Editing in a dark room: results

[Frans Waterlander]

Only to you. As other's have correctly pointed out.

So, did you actually run some tests?

[digitaldog]

So, did you and others actually run some tests?

It's as much a waste of time as trying to get you to prove your concept. I've been editing in dim and sometimes totally dark conditions for years and years; no dark images, no dark prints. So years of experience plus the recommendation of the ISO, NEC and two esteemed color scientists reinforce there's nothing to test. It works and works just fine.
Maybe you need your vision checked.

[Frans Waterlander]

It's as much a waste of time as trying to get you to prove your concept. I've been editing in dim and sometimes totally dark conditions for years and years; no dark images, no dark prints. So years of experience plus the recommendation of the ISO, NEC and two esteemed color scientists reinforce there's nothing to test. It works and works just fine.
Maybe you need your vision checked.

I do get dark images when I switch off the lighting, as I have reported. So it seems to me I have some prove of concept.

[digitaldog]

I do get dark images when I switch off the lighting, as I have reported. So it seems to me I have some prove of concept.

Again, only for you. So you know what? Use what works best for you, and consider getting your vision checked. Meanwhile, I'll stick with my decades of experience editing image and the recommendation of people who know far more about this subject than I do, and certainly you do. And no, there's not a lick of proof you've provided for others.

[N80]

I do get dark images when I switch off the lighting, as I have reported. So it seems to me I have some prove of concept.

With the test you have cited you have only proven the concept for you and that particular image. While I'm not necessarily disagreeing with your point, the test you cited has no validity beyond you, in your room with that print.

I will say this again, as I wade into waters that are way over my head, that the vast majority of this sort of discussion exposes a certain level of fetishism (on both sides of the argument) that probably has little to do with actual print outcomes.

[Doug Gray]

Here's an approach that eliminates most of the experimenter bias:

Frans could select a group of 5 photographs.

Then have another person unware of this discussion or hypothesis tweak them using only the midpoint in curves in both lighting conditions to best match what they recall the printed images appearing like and saving the results.

These could then be examined to determine whether the adjustments were as significant as Frans indicates from his experience.

It would be best to have more than one person do this. However, if the results are anywhere near the differences in the images Frans posted this process should easily demonstrate that.

In my experience there isn't much, if any, difference from editing in any ambient light so long as it is well below that of the display or proofing table.  However, I don't have any experience working in near darkness. It would be an interesting experiment.

[digitaldog]

Here's an approach that eliminates most of the experimenter bias:
Frans could select a group of 5 photographs.
Then have another person unware of this discussion or hypothesis tweak them using only the midpoint in curves in both lighting conditions to best match what they recall the printed images appearing like and saving the results.
These could then be examined to determine whether the adjustments were as significant as Frans indicates from his experience.
It would be best to have more than one person do this. However, if the results are anywhere near the differences in the images Frans posted this process should easily demonstrate that.

Good suggestions, I'd add a few and a caveat. The images should be carefully selected and scene referred so no subjective edits have been applied. They should contain differing amounts of highlights/whites in the mix along with darker tones. Ideally the group would be both men and women of differing ages to account for their quality of color vision. The caveat is this would take a good deal of effort and the person designing the tests, selecting the images and observing the process needs not be prejudiced and is willing to undertake the effort. IMHO, that eliminates Frans from the testing. 

[Jim Kasson]

As covered in another thread, I have maintained that editing in too dark a room will result in a tendency to tone down the image brightness, resulting in prints that are too dark and here is the proof. My digital darkroom lighting causes the viewing area on the right site of my monitor to be around 28-30 lux; on the left side it is around 5-7 lux. With the lighting off, it's 0 lux everywhere. I edited the same image with the lights on and off, using Curves to vary the midpoint, trying to make the image look as I remember it. The results in % brightness:
lights on                            lights off
average: 54%                    average: 49%
wall: 97%                          wall: 89%
sky: 82%                           sky 58%
tailgate: 72%                     tailgate: 47%
shadow: 12%                     shadow: 6%
The averages were obtained by selecting the whole image and using filter/blur/average.

This suggests that the room brightness at lower levels (below 32 lux) does indeed have a big impact on brightness perception and editing results.

The state of adaptation of the viewer affects the way that the colors are perceived.

Take look at Hunt's Reproduction of Colour, 4th Edition, on page 56.

