How to visually gauge absolute black in nature and tone map it to output.

[Steve House]

I have to confess I've been trying to follow this thread and try as I might I fail to see the point.  Unless you are designing image sensors or image processing software why do the numbers matter?  In the field you stand in front of a scene and form an emotional/intellectual impression that you would like to convey to an audience.  You make your exposure and return to post to prepare the image.  You manipulate it using your preferred software until the resulting image looks like the impression you formed looking at the scene.  What counts is when you present that image to an audience in the form you wish to display it (screen, print, whatever) they receive the impression you want - it really doesn't matter if the darkest tone in the image is 0,0,0 or 5,5,5 or 20,20,20, as long as the image says what you want it to say you've done the job.  Indeed photography abounds with excellent images that have NO absolute black or absolute white.

[michael]

I just got back last week for printing my new book in China. The project was lead by Arturo Chapa, one of the world's top book designers (in my opinion at least, but since his customers include the King of Spain, the President of Mexico and the Guggenheim Museum I'm in some good company).

In any event, while doing one of the first signature checks Arturo asked the press master to lighten the black (K) ink by 7%. The press master (PM) replied that the densitometer reading was as low as he was allowed to go and that he would take no responsibility for printing the K ink any lighter.

Arturo insisted. The PM asked that he sign a waiver, and another test was run.

The images looked much better. The PM said...You were right. The image looks better even though the numbers are wrong.

Arturo's reply was that people look at images, not numbers, and if there's a disagreement, go with how it looks, not how it measures.

Seems about right to me.

Michael

[digitaldog]

Arturo's reply was that people look at images, not numbers, and if there's a disagreement, go with how it looks, not how it measures.

Exactly. Colorimetry is about color perception based on color measurements. It is not about color appearance and the differences are significant. The reason why viewing something is more valid than measuring it is because measurement is about comparing solid colors. While color appearance is about evaluating images and color in context which no measurement device can do.

[cybis]

As RGB 0 is mapped to darker and darker printed blacks, more and more light is needed to illuminate the prints. Most calibration processes will map RGB 0 to the darkest black possible and RGB 255 to paper white. Light intensity shinning on the print doesn't enter the equation. But maybe it should?

Lightness (the L in Lab) is supposed to be perceptually uniform, i.e. equal Delta L being perceived as equal steps. But clearly the model fails when light intensity being reflected from the print is very low, either because of dark ink or low illuminant. And with high D-max baryta paper, for example, the number of photons bouncing back into the viewer's eye from the deepest black can be very low indeed.

Now, the lightness model is not supposed to fail at very Y/Yn (relative luminance) values. It makes prediction about how we perceive differences in light intensity all the way to Y/Yn equal zero.

Am I missing something or are we all just working from a flawed model of how human perceive light? (I know it’s probably the former… just don’t know where I went wrong.)

And Tim, I’m sorry if this has nothing to do with your question (I saw 'Black' in the title and went with it)

Edit: change Y/Y0 to Y/Yn (relative luminance)

[digitaldog]

Lightness (the L in Lab) is supposed to be perceptually uniform, i.e. equal Delta L being perceived as equal steps.

Key here is 'supposed to be'.

Am I missing something or are we all just working from a flawed model of how human perceive light?

We're not working with a color appearance model, that's one issue.

[Eric Myrvaagnes]

I just got back last week for printing my new book in China. The project was lead by Arturo Chapa, one of the world's top book designers (in my opinion at least, but since his customers include the King of Spain, the President of Mexico and the Guggenheim Museum I'm in some good company).

In any event, while doing one of the first signature checks Arturo asked the press master to lighten the black (K) ink by 7%. The press master (PM) replied that the densitometer reading was as low as he was allowed to go and that he would take no responsibility for printing the K ink any lighter.

Arturo insisted. The PM asked that he sign a waiver, and another test was run.

The images looked much better. The PM said...You were right. The image looks better even though the numbers are wrong.

Arturo's reply was that people look at images, not numbers, and if there's a disagreement, go with how it looks, not how it measures.

Seems about right to me.

Michael

Right on!

[cybis]

Key here is 'supposed to be'. We're not working with a color appearance model, that's one issue.

Swoosh (sound of the rabbit hole opening under my feet).

[papa v2.0]

Swooooosh

Hi Tim
Yes understand your problem

For print output (display is another rabbit hole)



The 'problem' is that the current imaging chain is is designed around reproduction of prints to be viewed under D50 illuminant and at a luminance of between 500lux (P2) and 2000lux. and a surround reflectance of 20%

Any deviation from the standard chain can be problematic and needs to be done on a trial and error basis, which then becomes down to experience (as in Shewe's example).


