Pages: [1]   Go Down

Author Topic: Z3200 Profiling  (Read 2854 times)

johncustodio

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 72
Z3200 Profiling
« on: February 18, 2018, 01:08:09 pm »

My Z3100 that I had for 10 years finally died last year, so I recently bought a 24" Z3200. I've been following the thread on profiling large targets and have successfully profiled a 2371 patch target printed on Epson Proofing paper.
– I made a 2371 chart in i1 Profiler, saved the target as a pxf file, reloaded the file to get rid of the fractional bit values, then resaved it as a CGATS file (.txt).
– Following the instructions in the thread, I loaded the CGATS file into the HP Color Center, and made a 24"x36" tiff file for printing. I printed the chart using the Adobe Printing Utility.
– After letting the print dry overnight, I loaded the sheet back into the Z3200, and told the Color Center to measure an already printed file. I set the measurement file to CGATS, and to include only original RGB and Sample_ID values, and Spectral Measurements. (First 2 check boxes checked - see Select Measurement Type screen shot attachment).
– After the printer measured the chart, I modified the resulting CGATS file by changing the .csv extension to .txt, and with Text Edit changed the number of fields from 46 to 41 to discount the spectral data from 740 to 780nm which i1 Profiler doesn’t read.
– I loaded the modified CGATS file into the Measure Tab of i1 Profiler.

I have a few questions at this point:
– i1 Profiler asks for the calibration type. I'm assuming the spectro in the Z3200 is calibrated to XRGA, so I selected that in the drop down menu as well as no conversion. (See Import and Convert Measurements screen shot attachment.)
– i1 Profiler then asks to select the measurement condition: M0, M1, or M2. I selected M0 UV included. What measurement condition does the spectro in the Z3200 use? It may use M2 UV excluded because in the CGATS file header it says FILTER "uv". (See Select Measurement Condition screenshot attachment and GCATS Header screenshot attachment.) In my case, it may not matter whether I select M0 or M2 because the paper I'm using has no OBAs.
Logged

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Z3200 Profiling
« Reply #1 on: February 18, 2018, 05:42:48 pm »

John,

Mark it as uV cut (M2).

As for the conversion options probably the safest thing is XRGA and no conversion. The different options represent variation amongst internal standards for differing models and companies that later merged. It's entirely possible HP developed their own. They certainly have the tech to have done it. But the question remains which of the versions are closest to the HP?  One way to do this is to print the same target in a format read by an iSiS. Then compare those Lab values to the Lab values after conversion of the HP to XRGA using each of the selectable options. Then, in the future, use that option that produced the closest match when importing any CGTAGS into the I1Profiler.

BTW, congratulations on having cleared a lot of weeds to get this far.


BTW, if you have access to an iSiS I can walk you through steps to find out if XRGA is correct, or if not which one of the conversions to XRGA would be the closest. If you don't have access I'd be willing to provide a small target for you to print w/o color management and PM you my address. While I don't have the HP so can't evaluate it myself, I can scan an iSiS target you print and evaluate which is closest if not XRGA. That would likely be a useful service to many here with the HP printer.
« Last Edit: February 18, 2018, 08:02:56 pm by Doug Gray »
Logged

Ernst Dinkla

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4005
Re: Z3200 Profiling
« Reply #2 on: February 19, 2018, 06:24:11 am »

John,

Mark it as uV cut (M2).

As for the conversion options probably the safest thing is XRGA and no conversion. The different options represent variation amongst internal standards for differing models and companies that later merged. It's entirely possible HP developed their own. They certainly have the tech to have done it. But the question remains which of the versions are closest to the HP?  One way to do this is to print the same target in a format read by an iSiS. Then compare those Lab values to the Lab values after conversion of the HP to XRGA using each of the selectable options. Then, in the future, use that option that produced the closest match when importing any CGTAGS into the I1Profiler.

BTW, congratulations on having cleared a lot of weeds to get this far.


BTW, if you have access to an iSiS I can walk you through steps to find out if XRGA is correct, or if not which one of the conversions to XRGA would be the closest. If you don't have access I'd be willing to provide a small target for you to print w/o color management and PM you my address. While I don't have the HP so can't evaluate it myself, I can scan an iSiS target you print and evaluate which is closest if not XRGA. That would likely be a useful service to many here with the HP printer.

