Ctein Post On Profile Issue

[JimGoshorn]

Anyone seen this posting on the Online Photographer site?

http://tinyurl.com/3cfgec3

[Mark D Segal]

Jim, thanks for making us aware of this. It is fascinating. Makes me think those interested should do some controlled experiments with other kinds of files and see what happens. If Ctein's experience gets replicated often enough with different kinds of photos, that would raise questions about the effort and expense we put into printer colour management, but then again, it could be an exceptional circumstance - of what kind and why would then be the questions.

[RFPhotography]

He doesn't say what's been done with the image file in terms of size.  Not sure that would explain the difference though.  It's possible that the issue was there in the smaller print but not visible due to the actual print being smaller.  Could it be an issue of how the profiled and colour managed workflow deals with interpolation vs. how the printer managed workflow deals with interpolation?  Same # of pixels being spred out over a larger area.  Or uprezzing to create the same # of pixels for the larger print.  That would seem to be a reasonable starting point.  Maybe.

[digitaldog]

There is so little about the cm workflow not described its impossible to comment. One thing said seems odd to me:

The thing about custom profiles is that they can't fix everything perfectly, and they emphasize fixing some colors better than others.

Fix?

Knowing how the profile was built (target, spectro, package), what rendering intent, the original working space, if the  profile really was loaded for OOG overlay (which isn't all that useful, I'd plot the image and profile in ColorThink Pro) would have made the rant easier to diagnose.

[RFPhotography]

Should the method used to build the profile make a difference with the output size though, Andrew?

[jc1]

Hi,
 
Couple of issues here, imho.
 
a) Color smoothness vs color reproduction
    In general, generic driver or canned profile offers better color smoothness at the expense of better color reproduction.
b) Instrument accuracy/repeatability vs profiling test patch selection in creating custom profile
    Repeatability of spectrometer for measurement over dark or darker test patches used to be poor, custom profile with high target count may improve color reproduction over mid-tone and highlight, but often give rise to unfavorable shadow details.
c) Defective profile
   That profile works only 99.99% of the time, and you were lucky.
d) Implementation algorithm for profilers
    There are profilers that offer result more pleasing to the eye than the others.

jc

[Ernst Dinkla]

A long article with few facts on the printer settings, image specs, media used.

The blue<>purple shift is a common problem in profiling and the discussed flaw is about the same hues. The phrases about the editing of the upsampled image may conceal where it went wrong, a bit more blue hue + a slight brightening and this article would not have been written:
"No problem; moving shadow detail is routine. A little tweaking of a curves adjustment layer, a little playing around with a hue/saturation layer, some masking to limit the effects to the areas I wanted. Nothing out of the ordinary."
Often enough I have seen identical image parts on my monitor drift in opposite directions in the print. And I do not think you can compare the numbers in areas like Ctein describes.

A printer internal LUT color management process might fit this image and the conditions it is printed in (upsampling or less downsampling, the editing) just a bit better on the details mentioned. In that case it would be interesting to see what a simple profile (less patches, smoother profile) could do in an application controlled CM workflow.
 
A very slight misalignment of the printer droplets can have an effect like that in the described areas. Target patch colors will not be influenced by that. A realigned printer may suite the custom made profile better. This is related to the discussion on what resolution to feed to the printer, aliasing or other resampling artefacts can play a role, either in the small print and adjusted on color for it to get it right while the large print would not need that correction or the opposite.

The original small print made with printer LUTs probably would have been too much shifted to blue in the eyes of Ctein.

Reading Joseph Holmes' workflow I wonder whether a PC printer server system running Qimage Ultimate is not an easier solution. I do not question his knowledge on printing. I have seen very nice prints he made. The workflow says that he lost confidence in Mac CM and avoids it as much as possible.

Reading the comments there the article could create new tribes that reject custom profiling, application CM or worse CM at all.  That the image was saved by using the most basic printer CM does not imply that the flaw was in the CM part, Ctein could have added some warnings in the article like he does in his replies on the comments.


met vriendelijke groeten, Ernst

New: Spectral plots of +250 inkjet papers:

http://www.pigment-print.com/spectralplots/spectrumviz_1.htm

[digitaldog]

Should the method used to build the profile make a difference with the output size though, Andrew?

It might give some idea of what and why the issue is seen.

[Waeshael]

Yes, Joe Holmes knows more about color management than anyone, and his current method is obviously correct for his hardware and software. There has been a lot of discussion about Colorsync on the Mac which has changed over the generations, and techniques that worked before are no longer working in Lion and even in Snow Leopard  - some versions.
Joe has always taught us since the 90's that you let the printer manage the color after you have converted the image file to the color profile of the ink/paper you are to use. Turn off colorsync and make no adjustments to the printer controls, and let the printer take over the color management. It worked for me in Tiger. I don't print anymore, but my friends do and they still use Joe's technique under Mac OS Tiger and Leopard. Lion was a whole other issue.

here is his website which discusses these printing issues with the Mac.
http://www.josephholmes.com/

[Ernst Dinkla]

Joe has always taught us since the 90's that you let the printer manage the color after you have converted the image file to the color profile of the ink/paper you are to use. Turn off colorsync and make no adjustments to the printer controls, and let the printer take over the color management.