For dim viewing conditions as low level tungsten in living room, one has to use a colour appearance model to adjust the colormetric output to the new (as opposed to standard viewing conditions). The current CIE appearance model is CIECAM02. Worth reading up about.

This would change the mapping of the input values to the output cmyk values to achieve the same appearance under the new viewing conditions.

The colour appearance rendering in profiles at the moment is fixed to the above standard output.

What is need is the ability to insert a colour appearance model into the digital imaging chain. I think an attempt at this has been made in the form of a Photoshop plug in and in the raw rendering program, RAW Therapee.

Do remember that the lower the lighting levels, the more likely the model will fail.


I would refer you to the following links for some background reading,

http://www.color.org/iccprofile.xalter

Book:  Mark D. Fairchild, Color Appearance Models, 3rd Ed. Wiley-IS&T, Chichester, UK (2013). ISBN 978-1-119-96703-3

http://www.cis.rit.edu/fairchild/CAM.html

Iain

[Tim Lookingbill]

I just got back last week for printing my new book in China. The project was lead by Arturo Chapa, one of the world's top book designers (in my opinion at least, but since his customers include the King of Spain, the President of Mexico and the Guggenheim Museum I'm in some good company).

In any event, while doing one of the first signature checks Arturo asked the press master to lighten the black (K) ink by 7%. The press master (PM) replied that the densitometer reading was as low as he was allowed to go and that he would take no responsibility for printing the K ink any lighter.

Arturo insisted. The PM asked that he sign a waiver, and another test was run.

The images looked much better. The PM said...You were right. The image looks better even though the numbers are wrong.

Arturo's reply was that people look at images, not numbers, and if there's a disagreement, go with how it looks, not how it measures.

Seems about right to me.

Michael

I'ld like someone in this thread to explain what makes me edit my images to look too dark? I'm not a rank beginner. I have a trained eye.

I've spent the weekend lightening the white rose image for the third time (even the lightened version I posted here is too dark) as well as this orange flower image (see below) because I'ld like to be able to see them in my living room light. Below are the lab luminance readings between the two edits of an area I chose to use as a target among other areas to lighten because on my transmissive display my eyes have adapted to the darker version's shadow areas to the point I can see everything in the background so I assume it will be seen clearly on the print.

The business card below the 8x10 print is the original dark edit from the framed 8x10 hanging on my wall I posted earlier. Why should I choose the dark version over the lighter one? Is it a matter of aesthetics or practicality?

I have to use the numbers to tell how much my eyes have adapted to working on them for so long. The orange flower can't be just lightened with a simple slider edit or it will over saturate and blow out the detail in the folds of the petals. HSL edits are required including Luminance to the orange channel.

This is why I started this thread on black point mapping to original scene so I have some type of gauge on just how dark the shadows need to be. Now that I know the black point density doesn't really matter I'm going to use luminance numbers to override the adaptive effect on my eyes that get me to edit my images too dark.

[Steve House]

At the risk of being considered overly simplistic, if the image looks right on the screen while you're editing but the resulting prints are too dark, the editing screen's brightness is set too high.  Reduce the monitor's brightness control until the screen looks like your pilot print and leave it there.  Now when the image looks correct on the screen it will print correctly.  You don't have to get more mathematical than that.

[Tim Lookingbill]

At the risk of being considered overly simplistic, if the image looks right on the screen while you're editing but the resulting prints are too dark, the editing screen's brightness is set too high.  Reduce the monitor's brightness control until the screen looks like your pilot print and leave it there.  Now when the image looks correct on the screen it will print correctly.  You don't have to get more mathematical than that.

I get exact screen to print matches in both luminance and color with the print viewed under lights that make white paper the same brightness of white as 255RGB white on my display.

You're solution comes from not clearly understanding this simple question which I'll repeat again...

...explain what makes me edit my images to look too dark? This means look dark both onscreen and on the print but not noticing how dark the final edit really is on my transmissive display and how it will translate to a print that will be viewed in various lighting situations.

If you've ever worked in a dark room using a stouffer step wedge grayramp to tell you correct exposure and developing time and not rely on the eye as I have and then immediately switch to closely examining under a loupe color slides on a very bright light table, it will help you understand what I'm getting at.

Digital mashes these two optically driven phenomenon into one process where both the desire of the photographer to make the image look good collides with not knowing how far to go for output to a print for viewing under a wide range of lighting situations.

[digitaldog]

I get exact screen to print matches in both luminance and color with the print viewed under lights that make white paper the same brightness of white as 255RGB white on my display.
...explain what makes me edit my images to look too dark? This means look dark both onscreen and on the print but not noticing how dark the final edit really is on my transmissive display and how it will translate to a print that will be viewed in various lighting situations.

[SZRitter]

I get exact screen to print matches in both luminance and color with the print viewed under lights that make white paper the same brightness of white as 255RGB white on my display.

