Z3200 - Better ICC Color Profiles

[Brad P]

I need to print files from my mac to a Z3200ps with no color management at all.  I believe the right procedure is to go to the HP Utility, hit Submit Job>Start, set up the file for print, but in Color>ICC Color Management>RGB Source Profile, select "None (Native)" for an RGB based file, then select absolute colorimetric as the rendering intent.  Is that correct or is there a better way? 

For rendering intent, the only other choices are default, relative colorimetric, perceptual or saturation which I imagine might cause problems. 

The reason I'm asking is that I had an external file profiler create a profile (they said it would be better but it didn't work) and I have started creating my own color profiles using Argyll with larger samples than the spectrophotometer natively processes with its software (which seem to work fine).

Adobe's Color Print Utility is not updated properly to work with my Mac computer so I can't use that option.

[Mark Lindquist]

You can make larger target profiles directly from the Z3200.
Check out this page:

Make 1728 patch targets for ICC Profiles

Then you can use Argylle with Terminal or ICC Gen

Always use Relative Colormetric with ICC Gen.

You can use Apple Color Center to print a non-color managed print if you need to.

But printing a 1728 patch target internally from the Z3200 is a huge improvement over the normal target used by the Z natively.

Mark

[deanwork]

That is very cool Mark,

Thanks for sharing, as always.

Do you know if I can drop this file into my X-Rite i1 Pro 2 software to create the icc profile?

John

You can make larger target profiles directly from the Z3200.
Check out this page:

Make 1728 patch targets for ICC Profiles

Then you can use Argylle with Terminal or ICC Gen

Always use Relative Colormetric with ICC Gen.

You can use Apple Color Center to print a non-color managed print if you need to.

But printing a 1728 patch target internally from the Z3200 is a huge improvement over the normal target used by the Z natively.

Mark

[kers]

...
But printing a 1728 patch target internally from the Z3200 is a huge improvement over the normal target used by the Z natively.
Mark

Mark, i just made some profiles for a paper with the 'HP advanced profiling solution' (APS) that i bought with the z3100
I have a 1728 reference so it can do that too.

i made a 220, a 918 and a 1728 profile with the APS and a 440 patch profile that came with the printers software.

Result ... and for me a trending result:
1728 and 918 has some banding in places with gradients in areas with strong colors were the 440 and 220 have not.
the 440 profile was the better one in this case. ( the 220 profile was not accurate enough )
On the other hand the 918 and 1728 had more extreme colour.

so my basic idea is :  more patches- more accurate colour, but also a greater chance for banding issues

What are your experiences?
I know this is a difficult subject ( can of worms) for a straightforward answer.

[deanwork]

I'm not Mark but I do know from experience with other profiling systems that the more patches the hardware has to read the greater the potential for mistakes in the reading process itself. This is most likely why Hp has standardized on using less patches to read as well as the fact that it takes less time to produce the profile.

Mark, i just made some profiles for a paper with the 'HP advanced profiling solution' (APS) that i bought with the z3100
I have a 1728 reference so it can do that too.

i made a 220, a 918 and a 1728 profile with the APS and a 440 patch profile that came with the printers software.

Result ... and for me a a trending result:
1728 and 918 has some banding in places with gradients in places with strong colors were the 440 and 220 have not.
the 440 profile was the better one in this case. ( the 220 profile was not accurate enough )
On the other hand the 918 and 1728 had more extreme colour.

so my basic idea is :  more patches- more accurate colour, but also a greater chance for banding issues

What are your experiences?
I know this is a difficult subject ( can of worms) for a straightforward answer.

[digitaldog]

You can use Apple Color Center to print a non-color managed print if you need to.

ColorSync Utility (or even Preview).
Not all patch sets are created equally so when someone states a 1728 patch target, they differ. You really want Bill Atkinson's patch set, based on my testing.

[digitaldog]

so my basic idea is :  more patches- more accurate colour, but also a greater chance for banding issues

In a near prefect world, you'd make a 16.7 million patch target that represents every possible RGB value. But that would take forever to measure and the profile would be larger than many images.
In a prefect world, you'd make a target that only contains device values that define colors so you'd have far less than 16.7 million patches. Still probably a few million and the same issues as above.


