New article on color management

Started Apr 26, 2013 | Discussions
Jeff Schewe
Regular MemberPosts: 371
Like?
Re: Camera-->sRGB versus Camera-->ProPhoto-->sRGB
In reply to crames, May 1, 2013

crames wrote:

I'm pretty sure they both should be v4 in order to make use of the perceptual Profile Connection Space.

Actually, the profile connection space has nothing to do with rendering intents...in the case of the sRGB Perceptual profile, the rendering intent is used in the receiving color space. So you can convert from a standard v2 ProPhoto RGB into the v4 sRGB perceptual profile and take advantage of te gamut compression of the perceptual rendering intent. I have it an use it...it's useful for images with highly saturated red and blues.

-- hide signature --

Regards,
Jeff Schewe

Reply   Reply with quote   Complain
crames
Regular MemberPosts: 192
Like?
Re: Camera-->sRGB versus Camera-->ProPhoto-->sRGB
In reply to Jeff Schewe, May 1, 2013

Jeff Schewe wrote:

crames wrote:

I'm pretty sure they both should be v4 in order to make use of the perceptual Profile Connection Space.

Actually, the profile connection space has nothing to do with rendering intents...in the case of the sRGB Perceptual profile, the rendering intent is used in the receiving color space. So you can convert from a standard v2 ProPhoto RGB into the v4 sRGB perceptual profile and take advantage of te gamut compression of the perceptual rendering intent. I have it an use it...it's useful for images with highly saturated red and blues.

Thanks for the info, Jeff. Good to know it works with v2 ProPhoto RGB.

I was thinking of this statement on the ICC's sRGB download page :

"The sRGB v4 ICC preference profile is a v4 replacement for commonly used v2 sRGB profiles. It gives better results in workflows that implement the ICC v4 specification. It is intended to be used in combination with other ICC v4 profiles."

I guess you don't have to use a v4 source profile, but potentially the results are better if you do?

More details here: Using the sRGB_v4_ICC_preference.icc profile

Cliff

Reply   Reply with quote   Complain
Jack Hogan
Senior MemberPosts: 3,727Gear list
Like?
Re: Camera-->sRGB versus Camera-->ProPhoto-->sRGB
In reply to gollywop, May 1, 2013

gollywop wrote:

As to a choice between the bottom two, each has its value (but there is no question about which is the easier to produce).

Good show Gollywop. This is the link to your flower dng file and this is the accompanying text:

"This file is underexposed about 1EV according to RawDigger.  It was shot with UniWB (hence the greenish tint).  A WhiBal card shot in the same light indicates a WB (for ACR) of 4900/3, and that is what I used for the image as posted.  It is also clear that the image as posted was cropped."

I've been thinking about the wild guesstimates during conversion mentioned earlier, and I've found that they can be really apparent.  Here is an example: note the clipping in the right petal in ProPhoto-->PP-->sRGB compared to sRGB-->PP (I know that GIF's are not ideal, but they make animation simple - hope it works).  Here are the Original TIFFs which are much better for comparison: ProPhoto , sRGB .

If the animation works the two images should alternate every 4 seconds

Perhaps the reason is that adaptive editing programs are a little more sophisiticated in dealing with boundary conditions than color space conversions/CMSs, which I assume blindly apply their formulas.  Here is the path that a color on that petal took in xyY space:

1. Opened NEF in CNX2 either in ProPhotoRGB and in sRGB
2. Changed Picture Control from Neutral to Landscape
3. Applied CEP3 Tonal Contrast at default values
4. Converted ProPhotoRGB image to sRGB

Path of sample color during processing

And here is another sample color from the left side of the flower: this one was luckier than the next.

More space waks

Here is a sample color's trip on the Blue Ball image with the purple shift going from ACR 6.7 Standard through CEP3 Tonal Contrast - this one was doomed from the start.  It seems to me that the hue shift was introduced by ACR when it applied the color spaces originally.

It seems to me that despite the fact that I selected 'perceptual' intent, when programs/CMMs convert to sRGB they simply bring whoever is out to left field back to the edge.  While instead if you are working within the smaller space to start with, the editing programs are a bit more mindful about not hitting the ceiling.  What do you guys think?

Jack

Reply   Reply with quote   Complain
Jeff Schewe
Regular MemberPosts: 371
Like?
Re: Camera-->sRGB versus Camera-->ProPhoto-->sRGB
In reply to crames, May 1, 2013

crames wrote:

I guess you don't have to use a v4 source profile, but potentially the results are better if you do?

Don't know if there's any benefit...but v2 PP RGB to v4 sRG allows the use of both perceptual and relative colorimetric renderings. I've seen some improvement in using perceptual with some images–particularly saturated reds & blues.