http://www.josephholmes.com/

I would think the last is not the right description. But OS-X could be more complex than I am used to. The printer has to do its job with its CM off. At least that is what I do in case I do not trust Qimage's handling of a QTR made profile, I make the color space to printer profile conversion in PS, save the file and load it in Qimage with CM off and let the printer print with CM off (or set on: let application handle CM, which it does not). The two will then not interfere on image data that is already converted. Windows CM has always been decent enough not to interfere.


met vriendelijke groeten, Ernst

New: Spectral plots of +250 inkjet papers:

http://www.pigment-print.com/spectralplots/spectrumviz_1.htm

[NickCroken]

I am having very similar issues to this, the full thread here http://www.luminous-landscape.com/forum/index.php?topic=56948.0

I tried emailing him with suggestions but he hasn't gotten back to me.

[Ernst Dinkla]

I am having very similar issues to this, the full thread here http://www.luminous-landscape.com/forum/index.php?topic=56948.0

I tried emailing him with suggestions but he hasn't gotten back to me.

And I wonder whether more reports on green shifts popping up have anything to do with an abnormal Photoshop 12-04 saving of files as discussed here:

http://feedback.photoshop.com/photoshop_family/topics/corrupt_jpg_from_12_0_4

http://ddisoftware.com/tech/qimage/off-color-problem/30/?PHPSESSID=3fjb11ea1e6j9m2e8bee71k3a1

It could be happening with more file formats than jpg.


met vriendelijke groeten, Ernst

Try: http://groups.yahoo.com/group/Wide_Inkjet_Printers/

[Ronny Nilsen]

Ctein have posted a followup on TOP on this issue:

http://theonlinephotographer.typepad.com/the_online_photographer/2011/09/when-good-profiles-turn-bad.html

Ronny

[digitaldog]

Ctein have posted a followup on TOP on this issue:

http://theonlinephotographer.typepad.com/the_online_photographer/2011/09/when-good-profiles-turn-bad.html

He writes: I still don't know for certain that there are out-of-gamut colors in that photograph for my printer. There's no test that I or Dave or Mike can run to determine that for certain.

Sure there are. You simply need something like ColorThink Pro. You can plot the image gamut over the gamut of the printer.

This is important because it means I can't entirely count on gamut previews to tell me if all my colors are properly printable.

I could have told him that! The old gamut overlay is far from useful. Consider it treats a color that is a tiny bit and one that’s way out of gamut the same in terms of the ugly overlay. The overlay was kind of useful prior to 1998 and Photoshop 5 which introduced soft proofing.

Without getting technical about it, the software that creates the profile that color management uses has to make some assumptions about how that mapping occurs. For the sake of computation, it also has to treat some out-of-gamut colors as if they were in gamut so that it can deal with them at all. Consequently, there is not really a clear and accurate boundary in the profile or in the gamut preview between colors that are in gamut and colors that aren't. The boundary is fuzzy.

Well that’s pretty simplistic and don’t take into account the differences in how out of gamut colors are treated by rendering intents. It doesn’t take into account there is no standard or fixed way to build a perceptual rendering intent and we have tools today that allow us to tweak this proprietary and custom process of dealing with OOG colors. Its why I initially asked if this fellow had provided any data about how the profile was created. Every profile building application creates a different perceptual rendering intent and many have loads of options (in i1P, three presets and several sliders making the creation virtually unlimited). Was this a D50 profile or was the ability to measure and create the profile with a custom illuminant available and options therefore over the chromatic adaptation algorithms as we have in i1Profiler?

The printer, though, doesn't think so. If it gets thrown a color to print that's outside its comfort range, it will print something, but it may not do it very well. ICC profiles don't understand a particular printer's idiosyncrasies very well.

This is oddly stated and isn’t really factual. It prints what the values the document describes so to say it throws a color outside a comfort range and that ICC profiles don’t understand the printers idiosyncrasies just doesn’t say anything useful or factual. Guy may be an amazing photographer and printer. But he should probably not attempt to teach or even blog about color management with such “terminology“. Its confusing enough without having to resort to this kind of language.

If profiles can cause problems like this and give misleading results, why not just avoid them all the time?

Clearly this is the profiles fault. Not the user. Not how the profile was built or what the user selected for a rendering intent. Or how the user ignored the soft proof presumably. Or how the user didn’t edit the document while soft proofing. I’m not suggesting this happened here for sure. As I wrote initially, there is far too little data supplied to come to any conclusions about the issues. I only wish this writer would do the same. Yes, a poorly built profile could be an issue. Or a really good profile could have been casually used without full understanding of what they can do and how to use them. I really don’t know, I wasn’t there. But the blog posts are less than useful in defining the real problems or the real solutions.

As I wrote previously, if you consistently get better results from printer-controlled color then full color management, something's wrong with your system.

