A classics revisited: My prints turn out too dark!

[nilo]

sure... confused - sorry!

while we're at it:
do you have the impression that the monitor is too bright all over the tonal range... or rather that the dark and mid tonal values are too bright (compared to the print, of course) wheras the white is okay?

It's more the darks and the mid tonal values that are brighter on the screen than on the print. And highlights are whiter on the paper. The print seems to have more contrast and saturation than the screen. Especially yellow is more saturated in the print. The dark blues and greens area little bit lighter on the screen. All  this seems to cause a slight blueish color cast, which is noticeable on all skin tones. But grey scales appear neutral.

[nilo]

Sometimes (depends on the paper of course, I usually only work with two) but the NEC SpectraView makes loading different target calibrations and their associated ICC profile a one button affair.

But the new NEC PAs offer even greater possibilities: You can actually load the paper-profile into the 3DLUT of the monitor, and also change between different  simulation with a click of the mouse.

That should give new perspectives to the above described work flow?  Especially regarding soft proofing?

[tho_mas]

It's more the darks and the mid tonal values that are brighter on the screen than on the print. And highlights are whiter on the paper. The print seems to have more contrast and saturation than the screen. Especially yellow is more saturated in the print. The dark blues and greens area little bit lighter on the screen. All  this seems to cause a slight blueish color cast, which is noticeable on all skin tones. But grey scales appear neutral.

My guess is that the calibration curve down from a certain luminosity is off... This can have several reasons. For instance it can result from a bad black calibration of the puck prior to the actual calibration.
But first let's check this...:
your calibration target is Gamma 1.8. ... please create a smooth gradation in Photoshop from black to white in ProPhoto RGB (or ECI-RGB or any Gamma 1.8 working space...). Now assign (not "convert" to!) your monitor profile. Does the dispersion of luminance remains the same or does the gradation change?

edit: and mabye you can post a validation chart... like this:

[nilo]

My guess is that the calibration curve down from a certain luminosity is off... This can have several reasons. For instance it can result from a bad black calibration of the puck prior to the actual calibration.
But first let's check this...:
your calibration target is Gamma 1.8. ... please create a smooth gradation in Photoshop from black to white in ProPhoto RGB (or ECI-RGB or any Gamma 1.8 working space...). Now assign (not "convert" to!) your monitor profile. Does the dispersion of luminance remains the same or does the gradation change?

edit: and mabye you can post a validation chart... like this:

Thank you Tho_mas for your patience and for taking the time to sort this out!

The dispersion of luminance remains the same, when in PS assigning the monitor profile to a ProPhotoRGB smooth gradient.

Regarding the validation chart, I don't have software to produce one. i1Match does not recognize the NEC monitor profile. SpectraviewII does validations, but gives not report chart... Maybe BasICColor?

regards

nino

[tho_mas]

SpectraviewII does validations, but gives not report chart...

so maybe you can post a screenshot of that validation?
and, again, which puck are you using?

[nilo]

so maybe you can post a screenshot of that validation?
and, again, which puck are you using?

Here is a screen shot of two measurements with the NEC MDSVSENSOR and one with the DTP94B. Very interesting indeed! I made all the profiles until now with the NECMDSVSENSOR that I especially bought because it is supposed to give better results with the wide gamut monitor as it is custom mated to the NEC wide gamut monitors...

EDIT added the attachments 

[tho_mas]

Now, basically those validations do not tell a lot in terms of "absolut" DeltaE … but they tell something about the difference of target->profile based on the measurment with the same puck.
For my taste the DeltaE of the greyscale is much too high. On a higher quality display the greyscale should be around 0.5 or so. In any case clearly below 1 DeltaE.
Not sure… but maybe the color shift you are seeing comes from the inconsistent greyscale.

Also not sure what to do here without seeing the actual monitor.
So just some random thoughts to consider WRT to calibration…

- reset the monitor and make sure that any "auto" feature is disabled (such as ambient light compensation etc.)

