Colors Come Out Muted Whenever I Print

[ajohnw]

The info I have says that high end printing can produce about 800 thousand colors, the best monitors 1.2 million, and the human eye can discriminate 1.7 to 2.5 million.

If this is the case why do some work in deep colour gamuts? Colin might clear this aspect up for you.

Opinions vary on the human eye. People who work with colour can become very sensitive to it much like conductors have more acute hearing tone wise than ordinary mortals.

My view is that monitors have evolved a lot in recent years but sRGB hasn't. That was cooked up to suite cathode ray type displays and fits them well. The 2 LED monitors I have calibrated struggle to fully cover the sRGB gamut, Dell quote 99% for my most recent one for instance however it's gamut volume is 128% sRGB. A lot of wasted colours. The none LED monitor I calibrated also barely covered sRGB.

Adobe RGB stretches the gamut to cover a bigger colour range along with bigger gaps but monitors can reproduce that colour range. They mostly do it by dithering colours - 8 + 2 bit it's usually referred too as. Once I believe 6+2 was used rather than the now usual 8.

As colour management is being added to all things on a PC that could need it I would say the next step is pretty obvious and doesn't need any really new display technology. The gamut will probably be the ones set by TV standards.

The resolution of a monitor is adequate, around 100 dpi so view from 30in / 750mm.

John
-

[Richard Lundberg]

We must have different information sources;

1. Pro photo printers can do a few hundred thousand colours (around half or even less than half of your numbers - the numbers I saw varied from around 250,000 to around 375,000). I can't find the article where I saw these numbers (thought it might have been at Luminous Landscape, but I can't find it right now;

2. According to some sources human eyes can see close to 10 million distinct colours http://hypertextbook.com/facts/2006/...erLeong.shtml; and

3. The best monitors (including dithering techniques) are at least 16 million distinct colours, and potentially much higher than that. At least that is what the manufacturers claim, but as their techniques are proprietary, that would be hard to validate. IPS monitors are natively 8-bits per channel and in theory can emulate 10-bits per channel.

There is a lot of disagreement about how many colors the human eye can discriminate, but I have never seen the reference you quote, I'll have to look that up. ( I did, the link isn't valid). There are 30 bit monitors that do 1 billion colors, that does not mean they are capable of actually putting 1 billion colors on screen. You can create all the numbers you want but the physical decice can't actually use them. IOW the numbers are there are but distinct they ain't. Same with 16 million. I am not sure which printing techniques generate the 800,0000 but I AM sure I can't afford them.

[Manfred M]

Try this link Richard; the semicolon screwed up the link:

http://hypertextbook.com/facts/2006/JenniferLeong.shtml

[ajohnw]

Richard may have been reading something like this which I suspect is a little out of date

http://www.graphicrepro.co.za/assets...-WideGamut.pdf

The digital people seem to think human vision needs 12 or 14 bits per channel. That number is probably part based on the smoothness of gradation. It a peculiar are when looking at an image. Take my new monitor. The calibration software can generate grey tone steps of 8bits from 0,0,0 rgb. The errors reported are reckoned to be below perceptual limits. I can just about discern the first 2. The reason for this is that my eyes have adapted to the average light levels that they see. Block of the rest and the steps would be a lot clearer - switch the lights off and even more so. If we did have 14bit per channel colour the 8 bits steps would have to go up proportional to the powers of 2 increase in the colour channel. This nothing like viewing a scene where our eyes can generally adapt to specific parts of it. Conclude what you will.

John
-

[Richard Lundberg]

Try this link Richard; the semicolon screwed up the link:

http://hypertextbook.com/facts/2006/JenniferLeong.shtml

http://hypertextbook.com/facts/2006/JenniferLeong.shtml does reach the page. However the page is only a collection of estimates and does not make any judgment on an actual number or range. This subject is still under active investigation and there are many references. I think I'll stick with CIE until someone revises it with full justification. I have collected about fifty interesting pdfs and pages. Color Perception by Kalionaitis looks good.

