Pages: 1 [2]   Go Down

Author Topic: Experiments with Interstitially Nested Grid Profiling Charts and the Z3200.  (Read 3590 times)

deanwork

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2400
Re: Experiments with Interstitially Nested Grid Profiling Charts and the Z3200.
« Reply #20 on: September 06, 2020, 10:29:14 am »

Yes, I see, print the tiff chart. It should be easily read on the Z3200 as the feed mechanism, borders etc are exactly the same on both printers.





I don't have a Z3100, so I'm not 100% sure whether the Z3100 will be able to directly import the Aardenburg 3315 reference file without the old APS software. The latest HP color utility version for your Z3100 may be on a par with the Z3200 such that the old APS software is no longer needed, but any printer can be profiled using your Z3200 as a a standalone automated spectrophotometer by first importing the reference file into the Z3200 driver, then using the "color measurement" features to export this reference file to a tiff image file. The exported tiff file is scaled and arranged so that the Z3200 will be able to scan it later. All you have to do is to print the tiff file with a no color adjust workflow on another printer, in your case, the other printer being your older Z3100.


cheers,
Mark
Logged

kers

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4388
    • Pieter Kers
Re: Experiments with Interstitially Nested Grid Profiling Charts and the Z3200.
« Reply #21 on: September 07, 2020, 08:21:27 am »

Attached is the 3315 Aardenburg Chart I described in the first post to this topic....

Thank you Mark, i will have a look;
I think it will work, or with some minor modification...
-
I have one question still; these targets are all 8bit and so 256 neutral grey patches are the most it can measure for BW.
For Bw it does not make a difference if you have the 918 chart with 256 gray patches or the 3315 chart with 256 grey patches .
Or am i wrong?
Logged
Pieter Kers
www.beeld.nu/la

MHMG

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1285
Re: Experiments with Interstitially Nested Grid Profiling Charts and the Z3200.
« Reply #22 on: September 07, 2020, 11:03:06 am »

...
I have one question still; these targets are all 8bit and so 256 neutral grey patches are the most it can measure for BW.
For Bw it does not make a difference if you have the 918 chart with 256 gray patches or the 3315 chart with 256 grey patches .
Or am i wrong?

A number of technical considerations are packed into this question, perhaps too many to totally unpack, but I'll try to answer the main considerations. First, an 8 bit color target doesn't negate a 16bit CMM workflow. It simply limits the choice of what RGB triplet values are used in the design of the color target to be printed. 8 bits is enough to offer 16 million discreet RGB triplet combinations which in turn are more than humans can observe as discreet colors or would ever use to design a color target, hence more than enough precision. Yet, you are right, out of those 16 million RGB choices, only 256 combinations are theoretically "pure neutral triplets" in the RGB source image color space. But these pure neutral triplets in the source image typically do not print as dead neutrals in the output destination color space, if for no other reason than the media white properties are often not neutral to begin with. This is where adding more near neutral triplets in the color target (or resorting to a two step color profiling approach) is also important because some of those near neutral combinations will ultimately be required to output a uniformly neutral tone scale or even a slightly tinted B&W ramp (e.g. for making warm tone, sepia, cool, or split tone B&W prints) with the chosen printer/ink/media output color space. This is one aspect (but not the only one) where the 3315 Aardenburg chart can improve monochrome printing performance on a color printer if the printer is stable and repeatable enough in its printed output.

