Pages: 1 ... 15 16 [17] 18 19 ... 24   Go Down

Author Topic: Color management myths and misinformation video  (Read 91677 times)

digitaldog

  • Sr. Member
  • ****
  • Online Online
  • Posts: 20944
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Color management myths and misinformation video
« Reply #320 on: September 03, 2014, 05:44:43 pm »

Andrew, you are going off-topic here, but since you opened the door, I'll play.
Since it's my topic, I'm OK with that  ;D
Quote
Thanks, by the way, for the provided links, as they prove my point, rather than debunk it.
I never disputed anything about his wife's skill's but thanks for adding data points that are not necessary nor up to dispute. The part about him failing math was very much up to dispute. Again, you seem to have difficulty separating these differing items out when going OT and disagreeing.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

digitaldog

  • Sr. Member
  • ****
  • Online Online
  • Posts: 20944
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Color management myths and misinformation video
« Reply #321 on: September 03, 2014, 05:51:30 pm »

I wonder what the kinks in the PP lines are about...
Yeah, that is quite odd! But I think at this point, we can probably move away from using CT and it's gamut volume metric to get to the bottom of the gamut's number of anything. And in defence of this fine product, the email I provided once again did states clearly it isn't intended for this analysis in the first place.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Jim Kasson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2370
    • The Last Word
Re: Color management myths and misinformation video
« Reply #322 on: September 03, 2014, 07:12:55 pm »

I agree. So the question is, should the video be aimed at beginners?

If you're up for it, sure. That makes it a lot harder, and forces the scope to be carefully managed, but it's a worthy objective to explain gamuts and color space conversions to beginners. I'm not sure the "myths" approach is the best way to do that, since it perforce requires you to explain why each myth exists, but, if it works, go for it, but select the myths for their pedagogical value, not their prevalence.

When I worked at IBM Almaden Research Center in the 90s, there was a tradition that some project leaders -- even if a project had only one worker -- would give a one-hour lecture to the Research staff on their research interest. These talks were open to the public, and were often attended not only by IBMers, but by people from Stanford, Berkeley, hp, and PARC. Presenters worked hard on their lectures, and the expectation bar was thus quite high. About two years after I decamped Rolm ahead of the invasion from Siemens and set up shop at ARC, my boss (or the closest thing I had to a boss -- when you’re an IBM Fellow it gets hazy) asked me to do a pitch on my color work. I spent a lot of time using InDesign [Edit: Micrografx Designer. I now see that there's a Corel program that will read the files, so maybe I can get the graphics up on the web someday] and preparing foils (yes, actual physical overhead projector sheets – IBM wasn’t modern about all things). I figured that no one in my audience would know much about color, but that my life would be made easier because they’d all be very smart, and, on average, have math skills that put mine to shame. That turned out to be wrong. Not the smart and mathematically sophisticated part, but the color-newbie part. Just before the lights went down Efi Arazi and four or five of his staff walked in and sat down in the back row. It was then I knew that I wouldn’t be able to get away with a thing.

Here’s an outline of my presentation. I know it’s not appropriate for the audience you’re talking about, but it does go from assuming nothing about color to a reasonable understanding of what color management (what we called in those days “device-independent color”) is all about. It may give you some ideas for your own presentation. I still have the Designer files, but I can’t figure out how to print them or turn it into, say, Adobe Illustrator files, so you’ll have to use your imagination.[see above addition]

First a statement of the problem.

I showed a diagram of an image capture and reproduction system, with a natural scene, a camera, storage, emissive display showing the scene, and a printer showing the scene. Each of the three images has a viewer, and each viewer is exposed to a set of – thus far undefined -- viewing conditions. The objective is for all three scenes to “look the same”.

I showed a similar diagram with a synthetic image, and displays in various locations. The objective is for all scenes to “look the same”.

Then I showed a block diagram of the then-prevalent method of managing color, where the colorants for the output device are determined at time of capture, and contrasted that to the now-prevalent model, where data is converted upon capture to device-independent form, and each output device is associated with software that converts the colors in the file to colorants appropriate to the device and the viewing conditions.

