Pages: 1 [2] 3   Go Down

Author Topic: Luminence value for web output  (Read 9065 times)

Simon Garrett

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 742
Re: Luminence value for web output
« Reply #20 on: February 23, 2015, 06:24:57 pm »

In LR/ACR this impression can result from the default S-curve which has a characteristic side effect on color saturation,
increasing it for the shadows and midtones while decreasing it for the highlights (insofar corresponding to the behavior of a RGB S-curve).
By varying the camera-exposure the data are initially placed under different sections of this curve.

Yes, that's certainly true.  I'd not really thought about that. 

I've read in several places (not just by Michael Freeman) the idea of the same saturation appearing lower in brighter images than darker images but perhaps it's not a physiological effect, rather it's an artefact of the way raw conversion works (whether in-camera or on a computer).  In other words: converting from the scene-referred raw image to an output-referred jpeg or screen rendering, in almost all cases an S-curve is applied, with the effect you describe. 
Logged

Tim Lookingbill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2436
Re: Luminence value for web output
« Reply #21 on: February 23, 2015, 09:47:13 pm »

Yes, that's certainly true.  I'd not really thought about that.  

I've read in several places (not just by Michael Freeman) the idea of the same saturation appearing lower in brighter images than darker images but perhaps it's not a physiological effect, rather it's an artefact of the way raw conversion works (whether in-camera or on a computer).  In other words: converting from the scene-referred raw image to an output-referred jpeg or screen rendering, in almost all cases an S-curve is applied, with the effect you describe.  

Don't forget to consider the influences of the Adobe Color Engine designed by Thomas Knoll that operates under the color management hood of both Adobe Raw and gamma encoded image editors. The contrast influenced hue/sat shift you're describing is a well known issue/feature? going back at least a decade when complainers were advocating a more Lab space or HSV? preview behavior when editing color. I remember reading Mr. Knoll designed the color engine to behave that way to emulate a more film like color appearance.

Try making that red ramp in Lab space and see how the steps change in appearance. Not sure but I don't think it will make that much of a difference because the preview still has to go through the ICC base color transform of the custom display profile.

I've often thought that what confuses me and most photographers who post process when describing how edited color previews behave is that we seem to think the contrast/brightness tools are suppose to behave like real lights on a real object or scene we've captured which I don't believe these tools are primarily designed to do. There's seems to be too much "English" designed into the algorithms that map the color to the displayed preview.

From my observations increase/decrease in exposure is a photographic concept which has no correlation to how a real light behaves due to the fact the original light that illuminated the original scene changes in character from scene to scene that no slider or curve can match. That's why we're given so many other tools to work with in order to create a match according to how we remember the character of light.
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20677
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Luminence value for web output
« Reply #22 on: February 24, 2015, 10:28:52 am »

Don't forget to consider the influences of the Adobe Color Engine designed by Thomas Knoll that operates under the color management hood of both Adobe Raw and gamma encoded image editors.
Actually do forget about it. Unless there's a bug in a CMM, the differences should be tiny (tests I've done were way less than a dE of 1 so invisible). Further, what ACE provides other CMM's may not is Black Point Compensation which really isn't a factor with this discussion.

Here is a 988 patches in Adobe RGB (1998) converted to sRGB using ACE and Apple CMM:

--------------------------------------------------

dE Report

Number of Samples: 988

Delta-E Formula dE2000

Overall - (988 colors)
--------------------------------------------------
Average dE:   0.02
    Max dE:   1.11
    Min dE:   0.00
 StdDev dE:   0.09

Best 90% - (888 colors)
--------------------------------------------------
Average dE:   0.00
    Max dE:   0.00
    Min dE:   0.00
 StdDev dE:   0.00

Worst 10% - (100 colors)
--------------------------------------------------
Average dE:   0.16
    Max dE:   1.11
    Min dE:   0.00
 StdDev dE:   0.25