I should also note that the Z series HP printers are unique in that HP uses essentially 100% gray component replacement. RGB neutral triplets printed in the no color adjust mode use only the gray inks when printing the color profiling target. Canon and Epson printers don't do that. Thus, HP's approach has an inherent advantage when trying to print accurate neutral and near neutral tones in a full color printing pipeline. It helps a smaller patch count target such as a TC918 color chart to deliver a pretty good B&W print without going to additional heroics, but a larger patch count target like the 3315 Aardenburg chart does refine HP's approach even more. It's subtle, but in my testing, I see it.  Epson, for example, takes a different approach than HP with regard to accurate B&W printing by resorting to an entirely different "advanced Black and white" (ABW) mode for monochrome printing. Nevertheless, HP's 100% GCR approach and its inherently neutral photo gray inks do deliver superior performance when printing B&W images in a full color output mode, and this is why the Z's have a well deserved reputation for making great B&W prints without having to resort to other methods like ABW, Quad Tone RIP, piezography ink sets, etc., to print excellent B&W output.   

cheers,
Mark
http://www.aardenburg-imaging.com
Logged

kers

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4388
    • Pieter Kers
Re: Experiments with Interstitially Nested Grid Profiling Charts and the Z3200.
« Reply #23 on: September 07, 2020, 11:22:52 am »

A number of technical considerations are packed into this question, perhaps too many to totally unpack, but I'll try to answer the main considerations. First, an 8 bit color target doesn't negate a 16bit CMM workflow. It simply limits the choice of what RGB triplet values are used in the design of the color target to be printed. 8 bits is enough to offer 16 million discreet RGB triplet combinations which in turn are more than humans can observe as discreet colors or would ever use to design a color target, hence more than enough precision. Yet, you are right, out of those 16 million RGB choices, only 256 combinations are theoretically "pure neutral triplets" in the RGB source image color space. But these pure neutral triplets in the source image typically do not print as dead neutrals in the output destination color space, if for no other reason than the media white properties are often not neutral to begin with. This is where adding more near neutral triplets in the color target (or resorting to a two step color profiling approach) is also important because some of those near neutral combinations will ultimately be required to output a uniformly neutral tone scale or even a slightly tinted B&W ramp (e.g. for making warm tone, sepia, cool, or split tone B&W prints) with the chosen printer/ink/media output color space. This is one aspect (but not the only one) where the 3315 Aardenburg chart can improve monochrome printing performance on a color printer if the printer is stable and repeatable enough in its printed output.

I should also note that the Z series HP printers are unique in that HP uses essentially 100% gray component replacement. RGB neutral triplets printed in the no color adjust mode use only the gray inks when printing the color profiling target. Canon and Epson printers don't do that. Thus, HP's approach has an inherent advantage when trying to print accurate neutral and near neutral tones in a full color printing pipeline. It helps a smaller patch count target such as a TC918 color chart to deliver a pretty good B&W print without going to additional heroics, but a larger patch count target like the 3315 Aardenburg chart does refine HP's approach even more. It's subtle, but in my testing, I see it.  Epson, for example, takes a different approach than HP with regard to accurate B&W printing by resorting to an entirely different "advanced Black and white" (ABW) mode for monochrome printing. Nevertheless, HP's 100% GCR approach and its inherently neutral photo gray inks do deliver superior performance when printing B&W images in a full color output mode, and this is why the Z's have a well deserved reputation for making great B&W prints without having to resort to other methods like ABW, Quad Tone RIP, piezography ink sets, etc., to print excellent B&W output.   

cheers,
Mark
http://www.aardenburg-imaging.com

Thank you again for your input; I have a z3100 and that is exactly why i asked. It only uses the LGray, Gray and Black ink(s) to make a BW print.
Indeed the BW prints are perfectly neutral and have a great tonality.
I read that the Canson Platine delivers very good looking prints- may i ask what (preset) ink load to use?
I think Canson choice was the photopaper ink-load/ preset.