[ajohnw]

http://hypertextbook.com/facts/2006/JenniferLeong.shtml does reach the page. However the page is only a collection of estimates and does not make any judgment on an actual number or range. This subject is still under active investigation and there are many references. I think I'll stick with CIE until someone revises it with full justification. I have collected about fifty interesting pdfs and pages. Color Perception by Kalionaitis looks good.

You will find many 1d drawings of the gamut of human vision about even if you just google the word and look at images. It's a very well researched aspect of colour. One problem though is that no screen can show it realistically.

Colour is an analogue entity. The only things that puts a number on it is digital implementations of it. The only other aspect that may put a number on it is the eyes ability to differentiate colours. That aspect is really difficult.

John
-

[Macmahon]

The info I have says that high end printing can produce about 800 thousand colors, the best monitors 1.2 million, and the human eye can discriminate 1.7 to 2.5 million.

The "number of colours" possibly displayable by a device is a function, I would guess, of bit depth, regardless of what the objective 'colour' at the maximum and minimum bit-depths are.

I used ColorthinkPro to measure and plot the colour gamuts of my Epson 3880, and my old iMac monitor, whose gamut is approximately equal to sRGB.

You can see the comparative gamuts
here (comparing the printer to 'human vision') and here comparing the printer's gamut to the monitor. (Roll your mouse over the pictures to see the comparisons)

What the 2D projections don't show is that the monitor is capable of a greater contrast range, particularly in greens and yellows, which probably accounts for the advertised greater 'number of colours'. Nevertheless, the printer's gamut is greater. The range of colour it can reproduce is greater than the range reproducible on the monitor even if the number of discrete steps may not be.

The situation is less clear cut when I compare the gamut of my NEC PA241W with the printer. The NEC's colour gamut is close to Adobe RGB and the printer only beats it in the deep cyans. But the 241W is a pretty darn good monitor.

For all that, the point simply was that the monitor is a critical part of the camera to print workflow. It's also potentially the weakest part. If it isn't adjusted (calibrated and profiled, and reasonably regularly recalibrated and reprofiled) to display colours as accurately as it can, a photographer will be editing in the dark, as it were, and monitor print comparisons are almost bound to be poor.

Cheers

Tim

[ajohnw]

This is a long video but does go through photographic gamuts rather well.

http://www.youtube.com/watch?v=n0bxSD-Xx-Q

What it doesn't cover well is why. aRGB and colour contrast primarily in prints. The video does show that aRGB has coarser spacings between colours than sRGB so actually misses some. Other sites will point out that it is possible to see that these colours are missing on a monitor. Prints and monitors have their own distinct problems.

Little if anything is ever mentioned in relationship to the eyes sensitivity to colour.

John
-

[Richard Lundberg]

Coarser spacing means you should use a 16 bit workflow.
Gamuts are usually shown at the 50% point in 2D and this doesn't represent the shapes very well. Printer gamuts are more like a rock and the device independent gamuts more like diamonds.
I think it's time for someone to create a color space with a very deep green and imaginary red and blue primaries in 18 or 20 bit, just to get ahead of the technology.

[ajohnw]

How ever many bit's in the gamut there is this problem to get round on monitors. This is a full sRGB grey scale in steps of 6, 7, 6, 7 etc with full white at the start just to make the 1st 2 steps a little more difficult to see. They are easier to see like this anyway as they aren't buried somewhere in a brighter picture.



Basically if monitors had 16bit colour channels steps of 6,7 bits at that scale would not show any visual difference. The same applies to colours.

Also if using the coarser aRGB gamut things do not come out correctly on sRGB and the same colours are missing from both when viewed in the wrong gamut.. It seems this is where procolour comes in and some odd PP methods to get round things finishing up being printed on very large gamut printers. Prints have there own problems hence rather bright viewing conditions that are usually mentioned.

As to some one should I think you will find Sony have - in terms of covering the visual spectrum sensibly. No one seems to be sure why but the term digital cinema is mentioned.

John
-

[Manfred M]

John the coarser separatations only make sense; if we are dividing a broader gamut into the same number of pieces as a narrower one, then the pieces must obviously be coarser. That being said, 8-bits per channel still gives you 16 million separate colours, and if best case our eyes can only detect around 10 million, so there are lots of colours to go around.