This is the kind of text that makes me wonder if the author really understands the role of ICC profiles, color management and process control. ICC color management assumes consistency otherwise we have to calibrate to use said ICC profile. Fortunately, modern ink jets, certainly Epson’s need no calibration because when functioning properly they are incredibly consistent! Use a profile, don’t use a profile, those facts remain. Profiles only tell us the condition of the device at the time we printed a target and measured it. If the printer changes its behavior, the profile is useless. A poorly built profile will consistency produce a poor print. A good profile consistently used in less than idea fashion will consistently produce less than ideal output. The same would be true using the Printer Manages Color process, again assuming in both cases, the printer is consistent. Having a good profile that one properly uses consistently, meaning soft proofing to pick the ideal rendering intent, editing prior to conversion based on the proper soft proof can consistently produce a superior print compared to PMC due largely to the fact that PMC gives no control over a rendering intent or soft proofing.

[madmanchan]

Hi Andrew, the interesting part that I'm not sure about is whether the driver mixes ink differently depending on whether you're using No Color Adjustment (NCA) or one of the other driver modes. This could explain why some specific color ranges are smoother in one driver mode vs another.

[digitaldog]

Hi Andrew, the interesting part that I'm not sure about is whether the driver mixes ink differently depending on whether you're using No Color Adjustment (NCA) or one of the other driver modes. This could explain why some specific color ranges are smoother in one driver mode vs another.

I think that is entirely possible. I know that if you print a target for a profile using NCA versus Epson controls, there’s a big difference in linearity as well as color gamut.

[Wayne Fox]

Yes, Joe Holmes knows more about color management than anyone, and his current method is obviously correct for his hardware and software. There has been a lot of discussion about Colorsync on the Mac which has changed over the generations, and techniques that worked before are no longer working in Lion and even in Snow Leopard  - some versions.
Joe has always taught us since the 90's that you let the printer manage the color after you have converted the image file to the color profile of the ink/paper you are to use. Turn off colorsync and make no adjustments to the printer controls, and let the printer take over the color management. It worked for me in Tiger. I don't print anymore, but my friends do and they still use Joe's technique under Mac OS Tiger and Leopard. Lion was a whole other issue.

here is his website which discusses these printing issues with the Mac.
http://www.josephholmes.com/

While there is no doubt many have had issues with printing beginning with apples change to the core printing and color management processes in Leopard and followed by Snow Leopard, this isn't necessary.  This information from Joe is quite old, and while I agree he is one of the most knowledgeable color management experts out there, it really doesn't apply anymore and convolute's what is now a pretty simple workflow.

Currently with a correct install of either 10.6 or 10.7  and CS5 (or Lightroom) printing to Epson printers will produce accurate color managed output.  While this isn't true when trying to print targets (and even Joe's methods there might fail), tricking the Mac's OS this way is only necessary if sending the OS an untagged unmanaged file, because by default the OS wants to handle that (and to be honest a good OS should try to handle that since 99.9% of the users have no clue about color management and for them some under the hood management can be a good thing).  Unfortunately Apple didn't think about the few of us  that sometimes needs to get a file through the OS to the printer and out on paper without any color management (a target for a profile).  Thus the birth of adobe's utility which is designed to do just that.

However, if you convert your images in photoshop to the target printer/paper profile, and then send that image to the printer with Adobe's utility which means neither the OS or the printer will do anything with the colors, the results will be identical to printing from the normal CS5 pipeline letting Photoshop manage the colors.  I had a customer that had problems like this and didn't believe me, so I literally did this Bill Atkinsons Lab Test Page - they were identical. The conversion is done by photoshop in both workflows,  and as  long as PS sends the image through the OS to the printer tagged, the OS will leave it alone and not step on it.

[John Nollendorfs]

Ctein finally concluded the printing problem was traced to out of gamut issues of the file, which was solved by letting the printer manage the color.

Interesting reading all the posts by a wide array of folk.

Bottom line, be careful when pushing your image's pixels around. Custom profiles do wonders in making most images really look great, but if you have strayed into out of gamut territory, beware! Those profiles will have trouble putting those values into printable form. Invariably when out of gamut shows it's ugly head, head for using the hue/saturation layer to carefully message that file back into printable gamut.

[artobest]

The old gamut overlay is far from useful. Consider it treats a color that is a tiny bit and one that’s way out of gamut the same in terms of the ugly overlay. The overlay was kind of useful prior to 1998 and Photoshop 5 which introduced soft proofing.

The problem, as I understand it, is that the image looked fine in soft-proofing. Only the final print exhibited the issues (abrupt colour and tonal boundaries).

I still use the gamut warning from time to time, depending on the image. You can make an educated guess about how far out of gamut the colours are by experimenting non-destructively with the saturation sliders.

[digitaldog]

A printer has a gamut which is what it is! A profile defines that gamut. As Eric pointed out, altering the driver could be the cause, it alters the gamut and ink delivery.

Using the NCA with Epson printers produces the widest gamut. So I still question if the issue here is a gamut issue. It could still be based on the rendering intent used, how the profile was built etc. Since we’ve had zero info from the source, its a big guessing game.

If you use another mode (Epson Color Controls), you get a more linear output with a reduced gamut. I’ve built profiles both ways (NCA and Color Controls). Its super easy to plot the differences and do so over an image.

Until there are more data points, its simply not fair to complain about using profiles instead of the Printer Manages Color options.