Another "My print doesn't match what's on the screen.."

[D Fosse]

When the author later says that the choice of a low gamma "produces noticeably lighter images on screen" that seems contradictory.  It seems he is ignoring the whole concept of a color space

Indeed he is. The article is really about monitor calibration from a non color management perspective. Most of the story is missing here. What's missing is the description of the monitor in its calibrated state, i.e. the profile. There's no color management here because there's no profile-to-profile remapping, and then yes, gamma makes a difference. Just as the native monitor color space makes a difference, even if it has been "linearized" through calibration.

[digitaldog]

When the author later says that the choice of a low gamma "produces noticeably lighter images on screen" that seems contradictory.  It seems he is ignoring the whole concept of a color space, which specifies the desired conversion of digital values into brightness or color.

He is probably referring to non ICC (color managed) applications. In ICC aware applications, the gamma's can all be different. Non ICC aware app's don't know what a color space is.

[hjulenissen]

I sure would be handy if my OS detected "non ICC aware apps" and switched the monitor into sRGB-emulated mode automatically.

As is now, I tend to view websites and videos with really off colors (over saturated). Too lazy to dive into the monitor menues, switch to sRGB mode, then back again when working in Lightroom.

-h

[Wayne Fox]

I take the simple approach. The calibration white point targets are set to match the paper - the perception of which obviously varies with the ambient light. I usually end up in the neighborhood of 6200K and 110 cd/m², but that's just me.

But from there on in, the display is king. That's my reference for what the file should look like.

Gamma doesn't really matter in a color managed workflow, the file is remapped into monitor gamma by the two profiles anyway. So my philosophy is just to stay close to the monitor's native gamma (2.2).

I use a similar approach. Printing a standard print that produces a known result (and which I have reference prints available that I have printed in the past)  I can verify my printing pipeline is producing a correct print.  If the print does not match my display, the only choice is to modify the viewing condition of the print to match (very hard to do some times), or I can vary the calibration settings for the monitor profile. This is assuming the viewing conditions of the print are adequate and consistent -  I use Solux 4100k bulbs in my viewing setup. However I also have a small GTI viewing station that holds smaller prints, and I feel that in either lighting condition the prints I produce now are a very close match.

I find it difficult to see the "white" point difference between a piece of paper vs a white document in Photoshop on the display to a great degree of accuracy, although I find luminance for this works pretty good.  I print a standard print and will tweak the white point of my calibration software until I get a visual color match between the display and the print.  This is almost always in the 6100k range or so.  If at 6500 my prints are warm.  Not unacceptably so, but they do not match the display.

With the SV software on my 301w, I do this by manually setting the white point ... using the xy adjustment.  It took 4 or 5 attempts to achieve a profile but now I'm quite confident that my print will be very close to what I'm seeing.

With the 301w I also set the contrast to 300:1 which seems to give me a better match in shadow detail as well.

[Rhossydd]

Here is a very good authoritative article.

Which doesn't suggest doing what you've suggested in reply #5 "First though, consider finding a "test image" of known character on the web, and print it.  Adjust your print driver to make that print look right."

How could you adjust a print driver to make a print "look right" if you're not using a reliable monitor to know what it should be ?

Now if you can find any credible reference in colour management literature to that workflow please supply it.

[Texas308]

Here's an update:

When I go in and change the profile used within LR 5 "Print Job" section to use "Epson Stylus Pro 3880_3885_3890 PremiumLusterPhotoPaper" the print comes out MUCH BETTER and along the lines of what I was looking for in a print.

[Manoli]

When I go in and change the profile used within LR 5 "Print Job" section to use "Epson Stylus Pro 3880_3885_3890 PremiumLusterPhotoPaper" the print comes out MUCH BETTER ...

What ICC profile were you using previously ?
or did you have  Print Job > Color Management > Profile: 'Managed by printer' selected ?

[Texas308]

I was using the "PA301W 25106369UW 2013-10-20 21-41 D65 2.20" profile... 

[Slobodan Blagojevic]

 .

[Manoli]

I was using the "PA301W 25106369UW 2013-10-20 21-41 D65 2.20" profile...  

This sounds like a monitor profile. What you should be using is an ICC profile for your chosen printer/paper combination. For the Epson 3800 that would be something like ' Pro38 PLPP ' . Not sure what it is for the 3880.

[digitaldog]

It is the display profile for the NEC.

[Floyd Davidson]

Interesting article.  The author does indeed advocate choosing some basic settings {brightness, gamma, color temperature} to make the monitor match an eventual print.  However, he also states in the very first paragraph that the purpose of monitor calibration is to match a digital reference, not an eventual print: "Proper calibration guarantees that the image shown on screen matches the numerical color data saved in the digital file."  