No if people say that they can detect breaks in AdobeRGB, I wonder if that is more an inidcation that there is an issue with the reproduction method that the smoothness of the colour space representation.

[ajohnw]

There is a lot of hog wash about Manfred.

Take we can see 10 million colours. That is based around us only being able to detect colour changes over a certain value of delta E. If 10 million is correct it's 10 million over the entire colour range of human vision. Neither aRGB or sRGB come any where near covering that gamut.

The problem with metrics based purely on delta E values is what number to choose. There is no doubt at all that people who work with colour can distinguish smaller changes than those that don't. All I have seen on this subject is that many comments about what is good, perfect etc seem to go back to a certain monitor manufacturer who makes them as good as they can be. On the other hand going back to the grey scale I posted if I approach the steps from the bright end the rgb 0,0,0 doesn't seem right. That's because it is effectively the screen back light leaking through. Delta e 0.04 and the next one 0.19, then 0.27. I shouldn't be aware of this according to some but I can see something is off. The software also does single r g b scales using steps of 12. I am more aware of the green and blue tinge than the red at this level. Green probably because our eyes are most sensitive to that colour and red, blue down to colour temperature correction. Lots of blue as our eyes aren't so sensitive to that so needs a lot and red cut to balance the native colours the monitor produces. Why some interest in this area. I had an interest in using power LED illumination in microscopes. Many people use day light types. I wouldn't touch one ever again due to the amount of blue they emit to achieve the colour temperature. They affected my colour vision for several hours. Even warm white ones around 5000k have to emit a lot of blue - more or less the same as the green level.

16bit processing is needed for aRGB to prevent banding during PP. The idea is to never loose any values. When a value is changed it's still presented as 8bit but the hidden bits remain so can be changed further without sudden jumps. I have only seen the effect once. Modifying one of Christina's jpg's with lots of blue sky. I suspect that a high compression ratio had been used so effectively the information needed to make changes was lost. Picture looks fine as it is but large areas have the same colour value so contrast type changes band. I usually work on 90 or 95% quality jpgs and rather large changes can be made.

There no getting round the fact that either of aRGB or sRGB can show shades of colour that the other can't. I suspect that if some one went into it more deeply they would find that sRGB meets the need of critical eyes more closely. aRGB is more aimed at printing and uses colour range as a substitute for contrast.

Personally I feel things will finish up with true 10 bit colour channels but that greens will as always be limited unless 2 are available. If not many recent changes in software are a bit pointless. Open source has decided to PP with 32 bit floating point colour channels for instance. That probably makes increased colour depth irrelevant for ever. Windows has been able to handle deeper colour for a long time now as well. Colour management is generally standard. Perhaps it will be 10bit aRGB like with the granularity of sRGB or something that doesn't relate to either. I get the feeling that existing monitor technology could cope with this. It already does but driven via 8 bits and on the face of it could finish up with 4 times more colours than sRGB driven via 10 bit.

Couple of links if people are interested. One brief, one long. Neither from memory even touch on the complications of the eyes sensitivity to colour. Best bear that aspect in mind as some styles of presentation normalise it all to 1. So



It could be argued that when it has fell to less than 10% viewers get little information. Various august people seem to carry on theorising though with no mention of this aspect. Actually it's not that good a graph. Blue sensitivity has a bit of a knotch/dip in it. Seems to be close to the blue power leds emit. Sort of inverse anthropic principle. There are also sunlight spectrum intensity graphs about. Much more anthropic.

http://dot-color.com/category/color-gamut-standards/

http://www.tftcentral.co.uk/articles/pointers_gamut.htm

It's all a horridly complicated subject.

John
-

[Richard Lundberg]

The Sony people are implementing Rec 2020 for UHDTV and the new cinema cameras. There are lots of references for Rec 2020, which has no imaginary primaries and reaches deep into green. Your tftcentral link , which I was about to post, is very comprehensive. The CIE research said 2.4 million colors back in 1931. I haven't examined that in depth. I suppose some discrepancies arise from luminance levels.

[Manfred M]

The CIE research said 2.4 million colors back in 1931.