-- hide signature --

Regards,
Jeff Schewe

Reply   Reply with quote   Complain
Jeff Schewe
Regular MemberPosts: 371
Like?
Re: Camera-->sRGB versus Camera-->ProPhoto-->sRGB
In reply to Jack Hogan, May 2, 2013

It seems to me that despite the fact that I selected 'perceptual' intent, when programs/CMMs convert to sRGB they simply bring whoever is out to left field back to the edge.

If you are working with device based matrix color spaces defined with v2 ICC profiles, there is no perceptual renderings–even if an app allows you to select it. It's always, only relative colorimetric.

If you want to use perceptual in a working space transform, you need to be using a v4 color space profile that HAS a real perceptual rendering intent. You'll want to get the sRGB_v4_ICC_preference_displayclass.icc profile available on the ICC web site. Otherwise, yes, a PP RGB>sRGB transform will always clip out of gamut color.

-- hide signature --

Regards,
Jeff Schewe

Reply   Reply with quote   Complain
Jack Hogan
Senior MemberPosts: 3,727Gear list
Like?
Re: Camera-->sRGB versus Camera-->ProPhoto-->sRGB
In reply to Jeff Schewe, May 2, 2013

Jeff Schewe wrote:

It seems to me that despite the fact that I selected 'perceptual' intent, when programs/CMMs convert to sRGB they simply bring whoever is out to left field back to the edge.

If you are working with device based matrix color spaces defined with v2 ICC profiles, there is no perceptual renderings–even if an app allows you to select it. It's always, only relative colorimetric.

If you want to use perceptual in a working space transform, you need to be using a v4 color space profile that HAS a real perceptual rendering intent. You'll want to get the sRGB_v4_ICC_preference_displayclass.icc profile available on the ICC web site. Otherwise, yes, a PP RGB>sRGB transform will always clip out of gamut color.

-- hide signature --

Regards,
Jeff Schewe

Ah, that's why.  Good to know, thanks Jeff.

Reply   Reply with quote   Complain
crames
Regular MemberPosts: 192
Like?
Re: Camera-->sRGB versus Camera-->ProPhoto-->sRGB
In reply to Jack Hogan, May 3, 2013

Jack Hogan wrote:

gollywop wrote:

As to a choice between the bottom two, each has its value (but there is no question about which is the easier to produce).

Good show Gollywop. This is the link to your flower dng file and this is the accompanying text:

"This file is underexposed about 1EV according to RawDigger.  It was shot with UniWB (hence the greenish tint).  A WhiBal card shot in the same light indicates a WB (for ACR) of 4900/3, and that is what I used for the image as posted.  It is also clear that the image as posted was cropped."

I've been thinking about the wild guesstimates during conversion mentioned earlier, and I've found that they can be really apparent.  Here is an example: note the clipping in the right petal in ProPhoto-->PP-->sRGB compared to sRGB-->PP (I know that GIF's are not ideal, but they make animation simple - hope it works).  Here are the Original TIFFs which are much better for comparison: ProPhoto , sRGB .

Is there a TIFF that hasn't been converted to sRGB and is still in ProPhoto?

Cliff

Reply   Reply with quote   Complain
Jack Hogan
Senior MemberPosts: 3,727Gear list
Like?
Re: Camera-->sRGB versus Camera-->ProPhoto-->sRGB
In reply to crames, May 3, 2013

crames wrote:

Is there a TIFF that hasn't been converted to sRGB and is still in ProPhoto?

Here is the original NEF and here is the equivalent TIFF in ProPhoto (opened in ProPhoto, Picture Control Landscape, CEP3 Tonal Contrast applied at default settings).

Cliff, the links to the two intermediate color spaces by jc mentioned in the LuLa thread no longer work.  I've sent you a PM with my email address.  Would you mind emailing them to me?

Jack

Reply   Reply with quote   Complain
xpatUSA
Senior MemberPosts: 3,007Gear list
Like?
Re: Camera-->sRGB versus Camera-->ProPhoto-->sRGB
In reply to Jack Hogan, May 17, 2013

Jack Hogan wrote:

crames wrote:

Is there a TIFF that hasn't been converted to sRGB and is still in ProPhoto?

Here is the original NEF and here is the equivalent TIFF in ProPhoto (opened in ProPhoto, Picture Control Landscape, CEP3 Tonal Contrast applied at default settings).

Jack

For what it's worth, I found a JPEG in ProPhoto of your flower up there and here's a 2D gamut diagram of it, courtesy of my new toy:

Lab color space, sRGB profile shown. Looks like a good candidate for perceptual intent?

-- hide signature --

Regards,
Ted http://kronometric.org
SD9, SD10, EF-500, GH1.

 xpatUSA's gear list:xpatUSA's gear list
Sigma SD9 Sigma SD10 Sigma SD14 Panasonic Lumix DMC-GH1 +2 more
Reply   Reply with quote   Complain
Iliah Borg
Forum ProPosts: 15,978
Like?
Re: Camera-->sRGB versus Camera-->ProPhoto-->sRGB
In reply to Detail Man, May 19, 2013

All raw converters ultimately have to take the demosaiced raw camera coordinates (RGB or GMCY) and map them to XYZ or Lab

Thу above is incorrect.