- monitor should run 2 hours prior to calibration

- the puck should be warmed up on the display. The i1Displays (the NEC is based on the i1Display, correct?) may very well need 15 minutes to warm up (mine definitely needs 15 minutes).
Why? The devices get more sensitive to dark tonal values when they warm up. So if they warm up during the calibration you may get a inconsistent calibration curve.
(for some reason this does not apply to the DTP94… at least not to my copy).

- black calibration. When the software is asking you to put the device on a flat surface (or so)… do not use a flat surface. Black Molton (or similar) is much better. Whatever you use, make sure that absolutely no light creeps under the puck.
Why? When the puck measures a black value during monitor calibration that is below the black level of the initial calibration the entire greyscale can be screwed up.

- you have certainly seen sometime when pressing your thumb on an LCD it will show a kind of "glow" that disappears after some seconds.
Now, the puck may also produce that kind of glow from its own weight when it is sitting on the display (which is ideally tilted back so that the puck sits securely on the LCD).
So make sure that you either don't move the puck as long as it sits on the LCD, or - if you have to remove it temporarily - that you place it on exactly the same position (otherwise it may measure in the area of the "glow").

If you can't get better results for the greyscale I'd contact your dealer. Maybe you have to exchange the monitor...

edit:
here are the greyscale values from the above attached validation of my 3 year old Eizo CG241W after 9200 hours of use.
Your brand new NEC should be able to come close.

another edit: I may very well have overlooked something! So take it with a grain of salt...

[nilo]

Now, basically those validations do not tell a lot in terms of "absolut" DeltaE … but they tell something about the difference of target->profile based on the measurment with the same puck.
For my taste the DeltaE of the greyscale is much too high. On a higher quality display the greyscale should be around 0.5 or so. In any case clearly below 1 DeltaE.
Not sure… but maybe the color shift you are seeing comes from the inconsistent greyscale.

Also not sure what to do here without seeing the actual monitor.
So just some random thoughts to consider WRT to calibration…

- reset the monitor and make sure that any "auto" feature is disabled (such as ambient light compensation etc.)

- monitor should run 2 hours prior to calibration

- the puck should be warmed up on the display. The i1Displays (the NEC is based on the i1Display, correct?) may very well need 15 minutes to warm up (mine definitely needs 15 minutes).
Why? The devices get more sensitive to dark tonal values when they warm up. So if they warm up during the calibration you may get a inconsistent calibration curve.
(for some reason this does not apply to the DTP94… at least not to my copy).

- black calibration. When the software is asking you to put the device on a flat surface (or so)… do not use a flat surface. Black Molton (or similar) is much better. Whatever you use, make sure that absolutely no light creeps under the puck.
Why? When the puck measures a black value during monitor calibration that is below the black level of the initial calibration the entire greyscale can be screwed up.

- you have certainly seen sometime when pressing your thumb on an LCD it will show a kind of "glow" that disappears after some seconds.
Now, the puck may also produce that kind of glow from its own weight when it is sitting on the display (which is ideally tilted back so that the puck sits securely on the LCD).
So make sure that you either don't move the puck as long as it sits on the LCD, or - if you have to remove it temporarily - that you place it on exactly the same position (otherwise it may measure in the area of the "glow").

If you can't get better results for the greyscale I'd contact your dealer. Maybe you have to exchange the monitor...

edit:
here are the greyscale values from the above attached validation of my 3 year old Eizo CG241W after 9200 hours of use.
Your brand new NEC should be able to come close.

another edit: I may very well have overlooked something! So take it with a grain of salt...

I will do as you say.

BTW, I forgot to mention that I made the test prints with more than one printer. It was the 3880 and the 9900. I also used more than one paper and profile. It was Canson Platine and Baryta Photographique, as well as Hahnemuhle PR Baryta and PR Pearl. The cast was an issue on all the papers and profiles. So I really think that there is a problem with the PA241W. Unfortunately I can't get it back to the dealer, as I had to bring it over by plane from Europe, because the local dealer does not sell those screens (When I asked them they told me a price that was something outrageously ridiculous)

[tho_mas]

Unfortunately I can't get it back to the dealer, as I had to bring it over by plane from Europe, because the local dealer does not sell those screens (When I asked them they told me a price that was something outrageously ridiculous)

is there no NEC service in your country or so? Maybe give it a try...

