Pages: 1 [2]   Go Down

Author Topic: Ctein's follow-up article on printer-managed color  (Read 9942 times)

Tony Jay

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2965
Re: Ctein's follow-up article on printer-managed color
« Reply #20 on: October 03, 2015, 09:37:26 pm »

What has had some "distinquished folks" (perhaps myself included) bent out of shape is not Ctein's favorable recommendation of a printer manages workflow. It's his resolute opinion that ICC profiles were an epic fail for him in some instances, and never really better in almost every other instance. That conclusion flies in the face of what my own hard fought hard won experience is with an ICC profiled workflow. I'd be the first to tell you a really well calibrated ICC workflow, complete with highly accurate soft proofing is definitely not easy to set up. Unlike what some marketing guys or amateur photography forums will tell you, you don't truly get all the way there to a precise soft proof by taking any old monitor, buying some inexpensive calibrator puck, and using a generic profile provided by some third party media vendor. I wish we could all get there easily following that recipe, and sometimes that recipe works "good enough" for some folks to be happy, but when we are talking about a master printmaker pulling out all stops to work through a sophisticated ICC profiling approach and then reporting it didn't work, yeah, that's when the little antennae in the back of my head start to tell me "what the hell, what happened?".   So, I repeated Ctein's exercise with his one of his most troublesome images (Ctein kindly sent me his source file for  the Apollo Soyuz image bathed in searchlights under a nighttime sky).

1) I replicated the print he got using his printer manages color/Epson color controls workflow, and I agree it was a very good print.

2), I printed again using PS manages colors and an ICC profile (custom built using PM5 software using the default logo colorful setting). Again, I replicated Ctein's experience. It did not produce an "out of the box" print that was as good as the one created in step 1. Ctein and I totally agree on that score.

3). I then noted that the source file itself has unwanted encoded color errors in the hue constancy of the colors in the night time beams of light, such that some corrective action must be taken to get the print he wants (where the beams of light should trail off uniformly in hue throughout the image). That means the LUT which Epson hardwired into its Epson Color control workflow is doing some error correction in the blue sector of the color space to fix the classic "blue turns purple" problems common to many digital image color reproduction methods. The Epson "fix" just happens to help this particular image in a good way, but may cause errors in others. And on a wide gamut monitor, the aRGB source image definitely does not come close to matching the printer manages color prints. So, let's clear up that misconception right now. Ctein was editing an image in aRGB color space, but looking at his edits on a monitor generically factory calibrated to mimic a native sRGB-like gamut.  As such, should it surprise us that Ctein believes soft proofing is not very accurate? Printer manages color did not render a perfect reproduction of the source file, perfectly displayed on his monitor. It was just a method that happened to "play nicely" with this particular image that Ctein printed along with others to test his printer manages color hypothesis.

4). In step 2, I had verified that Xrite ICC default profiles' perceptual and relative rendering tags do in fact exaggerate the color errors of Ctein's Apollo Soyuz source image because Xrite's default profile recipes favor colorfulness at the expense of tonality (i.e. the typical Xrite-built profile favors colorfulness over the preservation of lightness relationships among the various image elements). As such, these profiles were not a good staring point for any subsequent image edits. Because Xrite's default profiling algorithm(s) are so pervasive in the industry, and because Andrew Rodney likely made Ctein a custom profile using said default settingss which Ctein also stated failed to fix his issues, it's fair to say that Ctein does not like ICC profile(s) renderings which favor colorfulness over tonal accuracy.  In my own work, I switch between Logo colorful and logo classic frequently, so I'm sympathetic to anyone who finds precise color accuracy lacking in many typical ICC profiles. Xrite and other profile making vendors really need to emphasize these aspects of "color mapping" more in their literature than is typically found.

5. Once I understood what Ctein's ICC profile problem was, namely that one needs to start with a profile that favors the preservation of tonal relationships at the expense of colorfulness. I reached into my "bag of ICC profile tricks' and applied Xrite's original PM5 "logo classic" to soft proof the Apollo Soyuz image on my calibrated high gamut NEC spectraview II display. The softproof showed me that logo classic was providing a much better starting point for final image edits needed to match or exceed Ctein's printer managed color output because log classic favors tonal (lightness) accuracy at the expense of hue and chroma.

6). I applied just two image edits in my carefully calibrated softproofing mode, one hue/sat layer in PS, and one curve correction. Using a sound softproofing workflow as my guide, I was able to make my corrections in a matter of minutes while working on Ctein's source file image.(note: while the edits were few, they were indeed tricky moves on this file, ones that would be very hard pressed to do without accurate soft proofing guiding the effort). When I had completed those edits, I had on screen a soft proofed image I actually thought was superior to the print I made with Ctein's method. Yah, I know, my word against his as to which was a preferable rendition of the source file, but bear in mind, what I liked better, I was seeing on screen, and I was expecting it to print the way I was seeing it on screen. Moreover, with just one more layer, and a couple of trims on the "fill" percentages of the PS layer sliders,  I knew I could get really really close to Ctein's preferred "printer manages color" rendition of this image if I wanted to, so I wasn't sacrificing any flexibility in color and tone reproduction and in being able to deliver just what anyone would expect in print from that source file, IMHO, by using an ICC profiled workflow.