1931 data?? I'm not sure I would expect data that old to represent more than some consensus approach, rather than a more technologically sophisticated approach to make the measurement.

[ajohnw]

The tftcentral link goes through what will eventually happen, the one he hopes catches on, but I have my doubts about 12 bit. Once 10bit panel numbers go up prices will come down and maybe some will start working on 12 but given the small picture effects related to the eye and luminance problems I'd have to wonder if 12bit was worth while. What surprised me about the article is that the base rgb colours seem to be obtainable.

John
-

[animalnz]

I'm sorry if I missed it in what is an interesting but somewhat wandering thread but I couldn't find a succinct answer to the OP. IMHO, the problem is likely to simply be that the OP does not use a printer profile to soft proof her work. Early in the thread someone mentions that a printer profile has information about the paper stored in it, which is true, but it had so much more. It tells your software what colors the printer will actually produce from your file using the process, inks, dyes, paper, lighting and viewing conditions used to make the profile.

Soft proofing takes that information, uses the method you have chosen (perceptual, relative etc) and does is best to display, on the screen, how the print will look on paper. That result will always look flatter than the raw screen image. It's the difference between transmission (screen ) and reflective light. It's how it should be and it gives you the possibility of adding more punch before you upload if that's what you want.

I suspect that the lab applies a color profile to your print when you upload it and displays the corrected print for you so that you get the best possible idea of the print you receive. It's what I would do. I may however be wrong about their process.

No matter what, as others have said, your best chance of getting a print that is a good match between screen and print is to use a calibrated monitor and the facility in your software to soft proof using an up to date profile provided by your printer.

Stopping now before I have to redefine succinct.

[Manfred M]

I'm sorry if I missed it in what is an interesting but somewhat wandering thread but I couldn't find a succinct answer to the OP. IMHO, the problem is likely to simply be that the OP does not use a printer profile to soft proof her work. Early in the thread someone mentions that a printer profile has information about the paper stored in it, which is true, but it had so much more. It tells your software what colors the printer will actually produce from your file using the process, inks, dyes, paper, lighting and viewing conditions used to make the profile.

Soft proofing takes that information, uses the method you have chosen (perceptual, relative etc) and does is best to display, on the screen, how the print will look on paper. That result will always look flatter than the raw screen image. It's the difference between transmission (screen ) and reflective light. It's how it should be and it gives you the possibility of adding more punch before you upload if that's what you want.

I suspect that the lab applies a color profile to your print when you upload it and displays the corrected print for you so that you get the best possible idea of the print you receive. It's what I would do. I may however be wrong about their process.

No matter what, as others have said, your best chance of getting a print that is a good match between screen and print is to use a calibrated monitor and the facility in your software to soft proof using an up to date profile provided by your printer.

Stopping now before I have to redefine succinct.

I think you are being optimistic if you think this problem can be solved by soft proofing.

I agree, the first step is to have a calibrated and profiled screen and you need to adjust your image to get it to look right and need to follow a colour-managed workflow through to the printing step; i.e. have the profile for the printer and specific paper that will be used.

Soft proofing is an emulator that attempts to use a transmitted light, additive, RGB process to show the user what the output of a reflected light, subtractive CMYK process. While I understand the theory, I've never seen a lot of difference in the colours from soft proofing versus what I see on my screen using a photo printer. If I look at press proofs, then yes, there can be some siginficant differences.

[animalnz]

Hello Grumpy Diver. Not sure if I've understood you correctly. "A lot" is a pretty subjective measure but if you mean that the image displayed without soft proofing isn't very different from the image on the monitor with soft proof on then I think it depends.

If your workflow is entirely within the gamut that your printer can produce, and the paper you are using had a very bright white point then, yes, there will probably be little difference. But my little may be your enormous.

I agree entirely about the difficulties of representing a subtractive model with an additive light source. I've been reminded of just how difficult it is lately while printing to metallic papers.

I do note however that some posters have difficulty seeing any difference between the OP's two images. It could just be that the OP's "lot" may be other posters "little".

[Richard Lundberg]

The OP lost interest weeks ago. Hope she resolved her problem.