The sentence you quote says exactly the same thing as use a "standard image".  Lacking the hardware to do a proper calibration, the OP can download a standard image and use that to get the correct "numerical color dtata saved in the digital file".  Granted it is better to actually measure it the output, but lacking the equipment the OP can get something workable without it.

When the author later says that the choice of a low gamma "produces noticeably lighter images on screen" that seems contradictory.  It seems he is ignoring the whole concept of a color space, which specifies the desired conversion of digital values into brightness or color.

He is precisely correct.

Does monitor calibration preserve the white point defined by monitor brightness and color temperature settings?  Will a monitor profile ever result in a display of pure white (RGB=255,255,255) with some pixels not at maximum brightness?  How does the chosen color space enter into this?

If monitor calibration preserves this white point, it makes sense that you want to set monitor brightness and color temperature to match an eventual print, under whatever normal lighting conditions you choose for print display.  

Monitor calibration will then take care of displaying non-white colors and gray levels correctly, relative to this white point.  This requires a digital reference (and color space) and has nothing to do with a print.  Similarly, printer calibration matches printer output to a digital reference (and color space) and has nothing to do with a monitor.  The way monitor calibration influences the final print is by inducing the user to edit the image until the monitor image looks good.  

Good questions, but that is for another discussion.  The OP doesn't have the equipment to generate a monitor profile.

[Floyd Davidson]

Which doesn't suggest doing what you've suggested in reply #5 "First though, consider finding a "test image" of known character on the web, and print it.  Adjust your print driver to make that print look right."

He recommended doing exactly that!  The article "encouraged all digital artists to invest in a monitor calibration system" which is to say, get a device that produces a standard test image.  The OP doesn't have the hardware needed, but that is a different discussion.  He can download a standard test image in a few seconds and get useful results that are very educational.  

How could you adjust a print driver to make a print "look right" if you're not using a reliable monitor to know what it should be ?

Now if you can find any credible reference in colour management literature to that workflow please supply it.

If you understand what a standard test print is and what it shows there is no need to compare it to a monitor.  However, if that is difficult for whatever reason, then one method would be to have it printed by a commercial print shop to then have a known "correct" print to use for comparison.

Edit:  Here are two URL's of interest that Wayne Fox posted to a different thread.  A very good "standard image", but also a nice discussion about each part of it and how it is useful.

  http://www.outbackphoto.com/printinginsights/pi048/essay.html
  http://www.outbackphoto.com/printinginsights/pi049/essay.htm

[hugowolf]

The OP doesn't have the equipment to generate a monitor profile.

The OP does. He is running SV (NEC's SpectraView calibration and profile generating software) which comes with NEC monitors when you buy the associated hardware puck.

Brian A

[Floyd Davidson]

Here's an update:

When I go in and change the profile used within LR 5 "Print Job" section to use "Epson Stylus Pro 3880_3885_3890 PremiumLusterPhotoPaper" the print comes out MUCH BETTER and along the lines of what I was looking for in a print.

The profile changes the monitor to make it look like the printer output.

Obviously that profile is specifically for use with the 3880 and Epson's Premium Luster paper.  You may or may not be able to find a profile specifically for the paper you want to use.   

This is much like the issue of monitor calibration, where by far the best solution is to purchase the necessary hardware equipment and generate your own printer profiles.  But absent that you can again "download a standard image" that will put you in the ballpark.  In this case the "standard image" is hopefully a profile specifically for the paper type and the printer model you use, but it that can't be found you'll probably be able to find a profile for a different but similar paper that works.  Even for different printers may be close enough, if they use the same ink set, and hence a profile for an Epson 4880 or 7890 might do if nothing else can be found.

[Floyd Davidson]

The OP does. He is running SV (NEC's SpectraView calibration and profile generating software) which comes with NEC monitors when you buy the associated hardware puck.

Brian A

Okay.  Thanks, I had missed that. (I didn't know what "SV" was.)

That just makes it easier, and more accurate.  Now that he is using something close for a printer profile, he might want to go back to trying different brightness, gamma, and color temperature parameters when he calibrates the monitor to start with.  Once again perhaps reading Dr. Perkins' article cited previously... which hopefully will make more sense after reading this discussion.

[Schewe]

Okay.  Thanks, I had missed that. (I didn't know what "SV" was.)

I guess you also missed the fact that the OP was trying to print out while using his display profile as the output profile? That's why his prints looked like crap.

[Schewe]

Here is a very good authoritative article.  Note that monitor brightness, contrast, and color temperature are all set manually.  The choices made are intended to provide a monitor calibrated to match a well calibrated printer.

http://blogs.scientificamerican.com/symbiartic/2012/01/17/how-to-calibrate-your-monitor/

Well, I finally read the article and I'm sorry to say that Jim has a lot of stuff wrong (or mostly wrong with enough right that it may escape a casual read). Jim claims that Apple has had a default gamma of 1.8 and that was only changed with Snow Leopard (10.6.. Sorry, but Apple quit "defaulting" to 1.8 about a decade ago (once Apple quit selling Laserwriter printers).