7). The resultant print matched my edits very accurately, and because I liked this "interpretation" of the image better than what I got when printed with the "printer manages color workflow" on my P600 printer, I ended up with an ICC profiled print off this file that I liked better as well. Moral of the story:  ICC profile managed workflows are not inferior to printer manages color as Ctein seems to assert, but you do have to understand ICC profiling strengths and weaknesses very well.  Indeed, when done properly, soft proofing with the right ICC profile rendering recipe gets you where you want to go and also tells you very closely what to expect in the final print. My faith in my ICC profiling workflow and all those years I have spent mastering it was restored after completing this little exercise :).

All that said, for many folks who don't want to take on all the overhead of a sophisticated ICC custom profile workflow, Ctein has done you a favor, and I admire him for telling us about his personal approach to digital printmaking. Just Use printer manages color with a newer Epson printer, and you probably will get pleasing prints without investing nearly the effort I and others have made in exacting ICC profiled workflows over the years. However, if you are willing to invest the time and take on the steep learning curve of a sophisticated soft proofing environment, you will indeed be rewarded by a richly deserved wysisyg image editing methodology that is extensible to all printers not just Epson printers, and even ones at remote sites not just in your immediate reach.

cheers,
Mark
http://www.aardenburg-imaging.com
Thank you Mark.
Great summary!

Tony Jay
Logged

Schewe

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6229
    • http:www.schewephoto.com
Re: Ctein's follow-up article on printer-managed color
« Reply #21 on: October 03, 2015, 10:00:41 pm »

Moral of the story:  ICC profile managed workflows are not inferior to printer manages color as Ctein seems to assert, but you do have to understand ICC profiling strengths and weaknesses very well.  Indeed, when done properly, soft proofing with the right ICC profile rendering recipe gets you where you want to go and also tells you very closely what to expect in the final print. My faith in my ICC profiling workflow and all those years I have spent mastering it was restored after completing this little exercise :).

I appreciate your efforts (cause there was no way I would go down this rabbit hole) to sus out the potential reasons why the Profile vs PMC seemed to fail. Sadly, many users don't know how to distinguish the differences in profile generating software.

But, it's encouraging that one CAN use a profile based workflow and meet or exceed  the PMC method.
Logged

jed best

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 398
Re: Ctein's follow-up article on printer-managed color
« Reply #22 on: October 03, 2015, 10:19:47 pm »

Mark,

Your response is both very insightful and interesting. It would be terrific if someone with greater knowledge than myself could  write a tutorial or article on LULa so that those who are interested, could use icc profiling to produce prints of maximum quality and be aware of the pitfalls in the process so that they may compensate for them. It seems to me that those who are reading and contributing to these forums are not interested in producing average prints but prints that reach 99% of what their file contains.

Just my two cents.
Logged

ashaughnessy

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 62
    • My wordpress blog
Re: Ctein's follow-up article on printer-managed color
« Reply #23 on: October 04, 2015, 03:14:43 am »

Mark, thanks for all that, but the take-away for me is your last paragraph. While the full ICC approach may work, your suggesting it only works with a great deal of money laid out (for very expensive monitors, very expensive profiling hardware). Many of us can't afford that. I currently use the ICC approach, I have a cheaper monitor, a cheaper puck to profile my monitor, and I've bought custom paper/printer profiles and I get reasonable results I'm happy with but for a very small set of papers. I can't really afford to try every new paper that comes out because of the cost and time of getting a new custom profile made for each paper. In the future, when my current printer expires, I'll consider a new P600 and I'll consider adopting the printer-manages-colour approach and see if that works for me. I think many of us thought that the ICC approach was the answer, it turns out to be just one answer and an expensive one at that.
Anthony

EDIT - just realised I'm on windows, so I'm stuck with ICC for the foreseeable future. Doh!
« Last Edit: October 04, 2015, 03:20:17 am by ashaughnessy »
Logged
Anthony Shaughnessy
https://anthonyshaug

Tony Jay

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2965
Re: Ctein's follow-up article on printer-managed color
« Reply #24 on: October 04, 2015, 05:57:51 am »

Anthony, I am not sure that your conclusions are completely correct.

