Pages: 1 2 [3] 4 5   Go Down

Author Topic: Apple and Colour Management – Just When You Thought It Was Safe To Go Out  (Read 43220 times)

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20614
  • Andrew Rodney
    • http://www.digitaldog.net/

Andrew: Your data suggest that the problem is i1Profiler on 10.6.8.

Well I’ve attempted to make them aware of this.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Eric Myrvaagnes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 22814
  • http://myrvaagnes.com
    • http://myrvaagnes.com

Pretty hard to profile consumer printers that don't respect color management nor offer a way of turning CM off. The one I've got you can't really profile is an Epson PictureMate printer. I just feed it sRGB and use Printer Manages Color.
Hey, Jeff,

Maybe it's time for you to upgrade to a serious printer, like, say the Epson 2200!   ;D
Logged
-Eric Myrvaagnes (visit my website: http://myrvaagnes.com)

Schewe

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6229
    • http:www.schewephoto.com

Maybe it's time for you to upgrade to a serious printer, like, say the Epson 2200!   ;D

Well, when I make quick prints for family and friends and want 4.5x6 borderless glossy, the PictureMate is just way easier...
Logged

tatuvaaj

  • Newbie
  • *
  • Offline Offline
  • Posts: 8
    • 500px

Hi all,

I'm completely new at this but I was just wondering why nobody hasn't mentioned the ColorSync Utility? If you open a file and then go to Print, it has an option to "Print as color target" (under Color).
Logged

Doyle Yoder

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 519

Hi all,

I'm completely new at this but I was just wondering why nobody hasn't mentioned the ColorSync Utility? If you open a file and then go to Print, it has an option to "Print as color target" (under Color).

Is this something new in 10.7? I don't see it in 10.6.8.
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20614
  • Andrew Rodney
    • http://www.digitaldog.net/

I'm completely new at this but I was just wondering why nobody hasn't mentioned the ColorSync Utility? If you open a file and then go to Print, it has an option to "Print as color target" (under Color).

I’ll have to test this. Yes, the option IS there and yes, it does invoke the correct settings for Color Matching (Apple as you would expect or maybe not, is implementing kPMApplicationColorMatching).
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

tatuvaaj

  • Newbie
  • *
  • Offline Offline
  • Posts: 8
    • 500px

Is this something new in 10.7? I don't see it in 10.6.8.

I think it is new in Lion (also available in first preview of 10.8 )
Logged

hjulenissen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2051

Win 7 initially was slated to support "High Color", a space with sRGB primaries but allowing values outside the [0 .. 1] range. Monitors that supported High Color would map untagged images into their native space, avoiding the wildly oversaturated look that wide-gamut panels can give. That initiative went the way of WCS.
It does seem like many "revolutionary" color features are pumped up in the PR departement, only to be silently forgotten in when it comes to shipping actual software. I wonder why that is.

-h
Logged

Simon J.A. Simpson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 584

Simon - as I understand it, the dialog you've posted looks that way when printing from Color Munki right? Have you tried printing from Photoshop using the Null Transform trick. To make a long story short that's how I think it should be done... There are otherwise too many variables and ways it can get screwed up.

Scott.  No I haven't tried this.  I will, over the next week and report back (might need a new pack of paper though !).  The great thing about the ColorMunki software (assuming it is giving the correct target) is that its a cinch to use (or may be I'm just a bit lazy … ?!).

Hi all,
I'm completely new at this but I was just wondering why nobody hasn't mentioned the ColorSync Utility? If you open a file and then go to Print, it has an option to "Print as color target" (under Color).

Hmm, interesting.  I'll try this too (takes time out to order another pack of paper).  May be it'll give me yet another choice of how to print targets !  (grins).  BTW, my A3+ printer is a "professional" printer – at least that's what Canon says.  Would that my wife would let have something bigger (a printer, that is).
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20614
  • Andrew Rodney
    • http://www.digitaldog.net/

Something is very screwy using the ColorSync utility so I’m going to recommend not using it for now. Here are the differences between it and Adobe Print Utility and Photoshop CS5 using the null transform trick. Note that it doesn’t appear to matter if you select RelCol or Perceptual or Black Point Compensation in Photoshop:


Photoshop CS5: RelCol without BPC versus Perceptual with BPC:
--------------------------------------------------

dE Report

Number of Samples: 918

Delta-E Formula dE2000

Overall - (918 colors)
--------------------------------------------------
Average dE:   0.18
    Max dE:   0.71
    Min dE:   0.01
 StdDev dE:   0.10

Best 90% - (825 colors)
--------------------------------------------------
Average dE:   0.15
    Max dE:   0.31
    Min dE:   0.01
 StdDev dE:   0.07

Worst 10% - (93 colors)
--------------------------------------------------
Average dE:   0.39
    Max dE:   0.71
    Min dE:   0.31
 StdDev dE:   0.08

--------------------------------------------------
--------------------------------------------------
Rendering intent and BPC do not appear to affect NULL transform.