Reply   Reply with quote   Complain
Detail Man
Forum ProPosts: 15,000
Like?
Re: Camera-->sRGB versus Camera-->ProPhoto-->sRGB
In reply to Iliah Borg, May 19, 2013

All raw converters ultimately have to take the demosaiced raw camera coordinates (RGB or GMCY) and map them to XYZ or Lab, though it may be only an intermediate step. This is especially true for raw converters that use ICC profiles, where either XYZ or Lab D50 is the working space. The final mapping from XYZ to any of the standard RGB spaces is just a linear transform plus an encoding curve. The lookup tables provide a mechanism for non-linear corrections (e.g., if you need to tweak deep saturated reds differently than less-saturated reds, something the earlier matrix-only profiles obviously could not do).

- Eric Chan, Reply #51 on: July 28, 2008, 08:31:26 PM

http://www.luminous-landscape.com/forum/index.php?topic=26524.msg211284#msg211284

Iliah Borg wrote:

All raw converters ultimately have to take the demosaiced raw camera coordinates (RGB or GMCY) and map them to XYZ or Lab

Thу above is incorrect.

(Perhaps) one exception being RAW Photo Processor (RPP) ?

Reply   Reply with quote   Complain
RudivanS
Veteran MemberPosts: 4,302Gear list
Like?
Re: Camera-->sRGB versus Camera-->ProPhoto-->sRGB
In reply to Detail Man, May 19, 2013

Yes, RPP is my go-to when time allows - well worth a donation for full access.

RPP-Overview

-- hide signature --

Rudi - freelancer
'15 years in bladerunner Tokyo - back in sunny Sydney now'

 RudivanS's gear list:RudivanS's gear list
Nikon D4 Nikon D800E Canon Pixma Pro-10
Reply   Reply with quote   Complain
Iliah Borg
Forum ProPosts: 15,978
Like?
Re: Camera-->sRGB versus Camera-->ProPhoto-->sRGB
In reply to Detail Man, May 28, 2013

Detail Man wrote:

All raw converters ultimately have to take the demosaiced raw camera coordinates (RGB or GMCY) and map them to XYZ or Lab, though it may be only an intermediate step. This is especially true for raw converters that use ICC profiles, where either XYZ or Lab D50 is the working space. The final mapping from XYZ to any of the standard RGB spaces is just a linear transform plus an encoding curve. The lookup tables provide a mechanism for non-linear corrections (e.g., if you need to tweak deep saturated reds differently than less-saturated reds, something the earlier matrix-only profiles obviously could not do).

- Eric Chan, Reply #51 on: July 28, 2008, 08:31:26 PM

http://www.luminous-landscape.com/forum/index.php?topic=26524.msg211284#msg211284

Iliah Borg wrote:

All raw converters ultimately have to take the demosaiced raw camera coordinates (RGB or GMCY) and map them to XYZ or Lab

Thу above is incorrect.

(Perhaps) one exception being RAW Photo Processor (RPP) ?

The "have" part is totally wrong. Fantastically wrong.

Reply   Reply with quote   Complain
Detail Man
Forum ProPosts: 15,000
Like?
Re: Camera-->sRGB versus Camera-->ProPhoto-->sRGB
In reply to Iliah Borg, May 28, 2013

Iliah Borg wrote:

Detail Man wrote:

All raw converters ultimately have to take the demosaiced raw camera coordinates (RGB or GMCY) and map them to XYZ or Lab, though it may be only an intermediate step. This is especially true for raw converters that use ICC profiles, where either XYZ or Lab D50 is the working space. The final mapping from XYZ to any of the standard RGB spaces is just a linear transform plus an encoding curve. The lookup tables provide a mechanism for non-linear corrections (e.g., if you need to tweak deep saturated reds differently than less-saturated reds, something the earlier matrix-only profiles obviously could not do).

- Eric Chan, Reply #51 on: July 28, 2008, 08:31:26 PM

http://www.luminous-landscape.com/forum/index.php?topic=26524.msg211284#msg211284

Iliah Borg wrote:

All raw converters ultimately have to take the demosaiced raw camera coordinates (RGB or GMCY) and map them to XYZ or Lab

Thу above is incorrect.

(Perhaps) one exception being RAW Photo Processor (RPP) ?

The "have" part is totally wrong. Fantastically wrong.

