Out of Gamut unfixed, compresses all hues

[skid00skid00]

If I *don't* fix out-of-gamut colors before printing, are only the OOG
colors compressed back into gamut, or are other hues also affected?

Are there times when you chose not to fix OOG before rendering, for
artistic reasons?

Thanks.

[francois]

If I *don't* fix out-of-gamut colors before printing, are only the OOG
colors compressed back into gamut, or are other hues also affected?

Are there times when you chose not to fix OOG before rendering, for
artistic reasons?

Thanks.
[a href=\"index.php?act=findpost&pid=65307\"][{POST_SNAPBACK}][/a]

OOG colors will be compressed using the rendering intent you choose. Relative Colorimetric intent compresses OOG colors only and Perceptual intent affects all colors. You can see this using Photoshop soft proofing.

[bjanes]

OOG colors will be compressed using the rendering intent you choose. Relative Colorimetric intent compresses OOG colors only and Perceptual intent affects all colors. You can see this using Photoshop soft proofing.
[a href=\"index.php?act=findpost&pid=65313\"][{POST_SNAPBACK}][/a]

The above statements are not quite correct. Relative colorimetric clips out of gamut colors--it does not compress them. Perceptual intent compresses all colors, but the extent of compression varies with the implementation. Some implementations compress all colors even though there may be no out of gamut colors, thereby limiting the gamut unnecessarily. In this situation, a wide space such as ProPhotoRGB should be used only when necessary. A more ideal implementation would use just enough compression to bring out of gamut colors into gamut. Usually, the results are between the above extremes.

[francois]

The above statements are not quite correct. Relative colorimetric clips out of gamut colors--it does not compress them. Perceptual intent compresses all colors, but the extent of compression varies with the implementation. Some implementations compress all colors even though there may be no out of gamut colors, thereby limiting the gamut unnecessarily. In this situation, a wide space such as ProPhotoRGB should be used only when necessary. A more ideal implementation would use just enough compression to bring out of gamut colors into gamut. Usually, the results are between the above extremes.
[a href=\"index.php?act=findpost&pid=65316\"][{POST_SNAPBACK}][/a]

Thanks for correcting my sloppy language, I kept skid00skid00 phrasing.

[digitaldog]

A picture is worth a thousands words.

On the left is the Perceptual  intent and on the right is Colorimetric.

[Tim Gray]

A picture is worth a thousands words.

On the left is the Perceptual  intent and on the right is Colorimetric.
[a href=\"index.php?act=findpost&pid=65329\"][{POST_SNAPBACK}][/a]

Thanks for the graphic.  Further to Francois' previous post - what happens to in-gamut colors in a Perceptual rendering (the picture only shows oog examples) - specificaly as implemented in PS CS2?

[digitaldog]

Thanks for the graphic.  Further to Francois' previous post - what happens to in-gamut colors in a Perceptual rendering (the picture only shows oog examples) - specificaly as implemented in PS CS2?
[a href=\"index.php?act=findpost&pid=65331\"][{POST_SNAPBACK}][/a]

All colors are affected (although note that there's no standards in how this is done). With a relative colorimetric rendering intent the colors that fall within gamut are not affected at all.

[bjanes]

Thanks for the graphic.  Further to Francois' previous post - what happens to in-gamut colors in a Perceptual rendering (the picture only shows oog examples) - specificaly as implemented in PS CS2?
[{POST_SNAPBACK}][/a]

Here is a link that addresses your question:

[a href=\"http://steves-digicams.com/techcorner/July_2005.html]http://steves-digicams.com/techcorner/July_2005.html[/url]

"The current ICC CMM (color management model) is not a "smart" model meaning that it cannot and does not examine the gamut of the actual image  before trying to compress it to fit in the printer's gamut"

Another consideration with PSCS2 is that with matrix profiles (such as aRGB and ProPhotoRGB) the rendering intent is always relative colorimetric--the lookup tables for perceptual rendering are not included in the profile and thus perceptual rendering is not possible even though Photoshop gives no warning if you choose perceptual rendering with these profiles.

[Charles Beasley]

Hi Andrew-Nice to find you here; I missed you over at RG's.

When I download the file from your link, and try to open in PS, I get an error message that "it is not a tiff file."

(PS 7, Win XP, IE 6.0.29)

Is something wrong at my end or yours?

[digitaldog]

Hi Andrew-Nice to find you here; I missed you over at RG's.

When I download the file from your link, and try to open in PS, I get an error message that "it is not a tiff file."

(PS 7, Win XP, IE 6.0.29)

Is something wrong at my end or yours?
[a href=\"index.php?act=findpost&pid=65361\"][{POST_SNAPBACK}][/a]

What file?

[Richowens]

Charles,

You need Quicktime to open the file.