X-rite  Gretag Macbeth merge was announced early 2006. The HP Z3100 was introduced mid/end 2006. X-rite first publication of XRGA dates back to 2010.  There are Z3100/Z3200 spectrometer similarities to both the old Isis (GMB when not updated) and the Color Munki (XRGA), the Color Munki Photo is from early 2008, old Isis dates before that. The Z3200 intro was end of 2008. I checked some old PDFs on the Z spectrometers but could not find a reference to the calibration standards XRGA or GMB for it. Will transfer them to Mark as they were not among the docs I already handed over and there is other interesting info in them. Proofing software tests Fogra EFI GMG do not give answers either.

My bet is GMB for the Z3100 and given little bug support from X-rite after the appearance of the Z3200 I expect it still is GMB whatever firmware upgrades became available since.

The Z spectrometer measurements go from 400 to 700 nm, 20nm intervals.  As shown before in ArgyllCMS developments, smaller step measuring than OEM software allows  is possible, the i1Pro can do that for example.  I have good reasons to think it is possible for the Zs too. 380 to 400 nm can be faked/extrapolated if profiling software needs that. However as mentioned already the lamp has no UV so no UV effect can be measured. So UV-cut    M2.


Met vriendelijke groet, Ernst

http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
March 2017 update, 750+ inkjet media white spectral plots



Logged

johncustodio

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 72
Re: Z3200 Profiling
« Reply #3 on: February 19, 2018, 10:23:22 am »

So the question is what standard is the Z's spectro using. The way I understand it is that we want to convert that standard to XRGA (see screenshot 1), and choose what standard the Z is using to convert from, GMDI, XRDI, or XRGA (no conversion) (see screenshot 2). Ernst says he suspects the Z is using GMDI. I don't have access to an iSIS, so Doug if you can send me the test target, I'll print it and send it back to you. My email is: mail@johncustodio.com

AS Ernst says about the Z's spectro only measuring in 20nm bands, take a look at screenshot 3. When making the measurement file in the HP Utility you have a choice of 5, 10, or 20nm bands, but 5 and 10 are interpolated.

Logged

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Z3200 Profiling
« Reply #4 on: February 19, 2018, 10:53:05 am »

So the question is what standard is the Z's spectro using. The way I understand it is that we want to convert that standard to XRGA (see screenshot 1), and choose what standard the Z is using to convert from, GMDI, XRDI, or XRGA (no conversion) (see screenshot 2). Ernst says he suspects the Z is using GMDI. I don't have access to an iSIS, so Doug if you can send me the test target, I'll print it and send it back to you. My email is: mail@johncustodio.com

AS Ernst says about the Z's spectro only measuring in 20nm bands, take a look at screenshot 3. When making the measurement file in the HP Utility you have a choice of 5, 10, or 20nm bands, but 5 and 10 are interpolated.

The I1Pro measures over about a 15nm window in increments of about 3nm and X-Rite interpolates it to 10nm intervals. Some degree of interpolation is intrinsic. HP may well also use a 15nm window if it is based on the I1 tech. HP tends to be rather precise in their wording. They may well make measurements only every 20nm. OTOH, the I1Pro makes measurements every 3nm and smooth's them to 10nm which changes little as the window is already 15nm. Interpolation inaccuracy is largely a factor only in illuminant measurement where spikes occur but I would expect some errors in the shorter wavelengths in all of these that use LEDs since they are very spikey around 470nm and also have a strong dip around 510nm.

John, What I need first is the CGATs file you made. This will have RGB values and from those I can select a subset to create  a target. It's important to do it that way because we are looking at fairly small differences and so the target needs to have an exact, but evenly distributed, subset of the RGB values to compare against the various conversion options I1Profiler has.

So here's the process:

1. You post the CGATS file here as an attachment.
2. I create a subset CGATs file and tiff target and post those as an attachment.
3. You print the target tiff file w/o color management and mail it to me (mailing address by PM)
4. I scan the target, try the I1Profiler options, and determine which option is the closest match to current XRGA standards.


Logged

johncustodio

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 72
Re: Z3200 Profiling
« Reply #5 on: February 19, 2018, 03:16:57 pm »

Doug, Here's the CGATS file attached.
Logged

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Z3200 Profiling
« Reply #6 on: February 19, 2018, 04:12:22 pm »

Doug, Here's the CGATS file attached.

John,
Here you go.

There are three files. 2 RGB CGATs files and 1 Tif file. The iSiS tif file is designed to be symmetric with a reference bar at the top and bottom. This allows me to read it from either end and is a quick check for consistency. The two CGATs files are the RGB reference files used for the normal top-down direction and the reverse direction. This is provided in case someone here with an iSiS wishes to read the files.

Print the tif image using U.S. Letter size paper or trim to that size. If you trim the paper leave at least 30mm on the top and bottom. Print using color management disabled. Consider making two identical prints. This would provide some information on print to print consistency.