Have you tried to create an LUT profile? Maybe this will improve things.
If the NEC software doesn't create table based profiles take BasICColor Display...
Last but not least you can also try to hardware-calibrate with the NEC software and additionally sofware-calibrate with BasICColor Display (LUT, 16bit) to the same targets... resp. to the "native" white point, to the "min. black point" (in this case, as the NEC software already set the black point...) and to Gamma 1.8.

[nilo]

is there no NEC service in your country or so? Maybe give it a try...
[...]

I'll try that, but they do not sell graphic high-end monitors here, and from my experience with the local distributor, they do not know what they talk about.

Have you tried to create an LUT profile? Maybe this will improve things.
[...]

? I don't know what exactly you mean!

[tho_mas]

Have you tried to create an LUT profile? Maybe this will improve things.

I don't know what exactly you mean!

In BC Display the 5th subitem of "adjustments", i.e. in the "profile" tab you can choose to create a TRC matrix profile (which is mostly okay) or to create a table based profile (LUT = "look up table").
In Quatos iColor Display it's the last menu "save profile as". Here you can also choose to safe the profile as matrix or LUT profile.
Possibly an LUT profile will improve things.

As to my "last but not least" advice above... maybe it's better to use iColor Display instead of BC Display for the additional software calibration as you can choose the correction table "generic IPS wide gamut" here.
But first I'd try to create an LUT profile with BasICColor in hadware calibration mode (normally BasICColor supports hardware calibration for all the NEC displays... not sure if the latest update already supports your model).

[nilo]

I don't know what exactly you mean!In BC Display the 5th subitem of "adjustments" in the "profile" tab you can choose to create a TRC matrix profile (which is mostly okay) or to create a LUT profile (LUT = "look up table").
In Quatos iColor Display it's the last menu "save profile as". Here you can also choose to safe the profile as matrix or LUT profile.
Possibly an LUT profile will improve things.

As to my "last but not least" advice above... maybe it's better to use iColor Display instead of BC Display for the additional software calibration as you can choose the correction table "generic IPS wide gamut" here.
But first I'd try to create an LUT profile with BasICColor in hadware calibration mode (normally BasICColor supports hardware calibration for all the NEC displays... not sure if the latest update already supports your model).

I was about to ask you why you preferred BC to Quato. There is only one thing. I have a European monitor, and I already double checked with BC, it does not support hardware calibration , because it has been blocked (which is not the case for the North-American models). SO I think I'll have to go with Quato and software calibration, if the above described retry and routine does not work out.

BTW we have this NEC because we had issues with the EIZOs 

[tho_mas]

I already double checked with BC, it does not support hardware calibration , because it has been blocked

man, those things are really totally unnecessary and annoying!

SO I think I'll have to go with Quato and software calibration, if the above described retry and routine does not work out.

yes. But first have a look if the NEC software can create an LUT profile. Only if not I'd try the "dual" calibration method...

[nilo]

Tho_mas,

Here are the calibration and validation DeltaE results obtained with the DTP94b

[tho_mas]

looks much, much better now!!
does it also look better visually?
A smooth gradient black to white in Photoshop looks good, i.e. shows no banding?

Did you create a LUT profile or is this just the result of my advices WRT to black calibration etc. ?

Now, it will get even better!
The blacks still look off way too much.
What black level did you set as target?

[nilo]

looks much, much better now!!
does it also look better visually?
A smooth gradient black to white in Photoshop looks good, i.e. shows no banding?

Did you create a LUT profile or is this just the result of my advices WRT to black calibration etc. ?

Now, it will get even better!
The blacks still look off way too much.
What black level did you set as target?