PS i just slightly altered the testchart textfile and the APS seemed to make a good profile. but it was not so; have to try again...
« Last Edit: September 08, 2020, 06:56:29 pm by kers »
Logged
Pieter Kers
www.beeld.nu/la

kers

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4388
    • Pieter Kers
Re: Experiments with Interstitially Nested Grid Profiling Charts and the Z3200.
« Reply #24 on: September 11, 2020, 08:59:21 am »

I just printed and measured a 5832 patch colorchart  and made an ICC-profile with the APS i bought from HP. I also made ICC-profiles with smaller targets, to compare.
All done on cheap paper, to reduce costs. The 5832 colorchart is my own made and is a simple 18x18 x18 evenly divided colorchart- steps of 15 RGB values.

i noticed the 5832 profile is not working perfectly. ; that same is true for the bill Atkinson 1728 and the official 918 colorpatch that i used.
The best one is the standard 440 chart from HP that comes with the printer in the HP designjet utility.
So now i need to check were the problem is.
could be the printer colorsensor.
could be the (2006) APS-software that made the ICC profile.
could be both...
(could be my own made target, but others show the same problem, so don't think so.)

Are there any programs that inspects ICC profiles that could help me out?
Do i have to take a deep terminal dive into argyllCMS? ( i have used the frontend DisplayCal , but it is only for displays)
if someone has an idea please let me know. for now i stick with the basic 440 profiles
« Last Edit: September 11, 2020, 09:09:31 am by kers »
Logged
Pieter Kers
www.beeld.nu/la

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Experiments with Interstitially Nested Grid Profiling Charts and the Z3200.
« Reply #25 on: September 11, 2020, 12:44:42 pm »

I just printed and measured a 5832 patch colorchart  and made an ICC-profile with the APS i bought from HP. I also made ICC-profiles with smaller targets, to compare.
All done on cheap paper, to reduce costs. The 5832 colorchart is my own made and is a simple 18x18 x18 evenly divided colorchart- steps of 15 RGB values.

i noticed the 5832 profile is not working perfectly. ; that same is true for the bill Atkinson 1728 and the official 918 colorpatch that i used.
The best one is the standard 440 chart from HP that comes with the printer in the HP designjet utility.
So now i need to check were the problem is.
could be the printer colorsensor.
could be the (2006) APS-software that made the ICC profile.
could be both...
(could be my own made target, but others show the same problem, so don't think so.)

Are there any programs that inspects ICC profiles that could help me out?
Do i have to take a deep terminal dive into argyllCMS? ( i have used the frontend DisplayCal , but it is only for displays)
if someone has an idea please let me know. for now i stick with the basic 440 profiles

One approach I've used is to verify the consistency of the patch measurements. You take all the surrounding, nearby patches of each patch that isn't at the gamut surface and average their lab values then compare the average to the one in the middle by calculating the dE76. This can be done in excel with some programming in VBA. I've written utilities in Matlab which is easier for this sort of thing.
Now you have a list of 27, RGB values and the dE76 between the center and mean of the outer ones. Sort the list and any printer or measurement anomalies become pretty obvious and you can them re-measure and/or reprint and measure the ones that have the largest dE76.

I've found this more useful since it more easily locates where the anomalous patches Lab values are located.
Logged

kers

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4388
    • Pieter Kers
Re: Experiments with Interstitially Nested Grid Profiling Charts and the Z3200.
« Reply #26 on: September 11, 2020, 03:53:19 pm »

OK thank you for sharing your thoughts; I will have a look. I am already busy with spreadsheets, bbedit and terminal commands, and still have a lot to learn...

PS how do i calculate the LAB values?
From each patch i have 36 spectral measurements of different wavelengts from nm 380 til nm 730.. Or am i missing the point...
« Last Edit: September 11, 2020, 07:26:45 pm by kers »
Logged
Pieter Kers
www.beeld.nu/la

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Experiments with Interstitially Nested Grid Profiling Charts and the Z3200.
« Reply #27 on: September 12, 2020, 01:28:02 am »

OK thank you for sharing your thoughts; I will have a look. I am already busy with spreadsheets, bbedit and terminal commands, and still have a lot to learn...

PS how do i calculate the LAB values?
From each patch i have 36 spectral measurements of different wavelengts from nm 380 til nm 730.. Or am i missing the point...

