Screen calibration without hardware

color_management

#1

I’m trying to get a bit of a color managed workflow. Having found some profiles for my monitor, I found there was still a difference between what I had printed and what is on screen.
Playing with the monitor settings got things a bit better. But my home is like a glasshouse and 2700K moody lighting is not suitable for photo editing.
So, get a 5500K “photographic” CFL lamp. Ahh, that is better. Then it dawned on me:

I can use my camera to measure the white point of my screen!

Using the manual white balance of the A6000 on a pievce of white paper gave a 5500K, 0 A-B and 0 G-M shift. Now on to the monitor. 5300K, A-B 0, G-M 2G it said. Lower the green, increase B to raise the color temperature. And a few adjustments later, I notice that midtones are off. Adjust monitor gamma…

And while things are not perfect, they are now at least decent.


(Mica) #2

If you don’t have the cash for a new calibrator, watch ebay/craigslist etc for a cheap one. Or find your local camera club and make a friend who will let you borrow on 2-3 a year.


#3

@Sympa Keep in mind that you don’t know how accurate is your A6000. Did you profile it with test chart and generate ICC input profile to see?
At what lightning did you measure white paper? Which paper? Paper is not usually white (unless it’s very good quality photo art paper). Probably it’s more or less blue/yellow.
While colorimeter/spectrometer is expensive test chart is not cheap either. X-Rite ColorChecker Classic costs like 1/3 of the i1 Display Pro (at least in Poland).
You have downloaded generic profile and apply it but you don’t know in what condition is your display. It may and should differ.

If you’re happy with the results that’s good. It’s all that counts (unless you sell your prints!).


#4

Screen measurement devices are not that expensive indeed I think. DisplayCal is the software to use (Linux here) but ambient lighting plays an important role.
But the thing is - my prints now compare reasonably well to my screen - as long as I have the right lighting even side to side. Lighting changes, the comparison changes.

I feel the camera white balance was very useful to get the white point. and the generic monitor profile was made useful by adjusting the white point via camera white balance. But it is not perfect…


#5

As I said if you’re satisfied with your prints then it’s all that counts. I guess that you fit in to The Pareto principle :slight_smile:


(Elle Stone) #6

Hmm, well, if the print matches what’s on the screen under the same “near D50” lighting, and @Sympa likes the print - then it really is the intended image. So the print that’s being sold would be the print @Sympa intended to sell :slight_smile: unless of course the digital file was sent out to be printed on some other printer.

Sometimes we forget that before there was ICC profile color management, the big print houses did very well using “color management by calibration to make screen and prints match”, and editing basically in “monitor color space”.

As @maf indicates, the problem with “color management by calibration to make screen and prints match” of course is when sharing digital files with people using devices that aren’t calibrated to match the devices where the digital file was first created. But as long as everything is kept in house, all is good. Until of course the screen or printer calibration changes - then the other has to be changed to match, and so the big print houses took steps to make sure everything was always calibrated to the same in-house standard. But that’s also a problem when using ICC profile color management - if the device changes, time for a new ICC.

Ideally @Sympa would indeed profile the monitor rather than using a generic monitor profile - as already noted the calibration devices aren’t that expensive. Just make sure they work with Linux. But even if you do this, the actual lighting conditions being used to view the print and the on-screen image would still need to be accounted for.

Ideally the monitor white and the color of the light used to view the prints would be the same “color of white”, which is what @Sympa has taken steps to accomplish. I’m saying this based on trying to recreate on the screen a photograph of a painting that actually matches the real painting sitting next to the screen. I know the common recommendation is calibrate the monitor to D65, apparently even when trying to evaluate prints under D50 lighting. But I don’t understand the rationale - anyone have input on this topic?


#7

@Elle http://www.xritephoto.com/ph_product_overview.aspx?ID=583&Action=Support&SupportID=3349

Also some people who knows about printing recommend 5800-6500.


(Brien Dieterle) #8

http://www.color-image.com/2012/02/monitor-calibration-d65-white-point-soft-proofing/

This is a pretty neat article. I didn’t realize that the brain was “short circuited” by the image simply being on the screen versus being an actual object.


