody@onbody.co> wrote:

>Golden wrote:
>
>If you use the widest gamut that is available, converting back into
>sRGB/AdobeRGB may look very ugly.
>
It can do, but not if you use the mechanism described in Don Margulis'
Professional Photoshop where you can end up converting between
WideRGB, CMYK, LAB and another RGB.

--

Hecate - The Real One
Hecate@newsguy.com
Fashion: Buying things you don't need, with money
you don't have, to impress people you don't like...

Waldo

2005-08-25, 4:16 am

Hecate wrote:
> On Wed, 24 Aug 2005 11:44:09 +0200, Waldo <nobody@onbody.co> wrote:
>
>
>
> It can do, but not if you use the mechanism described in Don Margulis'
> Professional Photoshop where you can end up converting between
> WideRGB, CMYK, LAB and another RGB.

Therefore I said the word "may". It really depends on the content of the
image.

Also, converting between different color spaces is always done via a
"profile connection space" (PCS), a device independant color space.
Mostly XYZ and LAB are used for this purpose.

Waldo

Johan W. Elzenga

2005-08-25, 7:15 pm

Waldo <nobody@onbody.co> wrote:

> If you use the widest gamut that is available, converting back into
> sRGB/AdobeRGB may look very ugly.

I doubt that very much. No matter how wide the color space, you are
looking at the image *in your monitors color sapce* already. That is
usually very close to sRGB and even the best monitors are not more than
AdobeRGB. So, you are already looking at a 'converted image' if you are
looking at it on your monitor. If it doesn't look ugly on your monitor
right now, there is no reason why it would look ugly after conversion to
sRGB or AdobeRGB. In fact, you probably wouldn't even notice any
difference, for the above reason.

--
Johan W. Elzenga johan<<at>>johanfoto.nl
Editor / Photographer http://www.johanfoto.nl/

Golden

2005-08-25, 7:15 pm

Johan W. Elzenga wrote:
> Golden <goldenlasky@hotmail.com> wrote:
>
>
> That is because you are looking at the image on a monitor, so you are
> not really looking at the original colors, but a representation of those
> colors in the monitors color space (which is close to sRGB).

That takes me back to my original question, how do I edit an image in a
working space which is narrower than the color space of the image? What
do you pros do? Is it a 6th sense or are there some practical things I
can draw on?

>
> No, see above. The scanner probably has a wider color space, at least
> AdobeRGB. The problem is that you cannot see the difference on the
> average monitor.

What I'm still confused about is how does this conversion happen?

Please bear with me because this is hard to explain. Let's just for the
sake of the argument imagine two gamuts: "Big" with 100 colors and
"Lil" (a subset of Big) with 50 colors. Let's also assume that scanner
gamut is identical to "Lil" gamut. Finally, let's say that the
available dynamic range can handle 50 colors.

Given all that it appears to me that using Lil gamut is actually better
than using Big gamut. Why? Because expressing the scanner's gamut with
Lil gamut will utilize the available dynamic range fully. Big gamut
will only use a fraction (50%) of its available dynamic range to
express Lil and the rest will end up unused. In other words, Lil gamut
will have to be compressed to fit the space allocated to it within Big
gamut.

Does that make sense?

Golden

Golden

2005-08-25, 7:15 pm

Timo Autiokari wrote:
> "Golden" <goldenlasky@hotmail.com> wrote:
>
>
> You should, in the first place, acquire and edit in the 16-bit/c mode.

Yes I definitely do that.

> In this case using
> the largest gamut that the acquire device provide is the very best and
> the most safest approach.

That's was my initial instinct but now I've been rattled.

>
> No, when you acquire in 16-bit/c.

What about the case I raise in my answer to Johan?

Each gamut utilizes the available dynamic range fully. Right?

If that's true then using a bigger gamut to express a smaller one will
only end up compressing the smaller gamut and so make matters worse.
Yes, there will be more room at both ends but I start with fewer
colors.

Golden

John

2005-08-25, 7:15 pm

"Golden" <goldenlasky@hotmail.com> wrote in message
news:1124978187.676787.16740@g14g2000cwa.googlegroups.com...
> Johan W. Elzenga wrote:
>
> What I'm still confused about is how does this conversion happen?

It is nornally done by an 'on the fly' profile to profile conversion between
your working colour space (e.g. Adobe RGB etc.) and your monitor's profile
(as defined by your monitor calibration) using relative colorimetric
rendering. Relative colorimetric rendering causes colours which are common
to both working space and monitor to display correctly, but colours in the
working space which are beyond the monitor's capability are clipped to the
nearest hue. Hence 2 distinct colours in your working space will look
identical on your monitor. However, the data in your file is not affected -
as I said, the conversion to monitor space is done 'on the fly'.