ACPU vs. Photoshop (Perceptual with BPC):

--------------------------------------------------

dE Report

Number of Samples: 918

Delta-E Formula dE2000

Overall - (918 colors)
--------------------------------------------------
Average dE:   0.26
    Max dE:   0.97
    Min dE:   0.03
 StdDev dE:   0.15

Best 90% - (825 colors)
--------------------------------------------------
Average dE:   0.22
    Max dE:   0.45
    Min dE:   0.03
 StdDev dE:   0.10

Worst 10% - (93 colors)
--------------------------------------------------
Average dE:   0.57
    Max dE:   0.97
    Min dE:   0.45
 StdDev dE:   0.11

--------------------------------------------------
--------------------------------------------------

ACPU vs. Photoshop (RelCol without BPC):
--------------------------------------------------

dE Report

Number of Samples: 918

Delta-E Formula dE2000

Overall - (918 colors)
--------------------------------------------------
Average dE:   0.26
    Max dE:   0.99
    Min dE:   0.02
 StdDev dE:   0.15

Best 90% - (825 colors)
--------------------------------------------------
Average dE:   0.22
    Max dE:   0.46
    Min dE:   0.02
 StdDev dE:   0.10

Worst 10% - (93 colors)
--------------------------------------------------
Average dE:   0.58
    Max dE:   0.99
    Min dE:   0.46
 StdDev dE:   0.11

--------------------------------------------------
--------------------------------------------------
Now the shocker (since the above three are close, I'm only selecting one to compare with ColorSync utility):

APCU versus ColorSync Utility
--------------------------------------------------

dE Report

Number of Samples: 918

Delta-E Formula dE2000

Overall - (918 colors)
--------------------------------------------------
Average dE:   0.47
    Max dE:   4.08
    Min dE:   0.03
 StdDev dE:   0.44

Best 90% - (825 colors)
--------------------------------------------------
Average dE:   0.35
    Max dE:   1.02
    Min dE:   0.03
 StdDev dE:   0.21

Worst 10% - (93 colors)
--------------------------------------------------
Average dE:   1.52
    Max dE:   4.08
    Min dE:   1.02
 StdDev dE:   0.52

--------------------------------------------------
--------------------------------------------------
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

tatuvaaj

  • Newbie
  • *
  • Offline Offline
  • Posts: 8
    • 500px

Andrew,

Thank you very much!

Care to speculate on why such a large difference in the worst 10%?
Logged

MHMG

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1285

While delta E= 4 isn't a huge error unless it's in a low chroma color, it is still puzzling to find, but just because Adobe Color utility is consistent with the null transform trick (which also relies on Adobe components) on Mac 10.7 , doesn't necessarily mean Adobe has found the "pure" unaltered data path and Apple hasn't with Colorsync Utility. If one actually completes making a profile with these two approaches, then makes prints using printer manages color (as Apple presumably thinks most of it's customers do if they choose colorsync settings), and lastly performs a delta E test comparing predicted to actual output, the outcome could possibly tip in favor of the Colorsync utility accuracy.  One reason I suggest this is that I never was able to get totally consistent target generation across Mac OS 1.5 on Power PC to 10.5 on Intel and 10.6 Snow Leopard (Intel only). Differences even larger than what Andrew has reported kept cropping up. That outcome leads me sadly to believe that nowadays custom ICC profiles may not only be printer/ink/media/driver-settings specific, they may be OS specific as well!  Also, Adobe Color Utility totally failed to "hijack" the printer pipeline when I tried it in conjunction with my Canon ipF8300's "Free layout" printer driver, even though the appropriate settings were grayed out.  So, like Onsight, I just use the Canon 16 bit plugin for target generation. That works great, but of course, only works for certain Canon iPF printers.
« Last Edit: March 13, 2012, 09:22:49 am by MHMG »
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20614
  • Andrew Rodney
    • http://www.digitaldog.net/

While delta E= 4 isn't a huge error unless it's in a low chroma color, it is still puzzling to find, but just because Adobe Color utility is consistent with the null transform trick (which also relies on Adobe components) on Mac 10.7 , doesn't necessarily mean Adobe has found the "pure" unaltered data path and Apple hasn't with Colorsync Utility.

Let’s put the ColorSync test on hold a bit longer. I printed out another target in the CS utility because this first result seemed so odd, and the dEs were lower but still higher than the others. I’ve been using i1Profiler, on an iSis, with the minimum patch size it supports (6mm). This worries me a bit. So last night I built a slightly smaller patch target but with a much wider (12mm) patch and printed using ColorSync, Photoshop and ACPU. The targets were made in ColorPort and I plan to measure them today. I just don’t trust i1P at this point! Let’s wait and see if there is some ‘bug’ or other issue using the 6mm patch.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Rhossydd

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3368
    • http://www.paulholman.com

Let’s wait and see if there is some ‘bug’ or other issue using the 6mm patch.
Hmm, it was released nearly a year ago now.
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20614
  • Andrew Rodney
    • http://www.digitaldog.net/

I’m giving up on trying to print targets with the ColorSync utility. It is SUPER buggy just trying to get the damn target to print correctly!