I showed an illustration of a natural scene, and made the point that the spectra observed by a viewer or a camera are the wavelength-by-wavelength product of the illuminant and the reflectivity of the object in the scene.

Then I showed a diagram of the eye, with the main elements identified. I explained the basic properties.

Another eye diagram, this one with the four types of light sensitive retinal elements identified, giving me a chance to talk about how their densities vary with retinal location and to remark in passing on the relative deficiency in the number of blue cone cells.

Then I showed the response curves of all three types of cone cells versus spectral excitation.

I’m not sure why I did this, except that it has always fascinated me as great adaptation, but I showed a diagram indicating the longitudinal chromatic aberration of the single-element lens in the eye, and showing how having the rho and gamma (although I didn’t use those names) cone cells spectral response so nearly alike minimized the issues associated with the LCA. I even showed text in various colors against colored backgrounds, showing that our visual acuity varies with color. I wouldn’t do this today.

I explained the Trichromatic Principle, and credited Le Blon and Maxwell.

Then I jumped right into a diagram of the color matching experiment, and explained its history and how it worked, introducing the terminology “tristimulus values” and impressing the audience with the fact that, presented with a sample color, 93% of the men and almost all the women set the knobs in about the same place. I showed how projecting colors on the sample side of the screen allowed all colors to be matched.

I showed a graph of the normalized color matching functions versus the wavelength of spectral stimuli, and made the point that color matching was virtually linear, a point that was not lost on this audience. That means that additivity applies, and color matching functions can be used as weighting functions to determine knob settings to match any color whose spectrum is known. I showed a graph of the unnormalized color matching functions versus the wavelength of spectral stimuli, and showed how they added to the photopic luminous spectral efficiency curve.

Because color matching is linear, any linear transformation of the color matching functions carries the same information as the functions themselves. I showed a graph of one interesting linear transform, CIE XYZ, versus wavelength of a spectral stimulus.

Then I showed a two dimensional projection of XYZ, xy, together with the conversion equations and spectral locus.

I showed another xy horse shoe with lines and points showing the “center of gravity” rule for calculating the chromaticity of mixed colors.

I plotted the primaries of an arbitrary CRT on the xy diagram, and showed how repeated applications of the center of gravity rule allowed any chromaticity in the triangle defined by the primaries to be created.

I expanded the primaries as far as they’d go inside the horse shoe, and showed that you couldn’t create all visible chromaticities with positive amount of any set of such primaries, although you could if you allowed negative amounts, such as was done in the color matching experiment.

Then I showed MacAdam’s ellipses plotted in xy, and remarked on how xy emphasized greenish differences and deemphasized bluish ones when compared to the human eye.

I introduced u’v’, gave the equations used for conversion from xy, and plotted the ellipses there.

Moving back into three dimensions, I defined Brightness, Hue, and Colorfullness, showing how they worked on the Munsell Tree, which unfortunately wasn’t visible to everyone due to the size of the room.

Since everyone was already up to speed with u’v’, it wasn’t much of a stretch to move to CIEL*u*v*. I showed them in visual form how that worked. I showed them how to calculate color differences, hue angle, and chroma.

I showed them the math to get from XYZ to CIEL*a*b*, and apologized for the heuristic nature of that space. I pointed out that Luv had one ad hoc moment in its derivation: the transition from xy to u’v’. I showed them how to calculate color differences, hue angle, and chroma.

I compared Lab and Luv. I think I was successful in not exhibiting my preference for Luv.

I showed a list of other important color spaces.

A talked about computational issues in color space choice. These issues seem quaint today, since they were predicated on limited computational resources and low bit depth..

In a foil titled: “We’re not done yet!”, I showed an optical illusion: you’ve seen it: low chroma middle grey yellowish X against a saturated yellow background and the same X against a darkish grey background, with the two X’s connected to prove that they are indeed the same color.