>
> Please bear with me because this is hard to explain. Let's just for the
> sake of the argument imagine two gamuts: "Big" with 100 colors and
> "Lil" (a subset of Big) with 50 colors. Let's also assume that scanner
> gamut is identical to "Lil" gamut. Finally, let's say that the
> available dynamic range can handle 50 colors.
>
> Given all that it appears to me that using Lil gamut is actually better
> than using Big gamut. Why? Because expressing the scanner's gamut with
> Lil gamut will utilize the available dynamic range fully. Big gamut
> will only use a fraction (50%) of its available dynamic range to
> express Lil and the rest will end up unused. In other words, Lil gamut
> will have to be compressed to fit the space allocated to it within Big
> gamut.
>
> Does that make sense?

Gamut is *not* about 'number' of colours or dynamic range - it is about
the *range* of colours available, i.e. how saturated a colour can be
represented. Technically, a scanner does not have a gamut - it will return a
set of RGB values for any colour it sees; only a calibration (i.e. scanner
profile) gives meaning to those RGB values by assigning them to *device
independent* colours as defined by Lab or XYZ. This calibration will then
define the *effective* gamut of the scanner.

As an example, let's say the scanner was calibrated with a Velvia
calibration slide. The slide will contain the full range of colours
available on Velvia film. I am sure you are aware that Velvia film can
produce colours far more saturated than the average monitor which has an
sRGB like characteristic. If the scanner has been calibrated with such a
target, it will have an effective gamut like Velvia film - far in excess of
sRGB. If you decide to use sRGB as your working space, the the colours from
your scanner need to be converted to fit that smaller colour space. This can
be done *perceptually* - i.e. by compressing the entire Velvia gamut into
the limits of sRGB (which will reduce the saturation of *ALL* the colours,
both in and out of gamut), or *colorimetrically* - i.e. by clipping *only*
out of gamut colours to the nearest sRGB hue and hence losing the separation
between some of the colours.

The range of colours available, as defined by the gamut of the colour space,
is divided up by the *number* of available colours as defined by the bit
depth (e.g. 8-bit, 16 bit). A wide gamut colour space divides a *wider*
range of colours into the same number of increments as a narrow gamut colour
space. Hence the *jumps* between individual colours in a wide gamut space
are greater than in a narrow space. Hence the recommendation to use 16 bit
colour when using wide gamut colour spaces.

--
John
Replace 'nospam' with 'todnet' when replying.

Johan W. Elzenga

2005-08-25, 7:15 pm

Golden <goldenlasky@hotmail.com> wrote:

> Please bear with me because this is hard to explain. Let's just for the
> sake of the argument imagine two gamuts: "Big" with 100 colors and
> "Lil" (a subset of Big) with 50 colors. Let's also assume that scanner
> gamut is identical to "Lil" gamut. Finally, let's say that the
> available dynamic range can handle 50 colors.

No, as John explained, it's not about MORE colors, it's about a more
saturated color range.

>
> Given all that it appears to me that using Lil gamut is actually better
> than using Big gamut. Why? Because expressing the scanner's gamut with
> Lil gamut will utilize the available dynamic range fully. Big gamut
> will only use a fraction (50%) of its available dynamic range to
> express Lil and the rest will end up unused. In other words, Lil gamut
> will have to be compressed to fit the space allocated to it within Big
> gamut.
>
> Does that make sense?

In theory, yes but in practise, no. If your scanner has a color gamut
that is close to AdobeRGB, and you scan in ProPhotoRGB, a lot of the
ProPhotoRGB space will remain unused. That is indeed true and it makes
sense: Using ProPhotoRGB does not somehow increase the maximum color
saturation your scanner can capture, so the most saturated colors of
ProPhotoRGB will not be really present. However, if you scan in 16 bits,
the 'excess' compared to 8 bits is so big, that this doesn't really
matter.

--
Johan W. Elzenga johan<<at>>johanfoto.nl
Editor / Photographer http://www.johanfoto.nl/

Timo Autiokari

2005-08-25, 7:15 pm

"Golden" <goldenlasky@hotmail.com> wrote:

>Each gamut utilizes the available dynamic range fully. Right?

Dynamic range usually refers to the gray axis, that axis is exactly
the same in what ever RGB gamut.

With colors (with all the other codes than the grays that are the same
as R=G=B) there is also dynamic range but it is better related to
saturation and to the maximum available luminance for the particual
hue. Larger gamut can hold both higher saturation and higer luminance
values for hues.