With no surround at all on the monitor, objects visible in the periphery of the monitor become, in effect, the surround, and thus affect the viewers adaptation and the way that colors are perceived. hus lighting levels will affect the surround.

However, soft proofing without a surround is not accurate unless your output will be displayed under conditions where the surround is the same as the room where the editing is done happens to be and the image appears to be self-luminous. Those conditions are so infrequently met and so unstable that soft proofing with a sizable surround on screen is really necessary. Under those conditions, with a light gray or brighter surround and a monitor white point of 80 cd/m2 or so, the difference between a dim room and a dark room should be minimal.

http://blog.kasson.com/?p=15602

Jim

[Doug Gray]

Good suggestions, I'd add a few and a caveat. The images should be carefully selected and scene referred so no subjective edits have been applied. They should contain differing amounts of highlights/whites in the mix along with darker tones. Ideally the group would be both men and women of differing ages to account for their quality of color vision. The caveat is this would take a good deal of effort and the person designing the tests, selecting the images and observing the process needs not be prejudiced and is willing to undertake the effort. IMHO, that eliminates Frans from the testing. 

That would be appropriate for a more thorough experiment. I was trying to keep it simple enough that Frans (or anyone else) could do a quick test. His adjusted example photos are very, very different so a simple test should counter his results easily. But if, in fact, this is a real phenomena then it won't take much to replicate and then a more complete test with different people like you suggest would be in order. That said, this strong of an effect should already have been noticed and studied long ago and I'm not aware of any such study. OTOH, I haven't researched it either so it might have been.

I'm inclined to chalk it up to confirmation bias but I could be wrong.

[GWGill]

I agree completely with Jeremy (kikashi) above. As a working scientist at the largest university in Canada, the idea that this is a well thought out experiment and/or "proof" of anything is ridiculous...

Seems a well enough thought out data point to me. In itself it doesn't prove anything, but it is certainly how one starts to investigate a phenomenon of interest.

[digitaldog]

That said, this strong of an effect should already have been noticed and studied long ago and I'm not aware of any such study.

Right, plus those who are much smarter and understand this topic better than I have already provided their suggestion, specifically Karl Lang, =/- 4 Lux.

[Doug Gray]

The state of adaptation of the viewer affects the way that the colors are perceived.

Take look at Hunt's Reproduction of Colour, 4th Edition, on page 56.

With no surround at all on the monitor, objects visible in the periphery of the monitor become, in effect, the surround, and thus affect the viewers adaptation and the way that colors are perceived. hus lighting levels will affect the surround.

However, soft proofing without a surround is not accurate unless your output will be displayed under conditions where the surround is the same as the room where the editing is done happens to be and the image appears to be self-luminous. Those conditions are so infrequently met and so unstable that soft proofing with a sizable surround on screen is really necessary. Under those conditions, with a light gray or brighter surround and a monitor white point of 80 cd/m2 or so, the difference between a dim room and a dark room should be minimal.

http://blog.kasson.com/?p=15602

Jim

Interesting. Frans could well have run his test in Photoshop with no surround or white border. If so then ambient lighting affects perception much more. I have a CG318 which is rather large and do editing with a gray, background. I also edit the image with a matching white border "canvas" if I intend to print with a white border. Similarly my proofing environment backing is a neutral gray that roughly matches what I use in Photoshop.

[GWGill]

But I do see this specifically on page 140: The Stevens effect indicates that as luminance levels increase, dark colors will appear darker and light colors will appear lighter. ]

Right. So when you are editing an image at low adapted luminance levels, you tend to compensate for the decreased contrast by reducing your darker colors, thereby increasing perceived contrast. i.e. you've reduced the overall brightness of your image.

But I prefer the elegance of the explanation in this paper: "A probabilistic explanation of brightness scaling", notably Figure 1. To summarize :- our local contrast sensitivity is highest around the brightness we are adapted to. Equivalently this makes the apparent brightness of anything above our adaptation level higher, and the apparent brightness of everything below our adaptation level lower.  So at the extremes, if we are looking at an image with bright ambient light (say almost as bright as the white of the image), the whole image range will appear darker, while if were to have a very low ambient light level, the whole image will appear lighter. While editing we would compensate for this appearance, making the image brighter or darker respectively.