--------------------------------------------------
--------------------------------------------------
SINGLE worst patch and ONLY one over a dE of one from all 988 color patches (RGB and Lab) values:
2.0   7.0   3.0   1.55   -1.82   1.07         
0.0   7.0   1.0   1.40   -2.59   1.64         
dE 1.11   

Basically forget about it Tim.
« Last Edit: February 24, 2015, 10:32:32 am by digitaldog »
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Tim Lookingbill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2436
Re: Luminence value for web output
« Reply #23 on: February 24, 2015, 02:00:22 pm »

Andrew, did you apply the same tests in other non-Adobe Raw converters which use a different color engine? (i.e. Iridient Developer, Capture One, etc.)? There is a difference unless things have changed in PV2012 with regards to preview behavior increasing and decreasing contrast in edits AND white balance adjustments in how blue/yellow & green/magenta hues and saturation levels are interpreted which also compounds the issue editing landscapes and interiors, not static patches of color in a profiling/calibration session.

This is a preview behavior not readily seen in controlled tests. Have you tried changing the contrast in a scene referred vs output referred landscape to notice a difference to hue/sat changes? Photographers don't use science to edit images, they use their emotions.

Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20677
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Luminence value for web output
« Reply #24 on: February 24, 2015, 03:33:07 pm »

Andrew, did you apply the same tests in other non-Adobe Raw converters which use a different color engine? (i.e. Iridient Developer, Capture One, etc.)?
ACE isn't available so it's moot. I did the conversions in Photoshop where switching from ACE to whatever other CMM's are available is accessible. But again, unless there is a bug with a CMM, the differences should and are tiny! So again, Actually do forget about it.
Quote
There is a difference unless things have changed in PV2012
That may be so, it's not ACE. That's not how CMM's work or are designed to work Tim.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20677
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Luminence value for web output
« Reply #25 on: February 24, 2015, 03:54:09 pm »

Photographers don't use science to edit images, they use their emotions.
Maybe that's why they think and report they see things that don't exist, like differences in CMM's. Colorimetry as shown below proves what they think they see doesn't exist.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Tim Lookingbill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2436
Re: Luminence value for web output
« Reply #26 on: February 24, 2015, 10:08:13 pm »

ACE isn't available so it's moot. I did the conversions in Photoshop where switching from ACE to whatever other CMM's are available is accessible. But again, unless there is a bug with a CMM, the differences should and are tiny! So again, Actually do forget about it. That may be so, it's not ACE. That's not how CMM's work or are designed to work Tim.

Can you provide proof because I read from the horses mouth Thomas Knoll that he designed the underpinnings that affect saturation in previews when editing contrast in order to render with a film like look in the very issue that is being discussed here and many years ago. He said he designed the color engine but I can't remember if he was referring to ACE which stands for Adobe Color Engine. Not CMM=Color Management Module. I'm not talking about color management here. I'm talking about contrast induced saturation changes when editing images.

So could you provide evidence that this behavior is not on account of TK's design of how edited previews behave this way? Maybe it's not called ACE but something else and I just assumed back then that ACE was where the design resided.
« Last Edit: February 24, 2015, 10:12:12 pm by Tim Lookingbill »
Logged

GWGill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 608
  • Author of ArgyllCMS & ArgyllPRO ColorMeter
    • ArgyllCMS
Re: Luminence value for web output
« Reply #27 on: February 25, 2015, 02:44:51 am »

He said he designed the color engine but I can't remember if he was referring to ACE which stands for Adobe Color Engine. Not CMM=Color Management Module.
ACE = Adobe's ICC CMM

But a CMM would be just one small part of an application like PhotoShops color code.
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20677
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Luminence value for web output
« Reply #28 on: February 25, 2015, 08:59:51 am »

Can you provide proof because I read from the horses mouth Thomas Knoll that he designed the underpinnings that affect saturation in previews when editing contrast in order to render with a film like look in the very issue that is being discussed here and many years ago.
Tim, I'm afraid you don't know what you are taking about! First off, ACE IS a CMM! It is accessible from the CMM dropdown in PS itself along with others. Been that way since 1998.