You're solution comes from not clearly understanding this simple question which I'll repeat again...

...explain what makes me edit my images to look too dark? This means look dark both onscreen and on the print but not noticing how dark the final edit really is on my transmissive display and how it will translate to a print that will be viewed in various lighting situations.

If you've ever worked in a dark room using a stouffer step wedge grayramp to tell you correct exposure and developing time and not rely on the eye as I have and then immediately switch to closely examining under a loupe color slides on a very bright light table, it will help you understand what I'm getting at.

Digital mashes these two optically driven phenomenon into one process where both the desire of the photographer to make the image look good collides with not knowing how far to go for output to a print for viewing under a wide range of lighting situations.

So, umm...

I understand your problem, and have no where near the technical expertise of most of the others in the conversation, but isn't this kind of like shooting for Zones? Wouldn't you conceptually, at the time of capture say this should map to Zone V, then using LAB or something else make sure that spot's values fall near a Zone V? At least, with my basic understanding of the Zone system, that would be how I would apply it. Or am I also completely missing the point?

And on the issue of prints and lighting, my understanding of LED based lights for the home is that, even though sold as "warm" to match tungsten, the light given off by these LEDs is different than true tungsten. If that is true, and the visible spectrum is altered (this is coming from analysis of LEDs and suitability for video production, albeit a few years ago), wouldn't this also negatively influence the appearance of your prints? Something that could be very hard to compensate for as we are in a transition phase from tungsten to LED in the US.

[Steve House]

...
You're solution comes from not clearly understanding this simple question which I'll repeat again...

...explain what makes me edit my images to look too dark? This means look dark both onscreen and on the print but not noticing how dark the final edit really is on my transmissive display and how it will translate to a print that will be viewed in various lighting situations.
...

So what you're saying is that the print matches the screen but when you edit on-screen you somehow see it as lighter than it really is and so you call it done when it's really too dark, that your eyes are being fooled and you don't realize it until after pulling a pilot print?  If that is what is happening the answer lies somewhere in your own psychology and the mathematics of RGB values and tone mapping isn't going to help.

How are you defining "too dark?"  Is it by some numerical standard of what the print density of certain tones theoretically "ought" to be or is it a more subjective measure of not having a tonal range (and biased toward the dark) that conveys the message you intend to convey?

There really is no way to print so it looks perfect under a wide range of different lighting conditions.  You can calibrate every part of the process to a specific illuminant for the final print but it will look different under any other illumination.

[papa v2.0]

Sorry Tim
I though I understood you but I am afraid you have lost me again.

[Tim Lookingbill]

So what you're saying is that the print matches the screen but when you edit on-screen you somehow see it as lighter than it really is and so you call it done when it's really too dark, that your eyes are being fooled and you don't realize it until after pulling a pilot print?  If that is what is happening the answer lies somewhere in your own psychology and the mathematics of RGB values and tone mapping isn't going to help.

How are you defining "too dark?"  Is it by some numerical standard of what the print density of certain tones theoretically "ought" to be or is it a more subjective measure of not having a tonal range (and biased toward the dark) that conveys the message you intend to convey?

There really is no way to print so it looks perfect under a wide range of different lighting conditions.  You can calibrate every part of the process to a specific illuminant for the final print but it will look different under any other illumination.

Steve, the images I've posted throughout this thread will show you what I mean by too dark both visually and by the numbers.

My eyes ARE being fooled from the adaptive effect caused by the transmissive display skewing my judgment of overall brightness and contrast. Specifically when I brighten an image it loses contrast and definition so I increase contrast which makes the shadows even darker and I just keep going back and forth and back and forth and think I've made headway when I haven't because my eyes of adjusted too quickly.

I have to then resort to scene analysis by the numbers according to how much light was in the actual scene and stop emotionally driven edits from skewing my judgement. Things look great when darkening and adding drama like some film noir/Marvel Comics rendering (i.e. Ansel Adams style tone map where daylight scenes of Yosemite look like night). But because my eyes adapt quickly to the darker version I don't think it's dark at all but it sure does look good. The lower dynamics of a print will make it seem even darker in dimmer light than a print matching booth.

But there are so many options available working in Raw to make an image look just as dynamic and dramatic without resorting to making it look dark and I've found a way to do just that.

[Tim Lookingbill]

So, umm...

I understand your problem, and have no where near the technical expertise of most of the others in the conversation, but isn't this kind of like shooting for Zones? Wouldn't you conceptually, at the time of capture say this should map to Zone V, then using LAB or something else make sure that spot's values fall near a Zone V? At least, with my basic understanding of the Zone system, that would be how I would apply it. Or am I also completely missing the point?