DeltaE 1976, the traditional one, is easy. It's the geometric mean of the difference between two Lab values:

dE76=sqrt((L1-L2)^2+(a1-a2)^2+(b1-b2)^2)
Logged

kers

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4388
    • Pieter Kers
Re: Experiments with Interstitially Nested Grid Profiling Charts and the Z3200.
« Reply #28 on: September 12, 2020, 07:12:32 am »

DeltaE 1976, the traditional one, is easy. It's the geometric mean of the difference between two Lab values:

dE76=sqrt((L1-L2)^2+(a1-a2)^2+(b1-b2)^2)

I understand that part, ( I think.)
my problem is more basic:

I have a reading from a patch like number 5095; R=195 G=0 B=225  ( strong purple colour)
the measured spectral values are like so:
nm380   nm390   nm400   nm410   nm420   nm430   nm440   nm450   nm460   nm470   nm480   nm490   nm500   nm510   nm520   nm530   nm540   nm550   nm560   nm570   nm580   nm590   nm600   nm610   nm620   nm630   nm640   nm650   nm660   nm670   nm680   nm690   nm700   nm710   nm720   nm730

0.0000   0.0000   -0.1067   0.0000   0.3134   0.2011   0.1922   0.1532   0.1201   0.0950   0.0615   0.0442   0.0207   0.0196   0.0071   0.0071   0.0046   0.0055   -0.0005   0.0179   -0.0001   0.1016   0.2149   0.3017   0.4022   0.4432   0.4745   0.5188   0.6111   0.6113   0.6635   0.6569   0.6730   1.3202   0.0000   0.0000


how do i calculate a Lab value (L, A and B) from these measurements?
« Last Edit: September 12, 2020, 07:34:09 am by kers »
Logged
Pieter Kers
www.beeld.nu/la

kers

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4388
    • Pieter Kers
Re: Experiments with Interstitially Nested Grid Profiling Charts and the Z3200.
« Reply #29 on: September 12, 2020, 08:10:26 am »

I just completed a 5832icc profile of a paper with a large gamut ( HPEveryday gloss) and on first observations the profile seems to work well.
The difference with the standard 440 seems small, but especially the strong colors at the borders of the gamut are better defined, more saturated.
At the moment it seems papers with small gamuts don't need, or are even better of with a small patch count profile, handling borders more smoothly. My first attempt on a coated paper showed that.
On BW i see an improvement in the local contrast, better defined fine structures...
Sonow I know now that i can make a large patchcount profile with the Z3100 and HP-APS half-automatically. HP-APS seems to have a limitation of about 3200 patches and/or about 1.20m print length.
The 5832 patch was about the best i could do in two different measurements -and then I have combined the results.

So Mark your 3315 patchlist almost worked, but when almost ready it gave an error - alas, but you put me on track!
Logged
Pieter Kers
www.beeld.nu/la

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Experiments with Interstitially Nested Grid Profiling Charts and the Z3200.
« Reply #30 on: September 12, 2020, 10:25:09 am »

I understand that part, ( I think.)
my problem is more basic:

I have a reading from a patch like number 5095; R=195 G=0 B=225  ( strong purple colour)
the measured spectral values are like so:
nm380   nm390   nm400   nm410   nm420   nm430   nm440   nm450   nm460   nm470   nm480   nm490   nm500   nm510   nm520   nm530   nm540   nm550   nm560   nm570   nm580   nm590   nm600   nm610   nm620   nm630   nm640   nm650   nm660   nm670   nm680   nm690   nm700   nm710   nm720   nm730

0.0000   0.0000   -0.1067   0.0000   0.3134   0.2011   0.1922   0.1532   0.1201   0.0950   0.0615   0.0442   0.0207   0.0196   0.0071   0.0071   0.0046   0.0055   -0.0005   0.0179   -0.0001   0.1016   0.2149   0.3017   0.4022   0.4432   0.4745   0.5188   0.6111   0.6113   0.6635   0.6569   0.6730   1.3202   0.0000   0.0000


how do i calculate a Lab value (L, A and B) from these measurements?