(Elle Stone) #9

Yes, I believe the recommendation from Bruce Fraser also was to calibrate to D65, because of how dark the monitor was when calibrated to D50. Rodney’s paper was written in 2005 when we were all using CRTs. And I’m pretty sure Fraser’s Real World Color Management (hope I got the title correct) also predates LCD monitors.

The native color temp of my NEC LCD monitor is (was, the monitor is getting a bit old) around 5800, which I think is/was fairly typical of LCD monitors. Right now I have it calibrated to D55 because the native white is getting progressively greener (sign, I believe, of an impending need to replace the display).

Also today’s displays can be calibrated to be much brighter than the old CRT monitors. So I wonder what/how recommendations would change/have changed for current technology.

Hmm, reading through that article, and focusing on the last two paragraphs, it sounds like the user simply should never compare the print to what’s on the screen which should be calibrated to D65. Or else the ambient lighting under which the print is viewed should also be D65.

This 2014 thread over on the ArgyllCMS mailing list archive also mentions the color-image.com paper.:

In the particular post I just linked to @gwgill had some comments on the color-image.com paper. So I’ll just ask - @gwgill - is calibrating the monitor to D55 for soft proofing under D55 lighting a reasonable thing to do?

I did manage to get a good match between “actual painting under halogen lighting” and “photograph of painting on screen calibrated to D55” using RawTherapee’s CIECAM02 module, when the painting was done under halogen light (much lower color temperature). But I seriously wonder if anyone else would see “what Elle saw” on their own monitors calibrated to D65 - one of these days I’m going to set up two calibrated monitors side by side and do some experimenting :slight_smile: .


#10

I’m back to my screen at around 5000K. I can hold a print in front and the color matching is quite good actually (except blacks and really bright colors).
Indded I was wondering about CIECAM02 - which sounds like a great tool to adjust my workspace to the environmental lighting - but I think CIECAM will change the saved picture, correct? So I should disable it before saving. Hmmm… sounds error-prone.


(Graeme W. Gill) #11

I’ve always been a little dubious about the claims that there is some sort of inherent psycological effect that prevents an emissive display being a reasonable facsimile of a reflective print, simply because I’ve personally seen contradicting examples, particularly using LCD displays.
That doesn’t mean that such an effect doesn’t exist, but if it does, the strength and circumstances in which it is evident don’t seem to make it an insurmountable barrier.

The most obvious barrier to side by side comparison of reflective and emissive comparison is white point - traditional illumination is/was incandescent lights at about 2700K, while the display standard (set by TV history) is 6500K. Graphic arts standards is 5000K, although this has historically been difficult to achieve, since high spectral quality 5000K light sources have been hard to get. Examining two images side by side in a mixed white point environment is basically impossible.

So first aspect is setup of your print lighting to be something reasonably close to 5000K, then make the display visually match it. Using an instrument is a good way to get close, but then tweaking the display aim point by eye for the best possible match is perfectly acceptable, since this can compensate for field of view effects and observer CMF variation.

The second aspect which should really be addressed first, is the lighting situation, including brightness. There is conflict - you want bright, uniform light falling on the print, but you don’t want it reflecting from the display, nor causing the viewer glare. So the physical arrangement of the print evaluation position, the display, the surround color and the illuminant is important.
[ Recommendations are for the surround (i.e. walls etc.) to be a 50% visual neutral gray. i.e. around 20% flat reflectivity. ]

The third related point is matching brightness. You want the white of the print to be the same brightness as the white on the display. Again, an instrument can help here.

Get all that right, and you are 90% of the way towards being able to compare prints and soft proofs and expect good matches.

Black point matching is probably the next step.

A camera is generally a poor substitute for an instrument, because they don’t have spectral sensitivities that are like a human observer.

CIECAM02/viewing condition adjustment can be used to change output if it is used to change the perceptual table of an ICC profile.