I proved that there's tiny, tiny differences between it and Apple's CMM below using good old colorimetry. Of 988 colors, only one is 0.1 dE over a value of 1, 987 patches are invisible in terms of their differences. Further, of the 988 patches, about 800+ are 0.00 dE apart. Care to explain that?

2nd, if you understand the actual role of ACE as a CMM, you'd know you are being silly suggesting it's the causes of any numeric or visual differences from any other CMM. It would be just chaotic if using the same data, profiles and software (or different software), a conversion would be much different solely due to a CMM. UNLESS that CMM has a bug.

3rd, it is you sir that needs to prove your point which falls in the face of logic based on the above by stating: Don't forget to consider the influences of the Adobe Color Engine designed by Thomas Knoll that operates under the color management hood of both Adobe Raw and gamma encoded image editors. You have as yet no means to back that up yet you've got the balls to ask me for proof after I did so with a very simple test that shows ACE and Apple CMM with the same profiles and data produce colorimetrically identical visual output from the two conversions? You serious bud?

What ever Thomas told you, you obviously didn't understand. It's simply ridiculous for someone who thinks he understands the role and design of a CMM to make the point you did and you've so far have nothing to back it up.
« Last Edit: February 25, 2015, 09:11:51 am by digitaldog »
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20677
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Luminence value for web output
« Reply #29 on: February 25, 2015, 12:30:51 pm »

Oh Tim, here's a report converting from Adobe RGB to a print output color space (rather than to sRGB) using two different CMM's again. In this case NONE of the 988 color patches are visibly different!

--------------------------------------------------

dE Report

Number of Samples: 988

Delta-E Formula dE2000

Overall - (988 colors)
--------------------------------------------------
Average dE:   0.12
    Max dE:   0.85
    Min dE:   0.00
 StdDev dE:   0.15

Best 90% - (888 colors)
--------------------------------------------------
Average dE:   0.09
    Max dE:   0.31
    Min dE:   0.00
 StdDev dE:   0.11

Worst 10% - (100 colors)
--------------------------------------------------
Average dE:   0.41
    Max dE:   0.85
    Min dE:   0.31
 StdDev dE:   0.12

--------------------------------------------------
--------------------------------------------------
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Tim Lookingbill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2436
Re: Luminence value for web output
« Reply #30 on: February 25, 2015, 01:39:01 pm »

ACE = Adobe's ICC CMM

But a CMM would be just one small part of an application like PhotoShops color code.

I was understanding it that way as well but I seem to not communicate effectively enough to Andrew. The behavior of edits within a color managed preview (that involves transforms and connection spaces when using editing tools to translate to the video system through ICC display profile) has to be tracked/encoded/described in order for the CMM to commit it to the final tagged image in order for the preview to match across other CM apps.

My question is at what stage does Photoshop/ACR/LR color coding and whatever else is rendering the previews influences this contrast induced saturation behavior shown in the red ramps and other similar color preview mapping and hands it off to the CMM and if the CMM has any transform/preview controlling influences.

Andrew, to keep it simple could you explain simply what is making the darker reds appear more saturated and light reds less saturated in the red ramp posted here?
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20677
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Luminence value for web output
« Reply #31 on: February 25, 2015, 02:46:13 pm »

I was understanding it that way as well but I seem to not communicate effectively enough to Andrew.
Your communication was clearly wrong and understood by more than just me! ACE is a CMM. ACE has absolutely nothing to do with what you are referring to or understood from Thomas (still clearly undefined but you're quite sure it's spot on).
Quote
The behavior of edits within a color managed preview (that involves transforms and connection spaces when using editing tools to translate to the video system through ICC display profile) has to be tracked/encoded/described in order for the CMM to commit it to the final tagged image in order for the preview to match across other CM apps.
Yes and this has nothing to do with ACE.
Quote
Andrew, to keep it simple could you explain simply what is making the darker reds appear more saturated and light reds less saturated in the red ramp posted here?
It has nothing to do with the CMM named ACE. The statement: Don't forget to consider the influences of the Adobe Color Engine designed by Thomas Knoll that operates under the color management hood of both Adobe Raw and gamma encoded image editors. is simply incorrect. There's all kinds of proprietary image processing going on under the hood, about the only human here on LuLa what could know and explain it is Eric and that's unlikely to happen. Whatever those differences are, they have nothing to do with ACE as used as a CMM to convert color data. Got it?
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Tim Lookingbill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2436
Re: Luminence value for web output
« Reply #32 on: February 25, 2015, 04:41:18 pm »

Quote
Whatever those differences are, they have nothing to do with ACE as used as a CMM to convert color data. Got it?