And as I've noted previously, it's also well accepted in setting up video viewing environments that the gamma of the display should be increased as the viewing environment gets darker. What this is actually doing is applying a viewing conditions compensation due to the difference between the video encoding gamma (about 2.2) used for material captured in bright environments and the display gamma (somewhere between 2.2 for bright environments, 2.3 for dim, and 2.4 or even more for very dark environments).
Poynton covers this in some detail.
Now if the display is not compensated for this effect when we are editing, then we will tend to make this compensation ourselves, making the image darker in a dark viewing environment.

Nowhere do I see anything that states or suggests editing images on an emissive display in a dim environment produces images or prints that are too dark.

Seems pretty clear to me from the color scientific literature and experience in areas like Video, that this phenomena is quite plausible.

[digitaldog]

Seems pretty clear to me from the color scientific literature and experience in areas like Video, that this phenomena is quite plausible.

Plausible? Lots of ideas are plausible. Explain if you can, the ISO spec, specifically for editing our images and their recommendation for ambient conditions on a computer display and the comments from Dr. Lang? Or the NEC recommendation that confirms what the ISO suggests. Again, where is there proof that low ambient light conditions surrounding our display result in dark images and dark prints? Who besides Frans recommends not using an ambient condition lower than the suggestions above? Cause and effect.

[Mark D Segal]

One can over-complexify an issue till it all becomes totally counter-productive. Why work in a totally darkened room altogether? Very bad for one's eyesight and completely unnecessary. Why work in a bright room where the distraction from the monitor could be overwhelming - just silly. So work in a low-light environment. For example, I have two shaded lamps burning 60 watt bulbs ten feet away from the monitor (which does not face the lamps). It's fine. From there, look at four things: (1) is the monitor too bright relative to the environment so you adjust the prints too dark; (2) is the image surround for soft-proofing not too bright and not too dark, to roughly simulate average viewing conditions of the prints (unless you have other viewing conditions), (3) is your soft-proof set-up correctly and (4) when the print comes out of the printer and you examine it say under Solux illumination, does it satisfy - is it faithful enough to the softproof? If the answers are NO, YES, YES and YES, you're fine. Print away happily.

[digitaldog]

Why work in a totally darkened room altogether? Very bad for one's eyesight and completely unnecessary.

From the two closed posts:

(Karl Lang*) writes in the Adobe PDF on setting up a digital darkroom:
With the light output levels of current display technology, an ambient light level of 4 lux is an optimal compromise. While a lower light level would provide better results, it’s impractical. You need to be able to walk around your environment. Above 16 lux, the room ambient will have a very significant effect on the dynamic range of the display.


We should note that Frans has never told us 'what's too low' or provided any metric that he feels is so low, we'll begin to or will produce dark images and prints. Based on what Karl writes, Frans range at best is 4 lux or less.


* http://www.lumita.com/information/
Karl Lang is a color scientist, engineer and the principle of Lumita, Inc. With 20 years experience in the digital imaging industry, Karl has designed, developed and brought to market numerous products. The Radius PressView SR & XL, ProSense, ColorMatch, Separation Lab, and the Sony Artisan Color Reference Display System are just a few of the integrated color imaging products he created. Karl has extensive experience developing integrated systems that include hardware, software, industrial design and optical components. Karl enjoys the challenges and complex problems presented when integrating advanced color management into user friendly systems. A hallmark of all Lumita creations is the simplification of color management and the reduction of complexity. “Color management should just work.”

[GWGill]

Explain if you can, the ISO spec, specifically for editing our images and their recommendation for ambient conditions on a computer display and the comments from Dr. Lang?

The existing approach to dealing with viewing condition effects is to control them. Things like ISO viewing conditions have been developed with certain expectations :- maximum brightness of the display, the viewing conditions implied by the image encoding profile, etc.
Given the relatively low display brightness assumed (typically 100 cd/m^2), and the bad effects glare have on image appearance, it's easy to understand an emphasis on minimizing ambient light levels, so the standards end up specifying quite low levels. So yes, I would be surprised if going from an ambient of 10% of maximum image brightness to 0% will have much effect (so I'm agreeing with you in these specific circumstances). But there is every reason to believe that going the other way will have an effect - that's why ISO specify a maximum ambient level!

i.e. if you have been editing images on a display that perhaps is not setup to ISO (brighter maximum, ambient above 10%) and you have been happy with your resulting prints, and then you turn all the lights off and edit in complete darkness, your prints are likely to be darker.