>If that's true then using a bigger gamut to express a smaller one will
>only end up compressing the smaller gamut

Yes it is true.

>and so make matters worse.

Not so if you work in the 16-bit/c space.

There is more quantization for colors of course but the 16-bit/c space
(that in Photoshop is only 15-bit/c + 1 level) can easily withstand
that. For example, in the worst case, there is quantization of only 1
bit between the gamuts of adobeRGB and CIE1931 (that is a very very
large gamut). So converting image data in Photoshop 16-bit/c space
from AdobeRGB to CIE_1931_D65_Gamma_1 profile will result effectively
14-bit/c (since it really was in 15-bit/c space). And that
quantization does not happen all over the gamut but only locally here
and there.

Now, none of us have such image data that truly is 15-bit/c, typically
we have something like 10 to 14 bit/c A/D conversion _from_ sensor
data that is effectively comparable to less than 10bit/c. So there is
extra room in 16-bit/c to hold the image data in large gamut RGB
spaces. E.g in case you have the typical 12-bit/c A/D converter in
your scanner or camera then the image data that is destined to
Photoshop 16-bit/c mode is multiplied by 8, so :

original 12-bit level 0 becomes level 0*8=0 in the file Photoshop sees
original 12-bit level 1 becomes level 1*8=8 in the file Photoshop sees
original 12-bit level 2 becomes level 2*8=16 in the file Photoshop
sees

and so on. So nothing is lost if the quantization is doubled, it could
be doubled several times.

Timo Autiokari
http://www.aim-dtp.net/

Hecate

2005-08-25, 7:16 pm

On 25 Aug 2005 06:56:27 -0700, "Golden" <goldenlasky@hotmail.com>
wrote:

>That takes me back to my original question, how do I edit an image in a
>working space which is narrower than the color space of the image? What
>do you pros do? Is it a 6th sense or are there some practical things I
>can draw on?
>
Which brings me back to the question I asked you which you never
replied to: for what purpose? For example, I use AdobeRGB all the way
through my workflow (though I will convert in and out of other spaces
for specific reasons). Why? Because it maps most closely to CMYK and
I'm producing work for print.

--

Hecate - The Real One
Hecate@newsguy.com
Fashion: Buying things you don't need, with money
you don't have, to impress people you don't like...

Golden

2005-08-26, 7:16 pm

John wrote:
> Gamut is *not* about 'number' of colours or dynamic range - it is about
> the *range* of colours available, i.e. how saturated a colour can be
> represented.

That's what I mean even though my terminology is way off. This is what
I was thinking:

The number of colors in the real world is virtually infinite. But in
practical terms even the widest gamut is limited by the available
dynamic range used to express it because there is only a finite number
of colors available even in 16-bits. Ignoring specialist high
definition range image file formats this puts a ceiling on how many
visible colors can be expressed so in that sense this infinite gamut is
expressed using only 64K distinct colors. What I was confused about is
how different color spaces use this available limited range of only 64K
distinct colors and how these color spaces are mapped relative to each
other within this limited and finite range.

> Technically, a scanner does not have a gamut - it will return a
> set of RGB values for any colour it sees;

>From what I understand these values are returned when scanning in what
Nikon calls "scanner RGB" when there is no profile attached to the
resulting file.

What I did is open such as file in two working color spaces sRGB and
Wide using "Assign Working RGB" when the dialog box pops up. The sRGB
version looked fine but the Wide version looked weird. Actually it
looked exactly the same as when I open a file tagged with sRGB in Wide
working color space. (I can desaturate monitor colors but that gets me
even further away from what's really in the file.)

>From all this I drew the conclusion probably incorrectly that this
"scanner RGB", which as I understand contains all available scanner
colors, could be expressed fully in sRGB alone.

> The range of colours available, as defined by the gamut of the colour space,
> is divided up by the *number* of available colours as defined by the bit
> depth (e.g. 8-bit, 16 bit). A wide gamut colour space divides a *wider*
> range of colours into the same number of increments as a narrow gamut colour
> space. Hence the *jumps* between individual colours in a wide gamut space
> are greater than in a narrow space. Hence the recommendation to use 16 bit
> colour when using wide gamut colour spaces.

That's exactly what I was getting at only I wasn't able to express it
due to my lack of terminology and knowledge in general.

Thank you for taking the time to explain all this, John! Things are
slowly starting to make sense but I still have a very steep learning
curve ahead of me and that includes rereading this message a few more
times.