I got it you don't know and you don't have any evidence to prove what you're saying. But I do know you haven't been helpful in explaining the behavior of the red ramp saturation change. Got it?
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20677
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Luminence value for web output
« Reply #33 on: February 25, 2015, 05:46:15 pm »

I got it you don't know and you don't have any evidence to prove what you're saying. But I do know you haven't been helpful in explaining the behavior of the red ramp saturation change. Got it?
Now you're playing the fool and deserve an academy award.
You made a false and rather ridiculous statement about ACE. Both Graeme and I had to point out that ACE is a CMM. I don't know if you even understand what ACE or CMM's provide but the point is, I easily dismissed the incorrect post you made:
Quote
Don't forget to consider the influences of the Adobe Color Engine designed by Thomas Knoll that operates under the color management hood of both Adobe Raw and gamma encoded image editors.
Quote
He said he designed the color engine but I can't remember if he was referring to ACE which stands for Adobe Color Engine. Not CMM=Color Management Module.
ACE has no such influence. I proved that using colorimetry and a process anyone here can replicate and prove to themselves. Meanwhile and as your usual M.O. you have provided NO such evidence of what you incorrectly wrote. So telling! The red ramp is your red herring. It has zero to do with ACE.

If you were as smart as I used to think you were Tim, you could have said something to the effect of: "Andrew, I must have confused ACE with something else" and accepted the facts I've provided about ACE, CMM's and their tiny differences on the data they touch. But no, you have to continue looking foolish to others here by ignoring the colorimetric facts, the facts of what a CMM should be doing and then going down a red ramp rabbit hole that isn't getting you off the hook for a statement you made based on severe misunderstanding of CMMs.

You may have thought you heard (or read, not sure) from the horses mouth (Thomas Knoll) that the sun rises in the west and sets in the east but that's simply untrue. When shown the error of this mistake, you ask about the color temp of the sun before it sets, ignoring it never sets in the east. That's foolish but you are welcome to act that way and ignore the facts. So far you are unable or unwilling to backup the statements you've made about ACE. While asking me to back up their facts on something that has nothing to do with ACE. Again, very telling.  

I don't know nor care at this point about the red ramp. We're not ready to discuss that until you either admit you're wrong about ACE or except what you incorrectly think about it and the setting sun is factual. Neither are. Got it?

Why you refuse to make this a learning experience for yourself with respect to ACE and your incorrect statements is also a shame and rather telling.
« Last Edit: February 26, 2015, 04:20:01 am by digitaldog »
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Tim Lookingbill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2436
Re: Luminence value for web output
« Reply #34 on: February 25, 2015, 11:25:53 pm »

I've spoken my peace and counted to three. We will have no more talk on this matter, Andrew Jackson Rodney.
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20677
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Luminence value for web output
« Reply #35 on: February 26, 2015, 04:24:33 am »

I've spoken my peace and counted to three. We will have no more talk on this matter, Andrew Jackson Rodney.