Interesting. I wanted to attribute the statement to its author (Eric Chan). I wonder if any digital cameras exist that may take such approaches. You mentioned that in your understanding this might (at least in the case of the Olympus E-M5) be a typical in-camera processing scenario:

Iliah Borg wrote:

Yes. Essentially (simplifying and omitting any secrets of trade) what happens is:

Apply WB; run 1st stage NR and sharpening; perform demosaicking; convert from camera colour to XYZ; apply 2nd stage of sharpening on Y; convert to LCH; apply tone curve to L; apply saturation to C; apply "look" to H; apply 2nd stage of NR over L and C; smooth H; apply convert to output space (like sRGB or AdobeRGB).

http://www.dpreview.com/forums/post/42201474

Reply   Reply with quote   Complain
exdeejjjaaaa
Veteran MemberPosts: 6,728Gear list
Like?
Re: Camera-->sRGB versus Camera-->ProPhoto-->sRGB
In reply to Detail Man, May 29, 2013

Detail Man wrote:

Iliah Borg wrote:

Detail Man wrote:

All raw converters ultimately have to take the demosaiced raw camera coordinates (RGB or GMCY) and map them to XYZ or Lab, though it may be only an intermediate step. This is especially true for raw converters that use ICC profiles, where either XYZ or Lab D50 is the working space. The final mapping from XYZ to any of the standard RGB spaces is just a linear transform plus an encoding curve. The lookup tables provide a mechanism for non-linear corrections (e.g., if you need to tweak deep saturated reds differently than less-saturated reds, something the earlier matrix-only profiles obviously could not do).

- Eric Chan, Reply #51 on: July 28, 2008, 08:31:26 PM

http://www.luminous-landscape.com/forum/index.php?topic=26524.msg211284#msg211284

Iliah Borg wrote:

All raw converters ultimately have to take the demosaiced raw camera coordinates (RGB or GMCY) and map them to XYZ or Lab

Thу above is incorrect.

(Perhaps) one exception being RAW Photo Processor (RPP) ?

The "have" part is totally wrong. Fantastically wrong.

Interesting.

I think IB is nitpicking to certain extent on Eric... software developer certainly can avoid using XYZ or Lab for any internal postdemosaick operations and can avoid using any CMS transforms involving PCS = XYZ or Lab if he wants to (code something different all by himself) and Eric Chan for sure knows that but he was not writing that text for IB and so was not that careful w/ wording (or was carried away by what was/is generally accepted)

 exdeejjjaaaa's gear list:exdeejjjaaaa's gear list
Olympus E-M1 Sony Alpha 7 Panasonic Lumix G 20mm F1.7 ASPH Panasonic Leica DG Macro-Elmarit 45mm F2.8 ASPH OIS Panasonic Lumix G 14mm F2.5 ASPH +20 more
Reply   Reply with quote   Complain
Iliah Borg
Forum ProPosts: 15,978
Like?
Re: Camera-->sRGB versus Camera-->ProPhoto-->sRGB
In reply to exdeejjjaaaa, May 29, 2013

exdeejjjaaaa wrote:

Detail Man wrote:

Iliah Borg wrote:

Detail Man wrote:

All raw converters ultimately have to take the demosaiced raw camera coordinates (RGB or GMCY) and map them to XYZ or Lab, though it may be only an intermediate step. This is especially true for raw converters that use ICC profiles, where either XYZ or Lab D50 is the working space. The final mapping from XYZ to any of the standard RGB spaces is just a linear transform plus an encoding curve. The lookup tables provide a mechanism for non-linear corrections (e.g., if you need to tweak deep saturated reds differently than less-saturated reds, something the earlier matrix-only profiles obviously could not do).

- Eric Chan, Reply #51 on: July 28, 2008, 08:31:26 PM

http://www.luminous-landscape.com/forum/index.php?topic=26524.msg211284#msg211284

Iliah Borg wrote:

All raw converters ultimately have to take the demosaiced raw camera coordinates (RGB or GMCY) and map them to XYZ or Lab

Thу above is incorrect.

(Perhaps) one exception being RAW Photo Processor (RPP) ?

The "have" part is totally wrong. Fantastically wrong.

Interesting.

I think IB is nitpicking to certain extent on Eric... software developer certainly can avoid using XYZ or Lab for any internal postdemosaick operations and can avoid using any CMS transforms involving PCS = XYZ or Lab

And it is not nitpicking. It is a serious thing to consider.

if he wants to

Not the matter of what one wants but the matter of what is better.

Reply   Reply with quote   Complain
Iliah Borg
Forum ProPosts: 15,978
Like?
Re: Does Melissa "Gloss Over" Image Noise ?
In reply to Detail Man, May 29, 2013

and which is stated here by Eric Chan to precisely comport with that ProPhoto RGB specification, but has been measured by Bill Janes using Imatest to appear to not comport:

http://forums.adobe.com/message/4895109

He should have used a synthetic target instead.

Reply   Reply with quote   Complain
Keyboard shortcuts:
FForum MMy threads