I simply redid a calibration/profiling with SpectraViewII applying all the advices. A few major differences to my prior tries with the PA241W:

  - I warmed up the puck (I wanted to believe that, when placed on the screen right before the calibration/profiling, and the profiling being very quick, that there was no need to warm the puck up).
  - I used the DTP94b instead of the NEC MDSVSENSOR, NEC's custom mated i1d2 (I resisted this replacement because I paid so much and had so much trouble to get this NEC puck over from America. And it is supposed to be better!)
  - In the preferences I checked "Avarage low luminance measurements" ("Improves the accuracy of low luminance measurements by taking several measurements and averaging them."
  - I set 52 calibration and profile steps instead of 32.
  - I changed the calibration priority from "maximize contrast ratio" to "Best grayscale color tracking" (That was a winner I think?!)
  - I used the option "extended luminance stabilization time"
  - all other auto's are off

I can't see no banding on a gradient.

The test images show the same kind of difference to the print. The  screen is a little bit too bright and has a yellow color cast.

As Black Level I choose "Monitor Default", that's around 450.


I just validated the calibration again, and again and again. This time with the NEC puck, and things look like this:

White point Delta E: 2.95, 3.32, 3.12, 3.79
Maximum Grayscale Delta E: 2.95, 3.32, 3.12, 3.79
Average Grayscale Delta E: 2.19, 2.67, 2.55, 2.89


I have a third colorimeter, a regular i1D2. Here is the validation:

White point Delta E: 8.28, 7.64
Maximum Grayscale Delta E: 8.28, 7.64
Average Grayscale Delta E: 6.63, 6.17

New validation with DTP94b:

White point Delta E:  0.29
Maximum Grayscale Delta E: 1.09
Average Grayscale Delta E: 0.65


Hmmm...


So I did a new calibration with the NEC puck, and Delta E is at an average 0.86 and a max 2.86 (including dark values). DTP94b validation is White Point DeltaE:2.87, Max Grayscale DeltaE:2.87, Average Grayscale DeltaE: 2.04

 

To be honest, for soft proofing purposes, or any critical color work that is, this screen is not usable. Mainly because of the yellow cast. As I said, prints, on different printers and with different profiles, consistently come out darker and blueish, magenta.

[nilo]

Thomas,

I followed your advice and lowered the contrast a bit, to approximately 0.4. Consequently the Delta E readings of dark values got better. A got averages as low as 0.4-0.5. As a matter of fact the bad deep blacks, just got cut off.  Superstitiously, I also covered the monitor with a viewing cloth from the large format camera, though that wouldn't be necessary I think, as the digital darkroom is also a big, real old fashioned darkroom, just pitch black.

It took me lot of time, and I head to try that off business hours. The result is very good. Brightness and matches, insofar as the dark values are a bit darker on the print and the light values are bit lighter on the paper. So the paper seems to have, what I would call, a bigger dynamic range. I have the monitor at 125cd/m2 now and the gamma at 2.2.

The yellow cast of the screen, making the paper appear bluish and cold, got almost solved by again lifting the temperature. Its now 6000K and good match for now.

There is one last thing which I can't correct. That is that the print seems to show more red everywhere, more saturated reds, in the shadows, highlights and mid-tones. And the screen definitely lacks the reds.

I hope you have a good advice for this

regards

shaya

[tho_mas]

 - In the preferences I checked "Avarage low luminance measurements" ("Improves the accuracy of low luminance measurements by taking several measurements and averaging them."
  - I set 52 calibration and profile steps instead of 32.
  - I changed the calibration priority from "maximize contrast ratio" to "Best grayscale color tracking" (That was a winner I think?!)

these were the crucial changes, especially the first and the latter are essential.

I just validated the calibration again, and again and again. This time with the NEC puck, and things look like this:

White point Delta E: 2.95, 3.32, 3.12, 3.79
Maximum Grayscale Delta E: 2.95, 3.32, 3.12, 3.79
Average Grayscale Delta E: 2.19, 2.67, 2.55, 2.89


I have a third colorimeter, a regular i1D2. Here is the validation:

White point Delta E: 8.28, 7.64
Maximum Grayscale Delta E: 8.28, 7.64
Average Grayscale Delta E: 6.63, 6.17

New validation with DTP94b:

White point Delta E:  0.29
Maximum Grayscale Delta E: 1.09
Average Grayscale Delta E: 0.65


Hmmm...