For others Interested
Instructions for reading with an iSiS:
1. Load the CGATs rgb file. Then select US Letter size paper.
2. Use the iSiS defaults but change the patch width to 10.5 mm and the patch height to 7.0 mm. This will reformat the I1Profiler target image to match the provided tiff target.


Edit: Replaced 16 bit target with 8 bit target. RGB values are identical. I default to 16 bit because many of my targets use fractional rgb values. However, all the major profiling target print apps expect 8 bit targets.
« Last Edit: February 19, 2018, 06:16:34 pm by Doug Gray »
Logged

Ernst Dinkla

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4005
Re: Z3200 Profiling
« Reply #7 on: February 19, 2018, 04:14:51 pm »

The I1Pro measures over about a 15nm window in increments of about 3nm and X-Rite interpolates it to 10nm intervals. Some degree of interpolation is intrinsic. HP may well also use a 15nm window if it is based on the I1 tech. HP tends to be rather precise in their wording. They may well make measurements only every 20nm. OTOH, the I1Pro makes measurements every 3nm and smooth's them to 10nm which changes little as the window is already 15nm. Interpolation inaccuracy is largely a factor only in illuminant measurement where spikes occur but I would expect some errors in the shorter wavelengths in all of these that use LEDs since they are very spikey around 470nm and also have a strong dip around 510nm.


Doug,

The TechNotable: HP Embedded Spectrophotometer with i1 Color Technology. pdf says spectral bandwidth 20nm but there is a note at the end of the PDF;

7    All sensor performance figures under nominal measurement conditions. Sensor accuracy figures represent Mean Color Difference from
Standard and Maximum Color Difference from Standard respectively, on BCRA reference tiles. Sensor to sensor repeatability figures
represent the difference from one sensor to the population mean on BCRA tiles. A UV cut filter is integrated into the HP Embedded
Spectrophotometer. Both the HP Embedded Spectrophotometer and the X-RiteTM i1 ® sample the spectrum internally with an optical interval
of 3.5 nm.


That pdf is no longer on the web but I have archived it. In another HP pdf text an output format is mentioned for the Z3100 spectrometer; CGATS with 380-730nm @10nm spectral measurements.  I could have overlooked settings like that but I do not think it was a normal setting in either Color Center or the APS in Windows version.


Met vriendelijke groet, Ernst

http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
March 2017 update, 750+ inkjet media white spectral plots

Logged

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Z3200 Profiling
« Reply #8 on: February 19, 2018, 04:34:59 pm »

Doug,

The TechNotable: HP Embedded Spectrophotometer with i1 Color Technology. pdf says spectral bandwidth 20nm but there is a note at the end of the PDF;

7    All sensor performance figures under nominal measurement conditions. Sensor accuracy figures represent Mean Color Difference from
Standard and Maximum Color Difference from Standard respectively, on BCRA reference tiles. Sensor to sensor repeatability figures
represent the difference from one sensor to the population mean on BCRA tiles. A UV cut filter is integrated into the HP Embedded
Spectrophotometer. Both the HP Embedded Spectrophotometer and the X-RiteTM i1 ® sample the spectrum internally with an optical interval
of 3.5 nm.


That pdf is no longer on the web but I have archived it. In another HP pdf text an output format is mentioned for the Z3100 spectrometer; CGATS with 380-730nm @10nm spectral measurements.  I could have overlooked settings like that but I do not think it was a normal setting in either Color Center or the APS in Windows version.


Met vriendelijke groet, Ernst

http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
March 2017 update, 750+ inkjet media white spectral plots

Ernst,
Thanks for that tidbit. Graeme at Argyll, had noted that the bandwidth, even though sampled at 3nm intervals, appeared to be 15nm. So, at 10nm and 5nm it would be reasonable for HP to note that these are interpolated.

Should be interesting to see how the HP values compare to the various GMB/X-Rite versions. Wouldn't surprise me if they match most closely with XRGA since they have the same interval and LED illuminant.
Logged

johncustodio

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 72
Re: Z3200 Profiling
« Reply #9 on: February 19, 2018, 10:24:56 pm »

Doug, I printed 2 copies of the target. PM me where I should send them. mail@johncustodio.com
Logged

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Z3200 Profiling
« Reply #10 on: February 19, 2018, 10:31:05 pm »

Doug, I printed 2 copies of the target. PM me where I should send them. mail@johncustodio.com
You should have gotten the PM. Check the messages in the top right of the forum.