You can download it from Apple' website.

Rich

[PeterLange]

Can’t resist to smile,

Charles, to see you as beginner with 1 post.

So long!   Peter

--

[opgr]

Another consideration with PSCS2 is that with matrix profiles (such as aRGB and ProPhotoRGB) the rendering intent is always relative colorimetric--the lookup tables for perceptual rendering are not included in the profile and thus perceptual rendering is not possible even though Photoshop gives no warning if you choose perceptual rendering with these profiles.

CMMs are free to do whatever the heck they please with rendering intents, even for matrix based profiles. The host application can not usefully know if a CMM changes its behavior for different rendering intents.

You are correct that currently the Adobe CMM (ACE) more-or-less ignores the rendering intent for matrix based conversions, and always applies Relative/Absolute Colorimetric intent. This may not be true however for other CMMs, nor for future implementations of the Adobe CMM.

[Charles Beasley]

Andrew-the file your link refers to in your most recent post in this thread-

Figure1_11.tiff

Rich-Thanks, I was able to open it with Quicktime; I wasn't given a choice to do so in the browser, but could after saving it to disk.

Still, if it's a tiff, why doesn't PS open it?
Must be that it isn't a tiff but since it has the tiff extension, windows thinks it is.
Maybe should be .qtif?

Peter-And you with 53
Hello again to you also. I have missed you as well at RGs.
   

It's nice to find where all "the big boys" are nowadays.
Must have been hiding from me, as no one told me where you
were going. But I found ya

[digitaldog]

Andrew-the file your link refers to in your most recent post in this thread-

Figure1_11.tiff
[a href=\"index.php?act=findpost&pid=65417\"][{POST_SNAPBACK}][/a]

Its just a Tiff that's been attached to this site and it opens just fine for me.

[Charles Beasley]

Well, not on my windoze box. But it does if I save it to disk and change the extension to ".qtif"

Still, thanks for the diagram.

[jani]

When I check the file, I get:

Figure1_11.tiff: Adobe Photoshop Image

So it's probably using some of Adobe's TIFF-RGB extensions, layers or something. Maybe Photoshop 7 doesn't support that. ImageMagick, however, does, and you can download that for free from here:

http://www.imagemagick.org/script/binary-releases.php

Scroll to the bottom for the Windows versions, you probably want the ones supporting 16 bpc.

(Bonus: if you also download Cygwin and know a bit of shell scripting from Unix/Linux, you then have an excellent set of batch processing tools for images.)

[bjanes]

CMMs are free to do whatever the heck they please with rendering intents, even for matrix based profiles. The host application can not usefully know if a CMM changes its behavior for different rendering intents.

You are correct that currently the Adobe CMM (ACE) more-or-less ignores the rendering intent for matrix based conversions, and always applies Relative/Absolute Colorimetric intent. This may not be true however for other CMMs, nor for future implementations of the Adobe CMM.
[{POST_SNAPBACK}][/a]

Completely ignores is more accurate, as explained by Thomas Knoll in a post on the Abobe forum. With matrix profiles relative colorimetric is always used.

[a href=\"http://www.adobeforums.com/cgi-bin/webx/.3bbba392/0]http://www.adobeforums.com/cgi-bin/webx/.3bbba392/0[/url]

Perecetual rendering is not currently possible with matrix profiles because the matrix profiles lack the perceptual rendering tables necessary for the conversion. While it would be possible to have a "smart" rendering intent engine which would examine the contents of the image rather than the size of the color space and apply gamut compression, I don't think such a method would be ICC compliant.

[bjanes]

Completely ignores is more accurate, as explained by Thomas Knoll in a post on the Abobe forum. With matrix profiles relative colorimetric is always used.

http://www.adobeforums.com/cgi-bin/webx/.3bbba392/0

Perecetual rendering is not currently possible with matrix profiles because the matrix profiles lack the perceptual rendering tables necessary for the conversion. While it would be possible to have a "smart" rendering intent engine which would examine the contents of the image rather than the size of the color space and apply gamut compression, I don't think such a method would be ICC compliant. Perhaps Andrew can comment.
[a href=\"index.php?act=findpost&pid=65537\"][{POST_SNAPBACK}][/a]

[opgr]

Completely ignores is more accurate, as explained by Thomas Knoll in a post on the Abobe forum. With matrix profiles relative colorimetric is always used.

I suppose he wrote that for the sake of simplicity.

Photoshop doesn't do anything, it simply tells the relevant CMM what to do.

A matrix based profile primarily means that the corresponding colorspace (or device) acts linearly. It is completely irrelevant whether any tables are available. A CMM may create such tables from the linear matrix conversion if it wants to. It could emulate the conversion in a printer profile for example. That's irrespective of any image dependent conversion...