Golden

Golden

2005-08-26, 7:16 pm

Johan W. Elzenga wrote:
> Golden <goldenlasky@hotmail.com> wrote:
>
>
> In theory, yes but in practise, no. If your scanner has a color gamut
> that is close to AdobeRGB, and you scan in ProPhotoRGB, a lot of the
> ProPhotoRGB space will remain unused. That is indeed true and it makes
> sense: Using ProPhotoRGB does not somehow increase the maximum color
> saturation your scanner can capture, so the most saturated colors of
> ProPhotoRGB will not be really present. However, if you scan in 16 bits,
> the 'excess' compared to 8 bits is so big, that this doesn't really
> matter.

Which means that my instinct to just automatically go for the largest
available gamut without thinking is definitely wrong. While it may
represent a larger range of colors most of them will not be used. But
by using 16-bits, and I do, this becomes a bit of a moot point because
of the large number of colors available in 16-bits.

As someone said earlier, and I think now I'm starting to understand it,
the optimum gamut is the one which expresses all colors present in the
scanner. If the gamut is larger then the scanner gamut will get
compressed although that may not be much of a problem because 16-bits
have lots of overhead. If the gamut is smaller then it may not be able
to express all colors present in the scanner gamut but this will be
offset in part by getting a truer representation on the screen of
what's in the file which means fewer problems caused by editing and no
conversions for printing or display.

So the bottom line is there are no easy answers because it's all pretty
complicated with conflicting things pulling in different directions
depending on usage. Exactly what I did *not* want to hear! :-)

Golden

Golden

2005-08-26, 7:16 pm

Hecate wrote:
> On 25 Aug 2005 06:56:27 -0700, "Golden" <goldenlasky@hotmail.com>
> wrote:
>
> Which brings me back to the question I asked you which you never
> replied to: for what purpose? For example, I use AdobeRGB all the way
> through my workflow (though I will convert in and out of other spaces
> for specific reasons). Why? Because it maps most closely to CMYK and
> I'm producing work for print.

Sorry I got sidetracked by all these new things I'm learning.

My use is fairly pedestrian. I plan to keep all my scans in their pure
form before I do any touching up. I will then make various copies for
different uses for example reduce them to JPGs for viewing or print
them out for framing. But I want to keep the original scans as pure as
I can so that's why I wanted to capture as much as possible and why I
wanted to use the largest gamut.

Golden

Golden

2005-08-26, 7:16 pm

Timo Autiokari wrote:
> "Golden" <goldenlasky@hotmail.com> wrote:
>
>
> Yes it is true.
>
>
> Not so if you work in the 16-bit/c space.

Yes I do. That's a decision I made right on the outset.

> There is more quantization for colors of course but the 16-bit/c space
> (that in Photoshop is only 15-bit/c + 1 level) can easily withstand
> that. For example, in the worst case, there is quantization of only 1
> bit between the gamuts of adobeRGB and CIE1931 (that is a very very
> large gamut). So converting image data in Photoshop 16-bit/c space
> from AdobeRGB to CIE_1931_D65_Gamma_1 profile will result effectively
> 14-bit/c (since it really was in 15-bit/c space). And that
> quantization does not happen all over the gamut but only locally here
> and there.

How much difference is there between sRGB and adobeRGB, or even Wide? I
know you wrote that "sadRGB" is very limited but how many bits would I
lose by using it over adobeRGB or Wide?

My scanner is nominally 14-bits and I'm thorn between conflicting
things pulling in different directions. I don't want to sacrifice
accuracy my scanner can deliver but I also don't want the risk of
editing such highly accurate data by looking at the narrow information
my monitor shows me.

And thank you also Timo for a detailed explanation and for taking the
time.

Golden

Timo Autiokari

2005-08-27, 4:14 am

"Golden" <goldenlasky@hotmail.com> wrote:

>How much difference is there between sRGB and adobeRGB, or even Wide?

It took some effort to answer that question. First I explain what I
did:

Calculations were performed in double precision floating point in
Excel. For the smaller of the two gamut that are compared I created
29791 sample RGB patches, so each channel has 31 numerically equally
spaced points from 0.001 to 1. This means that I used the range of
1000:1 (this was to avoid zero channel values) .

Then converted this RGB dataset to the larger gamut.

Then took the divisions Rs/Rl, Gs/Gl and Bs/Sl for each of the 29791
RGB pairs where index "s" means the smaller gamut value and index "l"
means the larger gamut value. Because both the spaces have the same
amount of digital levels per channel these results are in essence the
multipliers with what the per channel quantisation steps in the
smaller gamut are multiplied for that particular color when looking at
the situation in the larger gamut.