Easiest way is to load the CGATs file into i1profiler (not sure if this requires a license) or Argyll. With I1Profiler you can immediately save the file as another CGATs and it will calculate the Lab and include it in the CGATs. Pretty sure Argyll offers the same but I haven't tried it. I don't have a HPZ printer and I've always gotten Lab along with the spectra.

Best to get the exact process from someone with an HPZ that's done it.
Logged

kers

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4388
    • Pieter Kers
Re: Experiments with Interstitially Nested Grid Profiling Charts and the Z3200.
« Reply #31 on: September 13, 2020, 05:27:38 pm »

OK, thanks again.
Iprofiler needs a dongle...  and the installer comes without an uninstaller... so i will have a look at Argyll...







Logged
Pieter Kers
www.beeld.nu/la

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Experiments with Interstitially Nested Grid Profiling Charts and the Z3200.
« Reply #32 on: September 13, 2020, 05:44:26 pm »

OK, thanks again.
Iprofiler needs a dongle...  and the installer comes without an uninstaller... so i will have a look at Argyll...

Turns out you don't need a dongle.

I just checked i1Profiler. Since it can print and read patches w/o a dongle or license, it turns out it can also read a CGATs file with only RGB and Spectral data. Then when you save the data using custom cgats, just select Lab to add in Lab values and it automatically converts the spectral data to Lab.
Logged

deanwork

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2400
Re: Experiments with Interstitially Nested Grid Profiling Charts and the Z3200.
« Reply #33 on: September 13, 2020, 06:54:42 pm »





The i1 spectro is the dongle


Turns out you don't need a dongle.

I just checked i1Profiler. Since it can print and read patches w/o a dongle or license, it turns out it can also read a CGATs file with only RGB and Spectral data. Then when you save the data using custom cgats, just select Lab to add in Lab values and it automatically converts the spectral data to Lab.


Logged

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Experiments with Interstitially Nested Grid Profiling Charts and the Z3200.
« Reply #34 on: September 13, 2020, 09:42:02 pm »

The i1 spectro is the dongle

Yep, if it's licensed it acts as a dongle.

However, this process to get lab values when you only have spectral data works without any dongle or i1pro.
Logged

kers

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4388
    • Pieter Kers
Re: Experiments with Interstitially Nested Grid Profiling Charts and the Z3200.
« Reply #35 on: September 15, 2020, 03:50:01 pm »

Yep, if it's licensed it acts as a dongle.

However, this process to get lab values when you only have spectral data works without any dongle or i1pro.

OK, i also cannot get it to work. I downloaded iOne profiler 3.2 ;I can't see a way to use my own patch chart ( restricted to some formats  )
I can't see how to get lab values from CGatt files
and i cannot  do anything without a dongle.
Maybe we are talking about different software...

Logged
Pieter Kers
www.beeld.nu/la

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Experiments with Interstitially Nested Grid Profiling Charts and the Z3200.
« Reply #36 on: September 15, 2020, 04:47:47 pm »

Same program. But you need to select Profiling, not Printer

Then drag and drop your CGATs file onto the measurement tab which is 3rd from the left on the bottom. The save as a custom CGATs file which allows you to check the components to include like XYZ, Lab, Spectral data.