I am unable to get the CS utility to exactly print a target from ColorPort correctly as far as scaling. I can’t figure out why! The original target out of i1Profiler printed just like those from Photoshop and ACPU and measured correctly. The newer target is scalled too large and will not measure.

Even if I resize the ColorPort TIFF to exactly match the other target, than copy and paste it into that document, I get wrong scaling. And the dialog here for size makes no sense to me (why is it defaulting to 70%)? Set it to 100%, way too big.

I did get measurements from CS5 and Adobe Color Print utility from dried targets using a wide, 12mm patch, the dE’s look pretty good:


--------------------------------------------------

dE Report

Number of Samples: 343

Delta-E Formula dE2000

Overall - (343 colors)
--------------------------------------------------
Average dE:   0.21
   Max dE:   0.70
   Min dE:   0.03
StdDev dE:   0.13

Best 90% - (308 colors)
--------------------------------------------------
Average dE:   0.18
   Max dE:   0.37
   Min dE:   0.03
StdDev dE:   0.09

Worst 10% - (35 colors)
--------------------------------------------------
Average dE:   0.47
   Max dE:   0.70
   Min dE:   0.37
StdDev dE:   0.09

--------------------------------------------------
--------------------------------------------------

I’d either print from ACPU or CS5 using the ‘null profile hack’ at this point.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Simon J.A. Simpson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 584

Andrew.  Many thanks for the answer, and for all your work.  Much appreciated.
Logged

shewhorn

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 537
    • http://

And yes, like you, I like to use Photoshop's Plugin when profiling Canon iPF printers. Quite fool-proof!

I'm not sure I'd say that as I'm seeing an average dE2k of 12.99 between ACPU and the 16 bit plugin (using CS5... I'm scanning another copy printed with CS4, but visually the CS4 and CS5 targets look identical, the ACPU version is WAAAAAAY different (OS version is 10.6.8):

dE Report

Number of Samples: 1005

Delta-E Formula dE2000

Overall - (1005 colors)
--------------------------------------------------
Average dE:  12.99
    Max dE:  49.37
    Min dE:   0.25
 StdDev dE:  10.30

Best 90% - (904 colors)
--------------------------------------------------
Average dE:  10.33
    Max dE:  29.74
    Min dE:   0.25
 StdDev dE:   6.66

Worst 10% - (101 colors)
--------------------------------------------------
Average dE:  36.84
    Max dE:  49.37
    Min dE:  29.93
 StdDev dE:   4.98

--------------------------------------------------
Logged

Scott Martin

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1315
    • Onsight

I'm not sure I'd say that as I'm seeing an average dE2k of 12.99 between ACPU and the 16 bit plugin (using CS5... I'm scanning another copy printed with CS4, but visually the CS4 and CS5 targets look identical, the ACPU version is WAAAAAAY different (OS version is 10.6.8):

If the ACPU version is the one that's different, why wouldn't you say the Plug-in is fool proof?
Logged
Scott Martin
www.on-sight.com

shewhorn

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 537
    • http://

If the ACPU version is the one that's different, why wouldn't you say the Plug-in is fool proof?

Because I woke up early (at least for me, 8AM is early)... and I've already taken some melatonin and it's beginning to take effect (I'm sleepy... ).  ;D

Printed yet another copy with a Null Transform and visually that matches the 16 bit plugin (I've yet to confirm that with some measurements). I'm hoping that's enough confirmation that all is well with the 16 bit Plugin but damn... these things don't instill a whole lot of confidence in... well, I'm still not exactly sure who it is that I should or shouldn't have confidence in. Apple writes the APIs, it's their job to maintain the codebase and gracefully deal with changes to said codebase so I feel like they deserve a big finger pointed at them but... When ACPU is supposed to solve such problems and it doesn't match something which is.. err... was... um... is??? a known quantity, I start questioning everything. 'Tis frustrating.

Time to sleep so I can look at this with a clear head tomorrow (profiling Innova's IFA-45... very very nice stuff)
Logged

Simon J.A. Simpson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 584

Which method of printing targets should we trust ?

Surely it would be the one that emulates what Photoshop does when ‘Photoshop Manages Colors’.  Although we hope (and pray) this does not introduce any colour “management” per se, for printing targets for profiling we need to replicate the conditions under which Photoshop/OS/printer–driver print our (colour managed) images.  Since we no longer seem to have any control over what a combination of Photoshop and the Mac OS do then somehow we need to determine which method of printing untagged target files appears to produce the minimum (or hopefully zero) difference error.  Evidence to date suggests which method … ?  At the moment ACPU has won out for me – subject to extended testing.

Of course Apple and Adobe engineers have the answer, but they're not telling us (or perhaps they don't know ?).

BTW Canon have a “Color Management Tool Pro” to produce targets for profiling.  Current version (3.0.0) works with the ColorMunki and i1 Pro.  Haven't played with it yet – but might.
Logged
Pages: 1 2 [3] 4 5   Go Up