So we need to use a vastly smaller set of patches and the software has to extrapolate a lot. Some do this better than others. Some products create better sets of patches than others. With a good set of patches from someone who knows how to define them, 1728 can do a good job. However, the product I use to create profiles offers an POST optimization process where another set of patches, generated from the data in the first profile, can in some cased improve the original profile. I'm working on a video now to show what that provides when used (it's subtle but it's visible).


Not all profiles are created equally and just one reason is, not all targets and the patches used are created equally:
Not all ICC profiles are created equally

In this 23 minute video, I'll cover:
The basic anatomy of ICC Profiles
Why there are differences in profile quality and color rendering
How to evaluate an ICC output profile
Examples of good and not so good canned profiles and custom profiles on actual printed output.

High resolution: http://digitaldog.net/files/Not_All_Profiles_are_created_equally.mp4
Low resolution (YouTube): https://www.youtube.com/watch?v=TNdR_tIFMME&feature=youtu.be

[Brad P]

Not all profiles are created equally and just one reason is, not all targets and the patches used are created equally:
Not all ICC profiles are created equally

In this 23 minute video, I'll cover:
The basic anatomy of ICC Profiles
Why there are differences in profile quality and color rendering
How to evaluate an ICC output profile
Examples of good and not so good canned profiles and custom profiles on actual printed output.

High resolution: http://digitaldog.net/files/Not_All_Profiles_are_created_equally.mp4
Low resolution (YouTube): https://www.youtube.com/watch?v=TNdR_tIFMME&feature=youtu.be

That is an awesome video.  Thanks.  I’m actually shocked to see the differences in the canned and custom profiles at the end.  It proves a good part of the time and money we allocate to lenses and camera gear would better allocated to color profiling if we want good prints. 

[digitaldog]

That is an awesome video.  Thanks.  I’m actually shocked to see the differences in the canned and custom profiles at the end.

Actually, it's the differences in the two canned profiles quality, from the SAME company that's rather shocking to me.

[Brad P]

Hi Mark - thanks for this.  I've been doing about the same thing for a few months but manually without ICC Gen.

I noticed you wrote to use RelCol with ICC Gen.  Is that an ICC Gen restraint or your preferred profile?  From memory, in Geraldo's original post somewhere someone noted that one could produce both RelCol or Perceptual ICC profiles using Argyll, but it was either one or the other with each icc file.  I've only made a perceptual profile to date but was about to reprint a 1728 sample today and make both perceptual and relative profiles (setting eachup different paper names in the printer, like "IGFGRelCol1728RGEon" and "IGFGPerc1728GEon").  Interested in your experiences so far.

You can make larger target profiles directly from the Z3200.
Check out this page:

Make 1728 patch targets for ICC Profiles

Then you can use Argylle with Terminal or ICC Gen

Always use Relative Colormetric with ICC Gen.

Mark

[kers]

ColorSync Utility (or even Preview).
Not all patch sets are created equally so when someone states a 1728 patch target, they differ. You really want Bill Atkinson's patch set, based on my testing.

Hello Andrew, thanks for joining,
I got the 1728 from Bill Atkinson-  the software however maybe does not match with the target...
I will watch your video, have a closer look at what is going on and come back...

[digitaldog]

Hello Andrew, thanks for joining,
I got the 1728 from Bill Atkinson-  the software however maybe does not match with the target...
I will watch your video, have a closer look at what is going on and come back...

IF you have a standard CGATs text file, virtually any software worth it's salt should import that to build a target.
Remind me again what product you're using to build a profile?

[Doug Gray]

Actually, it's the differences in the two canned profiles quality, from the SAME company that's rather shocking to me.

I'm less shocked. Unfortunately, only the Colorimetric Intents are well spec'ed by the ICC. And only those colors that are in gamut.  How out of gamut colors are mapped, as well as virtually all of Perceptual is left to the implementer. But for the same vendor to have such large differences in OOG rendering is rather odd.  Any idea how well in gamut colors are rendered by the two profiles? It may be the OOG ugly one does a better job rendering in gamut colors.

[Mark Lindquist]

That is very cool Mark,

Thanks for sharing, as always.

Do you know if I can drop this file into my X-Rite i1 Pro 2 software to create the icc profile?

John

Hi John,
Yes you can.  Should be no problem.