So I did a new calibration with the NEC puck, and Delta E is at an average 0.86 and a max 2.86 (including dark values). DTP94b validation is White Point DeltaE:2.87, Max Grayscale DeltaE:2.87, Average Grayscale DeltaE: 2.04

when you are calibrating with device A the validation with device B is absolutely irrelevent! You are wasting time!
Obviously the DTP94 makes the better greyscale, especially in the dark tonal values (which is no surprise).
So if the NEC puck and the DTP94 produce roughly the same results WRT to colors I'd use the DTP94.

I followed your advice and lowered the contrast a bit, to approximately 0.4. Consequently the Delta E readings of dark values got better. A got averages as low as 0.4-0.5. As a matter of fact the bad deep blacks, just got cut off.

Sounds very good!
Do you see what that means?
If you want to calibrate the monitor to (for instance) 5000K but the display has a generic white point of (for instance) 6500K (mostly even bluer in the blacks) the software has to adjust the hardware... i.e. it has to adjust the RGB channels of the monitor from RGB 0-0-0 to RGB 12-8-0 (or whatever... fictitious numbers). Strictly speaking this adjustment does not apply to the actual RGB channels of the monitor (only the white point is adjusted by the RGB channels) but it applies to the RGB channels in the LUT of the monitor.
So a neutral black - without any color cast and consequently whithout such a high DeltaE value - is always brighter than the absolut black (as said RGB 12-8-0 is simply not pure black).
It's up to you whether you prefer a bad calibration curve but higher contrast or prefer an accurate calibration curve but a lower contrast.
I think it's quite clear which one I would prefer.
BTW, a black level of 0.4cd/m2 is absolutely not uncommon for IPS panels. Maybe this new NEC is advertised as an IPS monitor with a lower black level. But if you can utilize the low black level only at the generic white point it's useless... IMO.

You have to get used to the somewhat lower contrast. But I promise you will get used to it after some time.
Now, as to a "match" with the prints, contrast wise, I bet the monitor still has a much, much higher contrast.
If you feel prints and monitor match better when you set the softproof to rel.col. + BPC (for the conversion) but leave the simulation of paper white and even the simulation of black ink unchecked, than go for it. It sounds a bit strange to me, as in re.col. viewing on the monitor (so without SPC) simulate black ink should match better.
But maybe, probably, this is something you have to get used to.
How does monitor and prints match if you move back from the display 1 or 2 meters? Maybe you are just far to close to the monitor. I don't know... you have to find out what works best. As mentioned several times in this thread monitor and prints will never really match 100%. Find out how to get the closest match for you...

The yellow cast of the screen, making the paper appear bluish and cold, got almost solved by again lifting the temperature. Its now 6000K and good match for now.

makes sense.

There is one last thing which I can't correct. That is that the print seems to show more red everywhere, more saturated reds, in the shadows, highlights and mid-tones. And the screen definitely lacks the reds.

I hope you have a good advice for this

yes, but first you should check if all your prints, i.e. all your papers show higher saturated reds.*
If so, you can adjust the colors of the monitor profile. Again, I don't know the NEC software but my software provides the adjustment of 6 color channels (hue and saturation). But actually one would fine tune the monitor profile for each paper individually. This is also not a totally easy task... when you say "red" the magenta and yellow might also play a role here. So boosting the saturation of the red channel will most likely not give you an accurate result. Most likely it will make things worse...
So my advice is to skip this for a while and first get accustomed to your new setup. It's possibly easier (less "destructive") to just create a color layer in Photoshop that you simply add for softproofing purposes, i.e. to simulate the actual print saturation (of course you have to delete or deactivate that layer prior to printing the file !!! ).

*edit: of course you should also check whether these colors are inside the actual gamut of your monitor. If you are talking about reds your display is not able to produce, than any adjustment won't help.
Check it this way: first convert the file to your printer profile (rel.col + BPC). Secondly set your monitor profiles as proof profile and enable color warning.

[nilo]

Thomas,

Thank you so much for your help and advice with this. I learn a lot!

I haven't got enough time now for a detailed reply. I'll be back after the Holidays, on Sunday.

kind regards

nino

[nilo]

Back to work ;-)

Since yesterday, I recalibrated and profiled the monitor to spectacular good results by following your additional advice.