I just checked "sent" messages and only one shows up from last summer. Weird. Might be a forum problem. Let me know if you don't have a PM showing here and I'll email you.
« Last Edit: February 19, 2018, 10:36:37 pm by Doug Gray »
Logged

johncustodio

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 72
Re: Z3200 Profiling
« Reply #11 on: February 20, 2018, 08:08:47 am »

Doug, OK, I got it. I'll send it out today.
Logged

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Z3200 Profiling
« Reply #12 on: February 23, 2018, 05:20:18 pm »

Got the charts. Format's good, they read fine. I'm reading them in (forwards and backwards) to check for consistency then will be comparing them to your CGATs file. Stand by.
Logged

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Z3200 Profiling
« Reply #13 on: February 24, 2018, 02:34:55 am »

Select the conversion from GretagMacBeth (import) to XRGA (export). This provides the closest match between the older, full profile and the 343 patch subset. The subset takes a 7x7x7 RGB from the full set. The next closest is the unchanged XRGA to XRGA and the worst is the Xrite legacy.

Average dE00 for the first chart is .6 and .8 for the second. These are quite close for two different instruments. They aren't unreasonable even for two of the same models.

Comparing only the two 343 patch charts produces an average dE00 of .4 which is a bit higher than I see printing two charts in succession unless my printers had not been operated in several days. Might be a good idea when printing profiles to print a smaller, throwaway to warm up the printer.

I'll provide statistical info tomorrow when I have a chance to get back to it and post the CGATS for anyone interested in looking over the raw data.
Logged

Ernst Dinkla

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4005
Re: Z3200 Profiling
« Reply #14 on: February 24, 2018, 04:06:26 am »


My Z3100 that I had for 10 years finally died last year, so I recently bought a 24" Z3200. I've been following the thread on profiling large targets and have successfully profiled a 2371 patch target printed on Epson Proofing paper.


With Doug's result and I presume your Z3200 is a B version as it is new, the Z3100/3200 and most likely the other Z's too, all have GMB-GDMI based spectrometers. Good to know, thank you Doug.

Met vriendelijke groet, Ernst

http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
March 2017 update, 750+ inkjet media white spectral plots

Logged

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Z3200 Profiling
« Reply #15 on: February 24, 2018, 03:11:25 pm »

Statistics for treating HPZ data as XRDI, XRGA, and GMDI then converting as required to current XRGA X-Rite standard.
This data compares the first printed iSiS compatible target sheet read with an iSiS 2 XL with its corresponding data from the HPZ spectrophotometer.

XRDI - Delta E2000:  Ave: 0.94, Max: 2.17, Median: 0.91, Worst 10%: 1.38
XRGA - Delta E2000:  Ave: 0.79, Max: 1.82, Median: 0.75, Worst 10%: 1.12
GMDI - Delta E2000:  Ave: 0.62, Max: 1.67, Median: 0.58, Worst 10%: 0.95


This is a comparison of the two, identically printed sheets read with the same iSiS 2 XL.
Sheet 1v2 - Delta E2000:  Ave: 0.46, Max: 1.08, Median: 0.44, Worst 10%: 0.74

It's pretty clear the HPZ spectrophotometer is based on the older GretagMacBeth tech, as Ernst had predicted, and that should be the selected conversion mode.
« Last Edit: February 26, 2018, 11:36:51 pm by Doug Gray »
Logged

johncustodio

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 72
Re: Z3200 Profiling
« Reply #16 on: February 25, 2018, 09:45:34 am »

Thanks, Doug, for doing this!
Logged

johncustodio

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 72
Re: Z3200 Profiling
« Reply #17 on: February 25, 2018, 11:13:53 am »

Doug,
Just to double check and make sure I'm doing this right, we're telling I1 Profiler that the Z3200's measurement file is GMDI, and we want to import it as (convert to) XRGA as shown in the attached screenshot.
Logged

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Z3200 Profiling
« Reply #18 on: February 25, 2018, 11:35:57 am »

Doug,
Just to double check and make sure I'm doing this right, we're telling I1 Profiler that the Z3200's measurement file is GMDI, and we want to import it as (convert to) XRGA as shown in the attached screenshot.

Yes John, that's exactly what to do.
Logged

Ernst Dinkla

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4005
Re: Z3200 Profiling
« Reply #19 on: February 25, 2018, 12:01:16 pm »

https://www.argyllcms.com/doc/XRGA.html

That page needs an addition too then.

Edit: emailed Graeme Gill.

Ernst, op de lei getypt
« Last Edit: February 26, 2018, 08:10:05 am by Ernst Dinkla »
Logged
Pages: [1]   Go Up