I then took max, average, variance and stdev over these multipliers
(over the 3*29791 result values).

An example (using 8-bit notation for simpicity): In case we have
RGBsmall=100,50,25 that converts to RGBlarge=50,80,30 then for this
color (the color that we see is the same in both the spaces, just the
RGB values are different) the red channel in the larger gamut has
100/50 = 2 times the quantization what the R channel of the same color
has in the RGBsmall. And for this color the green channel in the
larger gamut has 50/80 = 0.625 times the quantization what the G
channel of the same color has in the RGBsmall. And for this color the
blue channel in the larger gamut has 25/30=0.833 times the
quantization what the B channel of the same color has in the RGBsmall.

In addition I used D65 whitepoint, absolute colorimetry and since we
are interested about the effect of the gamut volume alone the gamma
was set to 1.0 for all the color spaces.

Now to the results, they are quite interesting:

sadRGB to adobergb(1998):
max=1.3978, average=0.9585, variance=0.049, stdev=0.2213

sadRGB to Widegamut:
max=3.1208, average=0.932, variance=0.0934, stdev=0.3056

sadRGB to CIE XYZ:
max=2.3017, average=0.9323, variance=0.1311, stdev=0.3621

adobergb(1998) to Widegamut:
max=3.4334, average=0.9245, variance=0.0578, stdev=0.2404

adobergb(1998) to CIE XYZ:
max=1.6471, average=0.9229, variance=0.1058, stdev=0.3252

(multiplier 0.5 means one bit gain, multiplier 1 means no change,
multiplier 2 means one bit loss, multiplier 4 means two bits loss to
the gradation).

So, when going from a smaller gamut to a larger gamut there are some
colors (some subvolumes inside the whole gamut volume) that suffer a
little (since the max value is > 1) however in al the cases the
average is a little less than 1 so, on average, there is a little
benefit (the gradation steps are, on average, a little smaller). The
small variance means that the multiplier are rather close to the
average so the subvolumes of the gamut where the gradation goes close
to the max are very small.

Now, these results seem to to be impossible, until it is realized that
the dataset that was used was limited to the to smaller gamut. This is
the only way how the gradation in the RGB channels can be examined
since the rest of the colors that the larger gamut holds are naturally
out-of-gamut colors for the smaller gamut (so can not be defined there
at all). This means that the gradation of those colors that can only
be defined in the larger gamut must be more coarse so that the larger
gamut size comes true.

So, this evaluation tells that there is absolutely nothing to worry
about when converting from a smaller gamut RGB working-space to a
larger one in regards of quantization as long as the data is kept in
the 16-bit/c mode. Even the very large "CIE 1931 D65 Gamma 1.0"
profile that I have been using, for over a year now, as my RGB
working-space does not introduce any problems.

I did this evaluation using my AIM.XLA, it is freeware Excel Add-In
for colorimetric and spectral color calculations, available at
http://www.aim-dtp.net/aim/technology/aim_xla/index.htm

>My scanner is nominally 14-bits

So it has 14-bit analog to digital converter. However the sensor in
your scanner has something about 10-bit effective dynamic range at the
best so the extra 4 bits do not contain useful image data.

Timo Autiokari
http://www.aim-dtp.net/

Golden

2005-08-27, 7:16 pm

Timo Autiokari wrote:
> "Golden" <goldenlasky@hotmail.com> wrote:
>
>
> It took some effort to answer that question.

Thank you very much for that, Timo! I feel guilty you went through all
that trouble. It will take me some time to digest it all and I really
appreciate the effort.

> So, this evaluation tells that there is absolutely nothing to worry
> about when converting from a smaller gamut RGB working-space to a
> larger one in regards of quantization as long as the data is kept in
> the 16-bit/c mode.

Thanks again for clarifying that. As I said I do use 16-bit mode.

> I did this evaluation using my AIM.XLA, it is freeware Excel Add-In
> for colorimetric and spectral color calculations, available at
> http://www.aim-dtp.net/aim/technology/aim_xla/index.htm

I have often been to your site. I particularly like the gamma curves
because they produce much more visually pleasing results than
Photoshop's gamma. I find that your gamma files produce smoother
gradients.

>
> So it has 14-bit analog to digital converter. However the sensor in
> your scanner has something about 10-bit effective dynamic range at the
> best so the extra 4 bits do not contain useful image data.

I did notice that because excessive noise is often a problem when
scanning difficult film and Coolscan V doesn't have multiscanning.

Golden