However, I just looked at a CGATs file from the HP z3200ps from the HP Printer Utility and it has Lab values in it. Here's the front end:

CGATS.5
FILE_DESCRIPTOR   "RGB 1728 Patches (12x12x12)"
CREATED   "November 16, 2017"
ORIGINATOR   "HP Printer Utility"
SERIAL   "016AD540-0C9E-4CA3-A44B-69329135A4EF"
INSTRUMENTATION   "Hewlett-Packard, HP Designjet Z3200ps 44in Photo (Q6721B), SN CN4BQ2K004"
MEASUREMENT_GEOMETRY   45/0
MEASUREMENT_SOURCE   "white LED"
FILTER   "uv"
POLARIZATION   "no"
SAMPLE_BACKING   "black"
PRINT_CONDITIONS   "Print Quality: Best_Graphics, Gloss Enhancer: Off"
MANUFACTURER   ""
MATERIAL   "Moab Entrada Natural 300gsm"
TARGET_TYPE   "RGB 1728 Patches (12x12x12)"
NUMBER_OF_SETS   1728
NUMBER_OF_FIELDS   44

BEGIN_DATA_FORMAT
   SAMPLE_ID   SAMPLE_NAME   RGB_R   RGB_G   RGB_B   LAB_L   LAB_A   LAB_B   SPECTRAL_380   SPECTRAL_390   SPECTRAL_400   SPECTRAL_410   SPECTRAL_420   SPECTRAL_430   SPECTRAL_440   SPECTRAL_450   SPECTRAL_460   SPECTRAL_470   SPECTRAL_480   SPECTRAL_490   SPECTRAL_500   SPECTRAL_510   SPECTRAL_520   SPECTRAL_530   SPECTRAL_540   SPECTRAL_550   SPECTRAL_560   SPECTRAL_570   SPECTRAL_580   SPECTRAL_590   SPECTRAL_600   SPECTRAL_610   SPECTRAL_620   SPECTRAL_630   SPECTRAL_640   SPECTRAL_650   SPECTRAL_660   SPECTRAL_670   SPECTRAL_680   SPECTRAL_690   SPECTRAL_700   SPECTRAL_710   SPECTRAL_720   SPECTRAL_730   SPECTRAL_740   SPECTRAL_750   SPECTRAL_760   SPECTRAL_770   SPECTRAL_780
END_DATA_FORMAT

BEGIN_DATA
   1   "1"   0.000000   0.000000   0.000000   19.982800   0.463143   0.768536   0.000000   0.000000   0.000000   0.014309   0.028617   0.028589   0.028623   0.028756   0.028875   0.028814   0.028768   0.028929   0.029132   0.029255   0.029357   0.029461   0.029554   0.029613   0.029690   0.029848   0.030038   0.030247   0.030433   0.030509   0.030596   0.030821   0.031086   0.031323   0.031565   0.031832   0.032092   0.032318   0.032520   0.016260   0.000000   0.000000   0.000000   0.000000   0.000000   0.000000   0.000000
   2   "2"   23.000000   0.000000   0.000000   23.632799   10.677700   8.439660   0.000000   0.000000   0.000000   0.016172   0.032344   0.029870   0.027810   0.026062   0.024935   0.024876   0.025376   0.026023   0.026979   0.028452   0.029891   0.030608   0.031219   0.031605   0.033278   0.038454   0.044909   0.051504   0.057200   0.060096   0.061647   0.062619   0.063180   0.063771   0.064341   0.065005   0.065711   0.066445   0.067213   0.033606   0.000000   0.000000   0.000000   0.000000   0.000000   0.000000   0.000000
   3   "3"   46.000000   0.000000   0.000000   26.231400   16.746401   12.005100   0.000000   0.000000   0.000000   0.017626   0.035253   0.032298   0.029712   0.027308   0.025648   0.025492   0.026066   0.026762   0.027840   0.029573   0.031310   0.032266   0.033120   0.033406   0.035771   0.043633   0.054221   0.066953   0.078676   0.085211   0.089058   0.091036   0.092002   0.093160   0.094287   0.095601   0.096985   0.098390   0.099833   0.049916   0.000000   0.000000   0.000000   0.000000   0.000000   0.000000   0.000000


So if you aren't seeing Lab values there may be an option you're missing in the HP Utility.
Logged
Pages: 1 [2]   Go Up