That gave me the opening to talk about viewing conditions, which I mostly ducked. I mentioned viewer adaption to surround and white point, the perception of self-luminosity, image size, and absolute brightness.

Leaving the hard stuff for (much. much) later, I returned to the color reproduction system diagram, talked about the interaction of the spectra in the field of a camera and the camera’s primary sensitivities. I introduced the concepts of illuminant metamerism and capture metamerism. I also introduced the idea of device gamut (for output devices only; the topic is the gamut of input devices is way too hard to get your head around), and made the point that gamut mapping is not an option. If you don’t do it, the device will.

I gave a survey of common gamut mapping algorithms. This was pretty easy to do since the audience understood Luminance/Hue/Chroma color spaces by now. I discussed smart clipping versus compression for OOG colors. I introduced the idea of neighborhood gamut mapping, which I later turned into a useful algorithm.

A lot to cover in an hour, and I elided the whole viewing conditions discussion, but from the questions, I seem to have gotten most of the ideas across.

Your problem is harder than mine was, Andrew, but some of this might help. The progression from the color matching experiment to Lab or Luv (it’s more elegant going to Luv, but you’ll want to take them to Lab) could be a useful introduction to 3D gamut maps.
« Last Edit: September 03, 2014, 11:39:28 pm by Jim Kasson »
Logged

digitaldog

  • Sr. Member
  • ****
  • Online Online
  • Posts: 20944
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Color management myths and misinformation video
« Reply #323 on: September 03, 2014, 07:25:40 pm »

Wow, thanks for taking the time to outline this Jim! I have a lot to think about and chew on. This may end up being a series, most likely more than one piece.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Tony Jay

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2965
Re: Color management myths and misinformation video
« Reply #324 on: September 03, 2014, 07:37:46 pm »

Andrew, as a follow up to Jim's post.

IMHO the level at which to pitch this is simple.

It needs to pitched at photographers who need a solid working understanding of colour management in their workflows.
A lot of us print, some don't, nonetheless we all need to make informed decisions in our workflow that require a good understanding of colour management.
A good discriminator for what to include is: does knowing this or that directly help one in making this or that workflow decision?
If it does - its in, if it doesn't - its out.

my 0.02c worth.

Tony Jay
Logged

Slobodan Blagojevic

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 18128
  • When everyone thinks the same, nobody thinks
    • My website
Re: Color management myths and misinformation video
« Reply #325 on: September 03, 2014, 07:54:34 pm »

... Adobe RGB doesn't have more color than sRGB. It has different colors...

Ok, let's work with that. Are you saying that, because they have different colors, Adobe RGB can not display colors that sRGB can?

digitaldog

  • Sr. Member
  • ****
  • Online Online
  • Posts: 20944
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Color management myths and misinformation video
« Reply #326 on: September 03, 2014, 07:55:00 pm »

my 0.02c worth.
Good points, thanks. It will help focus the presentation and I agree, it has to help folks make informed decisions. It will have go cover how to test this stuff too, from their own end. It doesn't matter what color space they pick, as long as they make the choice based on sound information and proper testing methodology.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Jim Kasson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2370
    • The Last Word
Re: Color management myths and misinformation video
« Reply #327 on: September 03, 2014, 07:55:14 pm »

Wow, thanks for taking the time to outline this Jim! I have a lot to think about and chew on. This may end up being a series, most likely more than one piece.

One thing I didn't talk about except in passing wrt the nonlinearity of the L* axis in Lab and Luv, was Weber's Law and the nonlinear nature of human visual difference perception. It's hard to fit in, but is important in that it explains not only L*, but why gamma corrected RGB encodings are efficient. But in an introductory video or videos, maybe you can duck that issue entirely.

Jim

Slobodan Blagojevic

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 18128
  • When everyone thinks the same, nobody thinks
    • My website
Re: Color management myths and misinformation video
« Reply #328 on: September 03, 2014, 07:55:35 pm »

... A good discriminator for what to include is: does knowing this or that directly help one in making this or that workflow decision?...