And on the issue of prints and lighting, my understanding of LED based lights for the home is that, even though sold as "warm" to match tungsten, the light given off by these LEDs is different than true tungsten. If that is true, and the visible spectrum is altered (this is coming from analysis of LEDs and suitability for video production, albeit a few years ago), wouldn't this also negatively influence the appearance of your prints? Something that could be very hard to compensate for as we are in a transition phase from tungsten to LED in the US.

Yes, SZRitter, I'm doing a reality check on my edits using a sort of zone system but for shadows which is the bottom end for establishing minimum density in image detail important enough to be seen in dim living room light. Not all images will have dark detail that needs to be seen in this sense.

The LED I'm using doesn't affect color or tone as you've indicated. Not all LEDs are alike. Check out the shot of the row of framed 8x10's hanging on my living room wall I posted earlier in this thread. I'm not talking about maintaining perfect print matching in this type of light but just to have shadow detail viewable so the idea of the subject can be conveyed.

[Tim Lookingbill]

Found this link from a Photo.net discussion on darkroom print editing notes...

http://petapixel.com/2013/09/12/marked-photographs-show-iconic-prints-edited-darkroom/

It helps explain the motives behind the decisions I'm discussing about editing an image based on an intent other than just taste.

Scroll down to the Audrey Hepburn image showing the lifted shadow detail in the reflection on the black card door. Is that important shadow detail that should be seen in all types of light as in a magazine at the library, airport or poorly lit bus? Or should they have left the reflections as they were? What would your aesthetics and taste dictate and what would you base it on?

I would've left the reflections as they are because they distract from Audrey's face framed by all the black surround. I wouldn't care if anyone could see it in any light. OTOH if thought it was important I'ld want to make it light enough so that it could.

[Steve House]

Steve, the images I've posted throughout this thread will show you what I mean by too dark both visually and by the numbers.

My eyes ARE being fooled from the adaptive effect caused by the transmissive display skewing my judgment of overall brightness and contrast. Specifically when I brighten an image it loses contrast and definition so I increase contrast which makes the shadows even darker and I just keep going back and forth and back and forth and think I've made headway when I haven't because my eyes of adjusted too quickly.

I have to then resort to scene analysis by the numbers according to how much light was in the actual scene and stop emotionally driven edits from skewing my judgement. Things look great when darkening and adding drama like some film noir/Marvel Comics rendering (i.e. Ansel Adams style tone map where daylight scenes of Yosemite look like night). But because my eyes adapt quickly to the darker version I don't think it's dark at all but it sure does look good. The lower dynamics of a print will make it seem even darker in dimmer light than a print matching booth.

But there are so many options available working in Raw to make an image look just as dynamic and dramatic without resorting to making it look dark and I've found a way to do just that.

'Too dark' visually I understand, though I found it impossible to get a meaningful impression of the images you posted, considering they're photographs of photographs and are further being filtered through my own screen settings on the laptop I use for web browsing. I can see one is darker than the other but can't go much beyond that in seeing into the image details. 'Too dark by the numbers' is a concept I can't relate to.  No matter what the numbers are, if it looks right it is right and if it looks wrong it is wrong, IMHO. While there's no question that there are some mechanical/technical elements involved, expressive print making is not a numbers driven process where you set the 'right' values for X and Y and everything else falls into place.

[Tim Lookingbill]

'Too dark' visually I understand, though I found it impossible to get a meaningful impression of the images you posted, considering they're photographs of photographs and are further being filtered through my own screen settings on the laptop I use for web browsing. I can see one is darker than the other but can't go much beyond that in seeing into the image details. 'Too dark by the numbers' is a concept I can't relate to.  No matter what the numbers are, if it looks right it is right and if it looks wrong it is wrong, IMHO. While there's no question that there are some mechanical/technical elements involved, expressive print making is not a numbers driven process where you set the 'right' values for X and Y and everything else falls into place.

So why mention you're viewing my posted images on a laptop? Why would that be an issue since if it looks right then it's right, if it looks wrong it's wrong? Is the laptop calibrated and does it match your higher end calibrated workstation? Just so we're comparing apples to apples, right? I edited those images on a calibrated system to match VERY close to what I see in my living room and I'm quite good at making a match by eye this way? Not so good at remembering how dark or light a scene I shot a week ago and now attempting to edit to look good and not realize I've made it too dark.

If you don't see the contradiction in your arguments in what I just indicated then there's no point in helping you understand what I'm conveying. And I never said I go strictly by the numbers but you seem to continue to misinterpret and add words to what I didn't say.

As some photographers use a gray target to establish 128RGB midgray for proper exposures I'm just advocating a minimum shadow number in relation to absolute black=(which constantly changes scene to scene>how it's seen on a transmissive display>to print depending on paper/ink) so that it will be viewable in a print and communicate at least what the image is about under the dimmest light. Nothing complex or technical about it.