Jim's also wrong'ish regarding the current thinking (in fact, a lot of Jim's thinking is pretty old school) about an optimal white balance. The further you try to move a display off it's native state, the more work the profile has to do...this is particularly true of any display that doesn't have internal calibration settings like the NEC has.

The other thing he doesn't mention at all is luminance targets for display calibration and profiling...so he's ignoring 1/3 of the display triad of major, critical factors.

Sorry, while Jim certainly comes off as "scholarly" (he is, after all, an academic) I'm not real sure he has a lot of real world experience (he even admits the only printer he has is a cheapo inkjet printer that came with his computer–that was in his first post).

Can you calibrate and profile a display to D50, Gamma 1.8 and get good results? Yes...but with current LCDs, unless you have a display that can calibrate internally (or you are using a 10-bit display pipeline and Photoshop on Windows) trying to calibrate a current LCD to D50 and gamma 1.8 is gonna be less accurate if your display pipeline is 8-bit/channel.

I have some friends at RIT (my alma mater) and I think I'll need to ping them and see if I can get in touch with Jim and teach him a few things...

So, a "very good authoritative article"? Nope, it's a very basic and somewhat flawed article at best. Sorry...

Now, if you want to learn how to properly calibrate and profile a display, just ask. Andrew (or I) are willing to help. But the very idea that you should wank on a display calibration to make your image on screen look like your print went out back in the old EFI Color Cachet which was around in the early 1990's. Modern thoughts are to calibrate to a standard: which I suggest be in the D65, gamma 2.2 with luminance output at about 150 cd/m² and match your viewing environment to match your display environment so the print can match your display (not the other way around. Oh, for these of you with NEC displays, I also suggest using a 250/1 contrast range which more closely matches the contrast range you can get off a glossy print.

:~)

[Floyd Davidson]

I guess you also missed the fact that the OP was trying to print out while using his display profile as the output profile? That's why his prints looked like crap.

I "missed" that, and so did you!  The OP posted that 25 messages into this thread...

Of course the procedure I recommended would have made it the obvious cause very quickly too!  Printing a test print to verify color management is, as I and others have suggested, a very useful tool.

[Floyd Davidson]

Well, I finally read the article and I'm sorry to say that Jim has a lot of stuff wrong (or mostly wrong with enough right that it may escape a casual read). Jim claims that [...]"

I'm never impressed with disrespect for someone who clearly does have significant credentials and knowledge about the topic of discussion.

Professor Jim Perkins has worked with the basis for how color management is done, and seems to understand the theory as well as some one specific implementation.  I may or may not agree totally with each and everything he said, but I have a great deal of respect for each point he makes.

Here's an example:

Jim's also wrong'ish regarding the current thinking (in fact, a lot of Jim's thinking is pretty old school) about an optimal white balance. The further you try to move a display off it's native state, the more work the profile has to do...this is particularly true of any display that doesn't have internal calibration settings like the NEC has.

Your statement of theory is exactly correct, but you seem to think that supports your set of parameters rather than those given in the cited article.  In fact, since gamma 1.8 and D50 will be closer to what the display will ultimately be, the profile will at best have less work to do, and at worst exactly the same work, than if your gamma 2.2 and D64 are the initial configuration parameters.  Jim Perkins is precisely correct!

Sorry, while Jim certainly comes off as "scholarly" (he is, after all, an academic) I'm not real sure he has a lot of real world experience (he even admits the only printer he has is a cheapo inkjet printer that came with his computer–that was in his first post).

He comes off as being very credible, and doesn't try to twist facts or other people's words.  He said he uses only one printer for his "professional work" as an artist.  We can also assume that he has access, in his academic role, to all of the fancy printers he has mentioned not needing to own personally.  Given his discussion it is pretty clear that he knows exactly what he's talking about, and stands on what he says rather than trying to denigrate those who might disagree.

I have some friends at RIT (my alma mater) and I think I'll need to ping them and see if I can get in touch with Jim and teach him a few things...

Learning is a better goal.

So, a "very good authoritative article"? Nope, it's a very basic and somewhat flawed article at best. Sorry...

Now, if you want to learn how to properly calibrate and profile a display, just ask. Andrew (or I) are willing to help.

Anyone wanting to learn should certainly read what you have to say, and should pay attention to the discussion it generates.  But they should also pay particular attention to more credible sources, such as Jim Perkiins, and weight their thoughts on the subject too.