+1

digitaldog

  • Sr. Member
  • ****
  • Online Online
  • Posts: 20944
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Color management myths and misinformation video
« Reply #329 on: September 03, 2014, 07:56:11 pm »

Are you saying that, because they have different colors, Adobe RGB can not display colors that sRGB can?
That isn't what I'm saying, no. The question you ask has been previously answered and I believe more than one.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

MarkM

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 428
    • Alaska Photographer Mark Meyer
Re: Color management myths and misinformation video
« Reply #330 on: September 03, 2014, 08:42:50 pm »

Ok, let's work with that. Are you saying that, because they have different colors, Adobe RGB can not display colors that sRGB can?

Again depends on how you are defining color. sRGB can encode finer gradations of color than AdobeRGB can. Several pages back I gave an example. In that example sRGB (1, 255, 240) and (2, 255, 240) both map to the same AdobeRGB color.

If you are defining color differences purely in terms of visual distinctions, then these are the same color (at least to my eyes). But if you are going by the numbers, the distinction between these two color values is one AdobeRGB can't make but sRGB can.
Logged

fdisilvestro

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1862
    • Frank Disilvestro
Re: Color management myths and misinformation video
« Reply #331 on: September 03, 2014, 08:53:49 pm »

Again depends on how you are defining color. sRGB can encode finer gradations of color than AdobeRGB can. Several pages back I gave an example. In that example sRGB (1, 255, 240) and (2, 255, 240) both map to the same AdobeRGB color.

If you are defining color differences purely in terms of visual distinctions, then these are the same color (at least to my eyes). But if you are going by the numbers, the distinction between these two color values is one AdobeRGB can't make but sRGB can.

Increase bit depth and you will be able to do it. This is an encoding problem, not a limitation of the color space per se, but I think we have covered that before

Regards,

MarkM

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 428
    • Alaska Photographer Mark Meyer
Re: Color management myths and misinformation video
« Reply #332 on: September 03, 2014, 08:57:15 pm »

Increase bit depth and you will be able to do it. This is an encoding problem, not a limitation of the color space per se, but I think we have covered that before

Yes, but if you increase the bit depth, there will be still finer gradations that sRGB can encode that AdobeRGB cannot at the new bit depth. Of course we are already at the point where the gradations of meaningless in practice. This is a purely theoretical point. 
Logged

digitaldog

  • Sr. Member
  • ****
  • Online Online
  • Posts: 20944
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Color management myths and misinformation video
« Reply #333 on: September 03, 2014, 09:01:26 pm »

Increase bit depth and you will be able to do it.
Not in terms of Mark's example. Take 16-bit sRGB using (1, 255, 240) and (2, 255, 240) both map to the same AdobeRGB color: 144/255/240.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

MarkM

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 428
    • Alaska Photographer Mark Meyer
Re: Color management myths and misinformation video
« Reply #334 on: September 03, 2014, 09:18:03 pm »

Not in terms of Mark's example. Take 16-bit sRGB using (1, 255, 240) and (2, 255, 240) both map to the same AdobeRGB color: 144/255/240.

Since photoshop reports 8-bit values even for 16-bit images, we can't really tell by looking at the info pallet because it's reporting rounded values back to us.
Logged

digitaldog

  • Sr. Member
  • ****
  • Online Online
  • Posts: 20944
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Color management myths and misinformation video
« Reply #335 on: September 03, 2014, 09:27:03 pm »

Since photoshop reports 8-bit values even for 16-bit images, we can't really tell by looking at the info pallet because it's reporting rounded values back to us.
FWIW:
In 16-bit, using the example, ColorThink reports them as 144.1/255.0/240.2 and 144.2/255/240.2.
In 8-bit, using the example, ColorThink reports  them  as 144.0/255.0/241.0 for both.
Save both out as a color list for CT for it's dE report. On is 0.06 dE, the other 0.24. That's using dE2000. As such, I think we have to agree, they are the same color.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

MarkM

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 428
    • Alaska Photographer Mark Meyer
Re: Color management myths and misinformation video
« Reply #336 on: September 03, 2014, 09:37:50 pm »

FWIW:
In 16-bit, using the example, ColorThink reports them as 144.1/255.0/240.2 and 144.2/255/240.2.
In 8-bit, using the example, ColorThink reports  them  as 144.0/255.0/241.0 for both.
Save both out as a color list for CT for it's dE report. On is 0.06 dE, the other 0.24. That's using dE2000. As such, I think we have to agree, they are the same color.