(Brien Dieterle) #12

This is an excerpt from the 2013 edition of Fairchild’s Color Appearance Models. I think Braun and Fairchild concluded that you’re better off just looking at the screen and print separately (not side-by-side) and “remembering” how they appeared. If LCDs changed this situation at all I would hope Fairchild would have noted this by 2013?

This is interesting to me since I don’t see much of a difference between comparing a print side-by-side with the screen and comparing a scene/object side-by-side with an emissive screen (iPad?) while digitally painting. That’s kind of a bummer.


#13

I’m guessing the following: On a CRT, ‘off’ was never black compared to the environment. Thus, the brightness had to be quite high to match the contrast ratio or a print. One would have the CRT under a dark hood (isolating oneself from the environmental lighting).
These days, I can stick a print on my monitor, and light it with equal brightness to my monitor. And it will look identical to what is on screen (well, it is not exact for some colors, but in general it is a nice match).
When working with CRTs I could never have imagined this.


(Graeme W. Gill) #14

Sounds like a project for some of the RIT P.H.D. students!

This is interesting to me since I don’t see much of a difference between comparing a print side-by-side with the screen and comparing a scene/object side-by-side with an emissive screen (iPad?) while digitally painting. That’s kind of a bummer.

If it works for you, then don’t be put off by such research conclusions. Time has moved on, and your situation may be materially different to the setup used in the previous research.


(Brien Dieterle) #15

I hope you’re right. I tried a crude version of the below also from Fairchild’s book:

image

I didn’t do this in the dark, but I matched the monitor white to the paper white visually; brightness and color. When viewed separately the paper looked white (of course) and monitor most definitely not white. So. . . seems to be legit. I encourage anyone to try a similar experiment if possible.


(Elle Stone) #16

@briend, in your experiment, to clarify what you did, did you calibrate the monitor to have a 2700K white point? And simultaneously illuminate the paper with incandescent lighting from a separate light source?

As an aside and slightly off-topic except in a practical sense, are “incandescent” and “ye old standard tungsten filament light bulb” lighting the same thing, along with halogen bulbs? That is, both are full-spectrum light sources with low color temp around 2700K?

There are so many different sorts of light bulbs being sold these days, including lights that have the same general shape as tungsten filament light bulbs, but have LEDs inside. And LED lights do have filaments, as the not-so-helpful people at the hardware store will stoutly proclaim when one is trying to find a standard tungsten light bulb.


(Dan) #17

@elle: are “incandescent” and “ye old standard tungsten filament light bulb” lighting the same thing, along with halogen bulbs?

Halogen bulbs burn a little hotter, they’re in the same general range as the standard incandescent bulb at about 3000K.

Related question: is the color temperature of newer “efficient” incandescent bulbs the same as “ye olde standard” ones?


(Elle Stone) #18

@dan - I hadn’t heard the phrase"efficient incandescent bulb" before, so I looked it up on the internet and found two different sorts of bulbs:

  • Research from MIT using nanotechnology, article dated 2016:
  • GE bulbs that replace the tungsten bulb with a halogen bulb, available considerably before 2016 but seemingly no longer available at our local big box hardware store (coincidentally we were there just last week actually looking for a halogen bulb that looked like a regular lightbulb and were told no such thing existed - all they had were floods):

Has the nanotechnology version made it to stores yet?


#19

Yes. At least in Sweden – but they are expensive as ****.
Also, it is very tricky to pick the correct wattage. I mean,
I know (please read knew) that I needed a 75W bulb, for instance.
Now they grade them in Lumen.

/Claes in Lund, Sweden


(Brien Dieterle) #20

Not at all. I visually adjusted the screen (entire background color) to match the white paper under whatever office lighting I have (fairly warm LED). I wasn’t trying to show as Fairchild did that an emissive and reflective object that are definitely the same color are not perceived as the same color. I wanted to try the inverse of that as suggested when Fairchild says “or, alternatively, predicted matching images that are unacceptable when viewed individually”.

So if the premise is that you should be able to compare a print against the screen as @Sympa shows in that screenshot above, I think first you should check if your white paper matches the screen white side-by-side AND in isolation. Maybe having D65 office lighting, as you mentioned earlier, would change things a bit?