Again, where is there proof that low ambient light conditions surrounding our display result in dark images and dark prints?

I've pointed you in the direction of understanding - the rest rest up to you.

[digitaldog]

The existing approach to dealing with viewing condition effects is to control them.

No argument.

Things like ISO viewing conditions have been developed with certain expectations :- maximum brightness of the display, the viewing conditions implied by the image encoding profile, etc.

Indeed. So you disagree with their suggestions for the ambient light conditions around the display?

Given the relatively low display brightness assumed (typically 100 cd/m^2), and the bad effects glare have on image appearance, it's easy to understand an emphasis on minimizing ambient light levels, so the standards end up specifying quite low levels. So yes, I would be surprised if going from an ambient of 10% of maximum image brightness to 0% will have much effect (so I'm agreeing with you in these specific circumstances). But there is every reason to believe that going the other way will have an effect - that's why ISO specify a maximum ambient level!

Yes it is easy to understand. What isn't easy to understand is Frans idea a too low setting, undefined by him results in images and prints that are too dark. Do they?

i.e. if you have been editing images on a display that perhaps is not setup to ISO (brighter maximum, ambient above 10%) and you have been happy with your resulting prints, and then you turn all the lights off and edit in complete darkness, your prints are likely to be darker.I've pointed you in the direction of understanding - the rest rest up to you.

Again, where's the evidence to support that a ambient light at 32 lux or below results in images that are too dark or prints too dark? It's a simple question. Where's the proof? Or the ISO, Karl and the other's suggesting low settings are wrong?

[GWGill]

I was having a think about the ISO viewing conditions in relationship to ambient light levels, and there is no reason to think that reducing ambient levels below 10% will have no effect - viewing conditions effects are a continuum after all. So this raises the question of why the ISO conditions are so loose as to be "10% or less", rather than a specific aim point ? I wondered if perhaps the effects of ambient light are so much worse at 10% or more, that they were content for the viewing conditions to be "somewhere under" this range.

Rather than attempt to carry out psycho-visual experiments that have been done more expertly by others, I'm happy to simply make use of their work by using the CIECAM02 appearance model as a basis for investigations.

If I take (say) an sRGB ICC profile and run (say) 50% RGB values through it while varying the ambient light level, I get the following J appearance values (analogous to L* values):

   Ambient %    J            Delta J to 10%
   0                   51.9       +3.3
   10                 48.6         0   
   20                 43.0       -5.6
   50                 37.4       -11.2
   100               29.6       -19.0

[ This was with zero glare, standard 20% image surround, and 100 cd/m^2 white. ]

So:

 1) As expected, the image looks brighter as you reduce ambient levels.

 2) There is a change if ISO 10 % ambient is reduced to zero.

 3) But this change is a bit smaller than differences amongst higher ambient light levels.

 4) ISO spec's are looser than they really should be for critical work.

If I then feed the 0% ambient J values back through the sRGB profile with the 10% ambient conditions, I get RGB values of about 53.5%, indicating that to (roughly) make the image edited at 0% ambient the same brightness as the one edited at 10%, one would have to increase the RGB values by about 3.5%. This is consistent with the numbers Frans Waterlander reported in his experiment, where his averages changed by 5%.

[digitaldog]

I was having a think about the ISO viewing conditions in relationship to ambient light levels, and there is no reason to think that reducing ambient levels below 10% will have no effect - viewing conditions effects are a continuum after all.

No one is arguing that isn't true. Where does the ISO speak of editing images per se and do not recommend a low ambient light level when the print isn't being viewed? And further doing so results in images and prints (then viewed properly) that are too dark? Further, there is at least one authority and color scientist who does define specific recommendations for this task; editing our images:
https://www.adobe.com/digitalimag/pdfs/calibrating_digital_darkroom.pdf
Specifically the recommendation is below 16 lux and as low as 4 lux (lower could result in bumping into things).
I think we must both agree with Karl that any ambient light striking the display affects the black and our perception of that display. High values, values not recommended make sense. Lower? Still no proof of concept.
Where are the masses of people who perhaps followed Karl's advise (this is an Adobe PDF that's been around awhile), complaining their prints are too dark while admitting they are working in conditions under 16 Lux or there about?
I suggest we must consider psycho-visual experiments, this is much to do about perception!