 

Obviously the DTP94 makes the better greyscale, especially in the dark tonal values (which is no surprise).
So if the NEC puck and the DTP94 produce roughly the same results WRT to colors I'd use the DTP94.

Surprisingly I tried the NEC sensor again and got very low DeltaE values, and an even closer visual match.

Sounds very good!
Do you see what that means?
If you want to calibrate the monitor to (for instance) 5000K but the display has a generic white point of (for instance) 6500K (mostly even bluer in the blacks) the software has to adjust the hardware... i.e. it has to adjust the RGB channels of the monitor from RGB 0-0-0 to RGB 12-8-0 (or whatever... fictitious numbers). Strictly speaking this adjustment does not apply to the actual RGB channels of the monitor (only the white point is adjusted by the RGB channels) but it applies to the RGB channels in the LUT of the monitor.

You are right, it defintly improved things!
But are you saying that even in the case of hardware calibration it is preferable to remain close to the generic white poit. BTW what do you mean by generic white point of the monitor.

So a neutral black - without any color cast and consequently whithout such a high DeltaE value - is always brighter than the absolut black (as said RGB 12-8-0 is simply not pure black).
It's up to you whether you prefer a bad calibration curve but higher contrast or prefer an accurate calibration curve but a lower contrast.
I think it's quite clear which one I would prefer.
BTW, a black level of 0.4cd/m2 is absolutely not uncommon for IPS panels. Maybe this new NEC is advertised as an IPS monitor with a lower black level. But if you can utilize the low black level only at the generic white point it's useless... IMO.

yes, so I stay around 0.4 for the moment.

You have to get used to the somewhat lower contrast. But I promise you will get used to it after some time.
Now, as to a "match" with the prints, contrast wise, I bet the monitor still has a much, much higher contrast.
If you feel prints and monitor match better when you set the softproof to rel.col. + BPC (for the conversion) but leave the simulation of paper white and even the simulation of black ink unchecked, than go for it. It sounds a bit strange to me, as in re.col. viewing on the monitor (so without SPC) simulate black ink should match better.
But maybe, probably, this is something you have to get used to.
How does monitor and prints match if you move back from the display 1 or 2 meters? Maybe you are just far to close to the monitor. I don't know... you have to find out what works best. As mentioned several times in this thread monitor and prints will never really match 100%. Find out how to get the closest match for you...

makes sense.
yes, but first you should check if all your prints, i.e. all your papers show higher saturated reds.*

That's the one thing I found out. The paper profile that I was struggling with is definitely not good. I did also ask around a bit locally, and that's what I heard from others too. I had to do quite some printing during the last day, where I tried to check the screen/print match for different papers. As tings are the match is outstanding, except for one paper/printer/ink combination, CIFA Baryta Photographique on 3880/k3vm, and that was the paper I was trying to match all along! So for last 24h of printing I changed the paper.

If so, you can adjust the colors of the monitor profile. Again, I don't know the NEC software but my software provides the adjustment of 6 color channels (hue and saturation). But actually one would fine tune the monitor profile for each paper individually. This is also not a totally easy task... when you say "red" the magenta and yellow might also play a role here. So boosting the saturation of the red channel will most likely not give you an accurate result. Most likely it will make things worse...

tried and true

So my advice is to skip this for a while and first get accustomed to your new setup. It's possibly easier (less "destructive") to just create a color layer in Photoshop that you simply add for softproofing purposes, i.e. to simulate the actual print saturation (of course you have to delete or deactivate that layer prior to printing the file !!! ).

*edit: of course you should also check whether these colors are inside the actual gamut of your monitor. If you are talking about reds your display is not able to produce, than any adjustment won't help.
Check it this way: first convert the file to your printer profile (rel.col + BPC). Secondly set your monitor profiles as proof profile and enable color warning.

thank you again for your help.

From what I understand, the best for me would be to set the monitor to the best possible paper independent state. This should be where I can get the lowest DeltaE. Only, the next step involves soft proofing and print to paper match, right?

regards

nino