EDIT:

It looks like it is doable John, but we have to determine how to make a minor modification to the Z series cgats header to enable x-rite to accept the files.  Working on it.  Will keep you posted.

Mark

[Mark Lindquist]

Mark, i just made some profiles for a paper with the 'HP advanced profiling solution' (APS) that i bought with the z3100
I have a 1728 reference so it can do that too.

i made a 220, a 918 and a 1728 profile with the APS and a 440 patch profile that came with the printers software.

Result ... and for me a trending result:
1728 and 918 has some banding in places with gradients in areas with strong colors were the 440 and 220 have not.
the 440 profile was the better one in this case. ( the 220 profile was not accurate enough )
On the other hand the 918 and 1728 had more extreme colour.

so my basic idea is :  more patches- more accurate colour, but also a greater chance for banding issues

What are your experiences?
I know this is a difficult subject ( can of worms) for a straightforward answer.

I’ve never had banding with 1728 patch target profiles using the Z.  I know that everything must be followed to the letter, (including advance paper calibration) and one must use the correct preset for the paper being profiled, and one must overwrite the original created paper not changing any names during install.  Also, it is an experiment to find the right combinations which can be time consuming and costly.  In using Argyll, rel col  works the best IMO.  It takes a lot of trial and error to work it out, but once you get it it is pretty good.  MHMG and I did some testing and he came up with excellent results. 
Mark

[Mark Lindquist]

Hi Mark - thanks for this.  I've been doing about the same thing for a few months but manually without ICC Gen.

I noticed you wrote to use RelCol with ICC Gen.  Is that an ICC Gen restraint or your preferred profile?  From memory, in Geraldo's original post somewhere someone noted that one could produce both RelCol or Perceptual ICC profiles using Argyll, but it was either one or the other with each icc file.  I've only made a perceptual profile to date but was about to reprint a 1728 sample today and make both perceptual and relative profiles (setting eachup different paper names in the printer, like "IGFGRelCol1728RGEon" and "IGFGPerc1728GEon").  Interested in your experiences so far.

I find very little difference in the two, and yes, do prefer relcol.  Argyl is what it is, other programs like RGB Drop do a very similar thing, in tests conducted with MHMG. In looking at comparitive graphs - very similar results.

Mark

[Mark Lindquist]

ColorSync Utility (or even Preview).
Not all patch sets are created equally so when someone states a 1728 patch target, they differ. You really want Bill Atkinson's patch set, based on my testing.

Thank you for correcting that Andrew - I did mean to say ColorSync Utility.

The target that the Z produces is the target the Z can and will read.  If it is possible to convert the Atkinson target to one readable by the Z, then that would be cool.  I think it can be done but I have yet to play with that.  I know that the Z is really picky about what it will and won’t read, however.  The 1728 patch target is much better than the standard one the Z produces automatically.  Not too bad considering it is all automated and in house as part of the actual cost of the printer.  Move away from that and it becomes money on an entirely new level for stand alone machines.

-Mark

[digitaldog]

The target that the Z produces is the target the Z can and will read.  If it is possible to convert the Atkinson target to one readable by the Z, then that would be cool. 

That's kind of true everywhere. So if you get a CGATs file defining the reference colors for the 1728 patches of Bills selection, the software needs the ability to create a custom target from that data. Very easy to do in a number of products like i1Profiler etc. No idea what your Z software can do. If it's worth it's salt and not some crippled product, it should allow you to create a custom target from the CGATs file. That custom target has all the colors values (device values) desired and formatted for whatever Spectrophotometer you need to use.

[Mark Lindquist]

That's kind of true everywhere. So if you get a CGATs file defining the reference colors for the 1728 patches of Bills selection, the software needs the ability to create a custom target from that data. Very easy to do in a number of products like i1Profiler etc. No idea what your Z software can do. If it's worth it's salt and not some crippled product, it should allow you to create a custom target from the CGATs file. That custom target has all the colors values (device values) desired and formatted for whatever Spectrophotometer you need to use.

Andrew, can you send me the Atkinson file?  I’ll check it out.
Mark

[digitaldog]

Andrew, can you send me the Atkinson file?  I’ll check it out.
Mark

Shoot me a PM with an email address. I have a CGATs file, formatted for ProfileMaker Pro but can format differently.