Being ignorant is not so much a shame, as being unwilling to learn.”
― Benjamin Franklin

Yes Tim, probably best to ignore the facts about ACE and run away. Hopefully in terms of this subject in relationship to ACE the CMM, all other readers here understand the facts and your inability to learn and accept them. Useful for your future postings on the subject. Good-by.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Tim Lookingbill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2436
Re: Luminence value for web output
« Reply #36 on: February 26, 2015, 12:17:15 pm »

You're the one that can't explain effectively the saturation behavior in the red ramp demo so I'ld say you're just as ignorant and I'll add quite tone deaf to humor.
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20677
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Luminence value for web output
« Reply #37 on: February 26, 2015, 12:32:36 pm »

You're the one that can't explain effectively the saturation behavior in the red ramp demo so I'ld say you're just as ignorant and I'll add quite tone deaf to humor.
That has zero to do with your erroneous statements about ACE. I know it's tough to look in the mirror Tim and admit you made a mistake and learn from it! Benjamin summed up your attitude perfectly.

I can't explain what took place prior to the big bang, but that doesn't alter the proof I provided that dismissed the two sentences you wrote about ACE. Two wrongs don't make a right. One unexplained process doesn't defend a severe miss understanding proposed by Mr Tim Lookingbill. That you continue to focus on a red ramp while ignoring a mistake you made about a CMM isn't going to disappear Tim. Get your facts straight if you can.

Now, as to what you think you heard or think understood from Thomas, you are so vague and confused I'm not sure what you are scrabbling about. I can tell share this:

Quote
A new message was posted by Thomas Knoll in

Adobe Camera Raw --
  Color reproduction in digital photography

While developing Camera Raw, I experimented with a pure luminance curve (as Simon suggests). However, based on my testing results, I rejected this algoirthm since it produced results that were most often visually worse looking that the tone curve algorithm actually used by Camera Raw (which is a special hue-preserving curve, NOT three indepent curves as Simon incorrectly assumed). The saturation effects that Simon considers a defect is actually something that most users actually want.

You have a short memory Tim: http://forum.luminous-landscape.com/index.php?topic=72369.new;topicseen
But again, this has nothing to do with your misinformation about the CMM Thomas built. Got it?
« Last Edit: February 26, 2015, 12:46:40 pm by digitaldog »
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20677
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Luminence value for web output
« Reply #38 on: February 26, 2015, 12:42:21 pm »

We will have no more talk on this matter, Andrew Jackson Rodney.
And yet, you keep on talking. You're not even good to your own word Tim.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Peter_DL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 544
Re: Luminence value for web output
« Reply #39 on: February 26, 2015, 04:59:36 pm »

Many thanks Andrew,

- for clearing the way
so that we can return to the OP’s question which I for one find interesting:


Ok, I know that this is mostly an unanswerable question but I gota at least try.
Think of it as the photographic equivalent of a controlled crash landing.
Does anyone have any thoughts about what to set  your luminance target for your display in preparing files for websites?

I am calibrating my monitors with  i1 software & hardware.
This is what I set up for targets...
White point = D65
Luminance = 120
 
To quote from the whitepaper which I had referenced earlier:

>> Transforming scene-referred data with a 100,000:1 dynamic range into a print is a complex and subjective process. Any two viewers will disagree on what looks best. Transforming a 400:1 output-referred image to 300:1 or 200:1 is pretty straightforward and most people will be happy with a simple fixed method that is the same for all their photographs. Although the photographic industry still has not fixed a universal standard, most JPEGs and other output-referred images have a target dynamic range of about 400:1. By design, this also compares favorably with computer displays, so the images look correct on your screen as well. ... <<

… which makes me believe that there must be a kind of standard for the white luminance [cd/m2],
e.g. with the monitors in the labs of a camera manufacturer, where the engineers work on the algorithms for in-camera-JPEG conversion.  At least that's what I would assume that happens.

So following the logic of the above quote, when prints correspond to an avg. 250:1 dyn. range,
and assuming that ca. 125 cd/m2 on screen work for print-image-matching,
it makes a black luminance of 0.5 cd/m2,

and with a 400:1 monitor dyn. range,
it finally points to a white luminance of 200 cd/m2.

A lot of assumptions here as I have to admit, maybe too simplifying (?),
however, in practice it corresponds (roughly) to my monitor's output and brightness setting for the given purpose.

Peter

--
« Last Edit: February 26, 2015, 05:07:09 pm by Peter_DL »
Logged
Pages: 1 [2] 3   Go Up