No argument there, they were chosen because they were very close together.

For what it's worth you can get Photoshop to report 16 bit values in the info pallet. Doing this reveals the two colors in sRGB as:
(129, 32768, 30840) and (257, 32768, 30840)

On conversion to AdobeRGB you have:
(18520, 32768, 30864) and (18527, 32768, 30864)

So it does make the distinction in 16 bits. That's a lot of angles on this particular pin.
Logged

digitaldog

  • Sr. Member
  • ****
  • Online Online
  • Posts: 20944
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Color management myths and misinformation video
« Reply #337 on: September 03, 2014, 09:46:38 pm »

So it does make the distinction in 16 bits. That's a lot of angles on this particular pin.
Yes indeed. But what Photoshop can't give us the dE differences so we can suggest there is a color difference we can see (and call them two colors). More proof that this 'color number' rabbit hole is messy.
Here's a fun one too. Take ProPhoto RGB and mess with high number of blue, like 255/0/0 and 250/9/9 and convert them to Adobe RGB let alone sRGB. Same values. Perhaps that's why we see blobs of colors (If that's a fair term), lack of color details as I see on my prints from the Gamut Test file going from ProPhoto to Epson RGB vs. sRGB to Epson RGB.
« Last Edit: September 03, 2014, 09:49:18 pm by digitaldog »
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

fdisilvestro

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1862
    • Frank Disilvestro
Re: Color management myths and misinformation video
« Reply #338 on: September 03, 2014, 10:48:42 pm »

Yes indeed. But what Photoshop can't give us the dE differences so we can suggest there is a color difference we can see (and call them two colors). More proof that this 'color number' rabbit hole is messy.
Here's a fun one too. Take ProPhoto RGB and mess with high number of blue, like 255/0/0 and 250/9/9 and convert them to Adobe RGB let alone sRGB. Same values. Perhaps that's why we see blobs of colors (If that's a fair term), lack of color details as I see on my prints from the Gamut Test file going from ProPhoto to Epson RGB vs. sRGB to Epson RGB.

That is expected. If you plot the 3D gamuts of both Adobe RGB and ProphotoRGB, you will see that there is a small surface in the blues that is shared between both color spaces.

Actually, if you don't do it the right way, which is not performing the chromatic adaptation of Adobe RGB to D50 before comparing it to Prophoto RGB (which is native in D50), you will see a region from Adobe RGB in the blues which is out of the gamut of Prophoto RGB

 

Rhossydd

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3381
    • http://www.paulholman.com
Re: Color management myths and misinformation video
« Reply #339 on: September 04, 2014, 02:09:44 am »

I can only speak for myself. I refused to take Gary's bait based on his absolutely silly described testing methodology.
In the process of that flounce, you just confirmed his prejudices.
Quote
Again, that's your opinion, I frankly found it quite different. The person who continued to look sillly was Gary (and anyone defending him) based on his flat earth color theories
If you weren't so personally involved and looked at that thread independently, you'd see a lot of people played into his hands by refusing to agree on anything so proving his point.
Quote
Adobe RGB doesn't have more color than sRGB. It has different colors.
<sigh> come on, look at 3D gamut plots. There's virtually no colour range of any significance that is only in sRGB. For all practical purposes Adobe RGB encompasses sRGB.
Trying to argue anything else just makes you look daft.
Logged
Pages: 1 ... 15 16 [17] 18 19 ... 24   Go Up