Not for one second would I suggest that high-end printing is not expensive to a degree however one does not need, for example, to do one's own ICC profiling.
Get yourself a decent monitor (my personal recommendation is one from the Spectraview family - as big as one can afford) and a decent puck to profile and calibrate it.
Depending on the paper one is using the paper manufacturer may or may not provide ICC profiles worth using. The Canson papers I use happen to give great results but that is not universal.
If the profiles were horrible and I wanted to continue using that paper I would ask someone like Andrew Rodney (our very own Digital Dog) to profile a target printed on my printer. Andrew Chan also used to do ICC profiles but I think he no longer offers this service.

Getting great printing results is not just a function of financial expense. I print really large (A2 is the smallest) and I accept that the printer is expensive but the outlay to get a really good softproofing workflow going is not out of reach.
The point I want to make about printing really large is this: If the result is subpar there is no hiding this when the print real estate is in the order several square feet.

Perhaps the biggest outlay for me has been time and experience as well as lots of reading and research to understand what I was doing. Nonetheless I believe that any current deficiencies in my printing results are in my aesthetic judgement rather than the processing pipeline per se.

Food for thought

Tony Jay
Logged

ashaughnessy

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 62
    • My wordpress blog
Re: Ctein's follow-up article on printer-managed color
« Reply #25 on: October 04, 2015, 07:37:14 am »

Tony, to moderate my post a little, my only real problem, which is a big one, with my ICC workflow is not being able easily to try new papers. One of the fantastic benefits of inkjet printing is the tremendous choice of papers available. However, to really see what they're capable of, you need to get a custom profile for each one. This is quite a big cost, both in money and time. So I limit myself to just the two papers I regularly use.

Maybe I need to change my attitude on this. Maybe I should try printing more with manufacturer canned profiles and not worry about the resulting problems and then get a custom profile once I've narrowed down the choice. It would be nicer though to have a setup where I could get closer to optimal results without having to wait a week and spend £30 or more on a custom-made profile from someone like Andrew. I should also be clear, I'm not "blaming" ICC for any of this, simply pointing out that finding out there may be an alternative is a good thing. I've had good results from the custom profiles I currently use.

Anthony
Logged
Anthony Shaughnessy
https://anthonyshaug

cortlander

  • Full Member
  • ***
  • Offline Offline
  • Posts: 158
Re: Ctein's follow-up article on printer-managed color
« Reply #26 on: October 04, 2015, 08:34:20 am »

What has had some "distinquished folks" (perhaps myself included) bent out of shape is not Ctein's favorable recommendation of a printer manages workflow. It's his resolute opinion that ICC profiles were an epic fail for him in some instances, and never really better in almost every other instance. That conclusion flies in the face of what my own hard fought hard won experience is with an ICC profiled workflow. I'd be the first to tell you a really well calibrated ICC workflow, complete with highly accurate soft proofing is definitely not easy to set up. Unlike what some marketing guys or amateur photography forums will tell you, you don't truly get all the way there to a precise soft proof by taking any old monitor, buying some inexpensive calibrator puck, and using a generic profile provided by some third party media vendor. I wish we could all get there easily following that recipe, and sometimes that recipe works "good enough" for some folks to be happy, but when we are talking about a master printmaker pulling out all stops to work through a sophisticated ICC profiling approach and then reporting it didn't work, yeah, that's when the little antennae in the back of my head start to tell me "what the hell, what happened?".   So, I repeated Ctein's exercise with his one of his most troublesome images (Ctein kindly sent me his source file for  the Apollo Soyuz image bathed in searchlights under a nighttime sky).

1) I replicated the print he got using his printer manages color/Epson color controls workflow, and I agree it was a very good print.

2), I printed again using PS manages colors and an ICC profile (custom built using PM5 software using the default logo colorful setting). Again, I replicated Ctein's experience. It did not produce an "out of the box" print that was as good as the one created in step 1. Ctein and I totally agree on that score.

3). I then noted that the source file itself has unwanted encoded color errors in the hue constancy of the colors in the night time beams of light, such that some corrective action must be taken to get the print he wants (where the beams of light should trail off uniformly in hue throughout the image). That means the LUT which Epson hardwired into its Epson Color control workflow is doing some error correction in the blue sector of the color space to fix the classic "blue turns purple" problems common to many digital image color reproduction methods. The Epson "fix" just happens to help this particular image in a good way, but may cause errors in others. And on a wide gamut monitor, the aRGB source image definitely does not come close to matching the printer manages color prints. So, let's clear up that misconception right now. Ctein was editing an image in aRGB color space, but looking at his edits on a monitor generically factory calibrated to mimic a native sRGB-like gamut.  As such, should it surprise us that Ctein believes soft proofing is not very accurate? Printer manages color did not render a perfect reproduction of the source file, perfectly displayed on his monitor. It was just a method that happened to "play nicely" with this particular image that Ctein printed along with others to test his printer manages color hypothesis.

4). In step 2, I had verified that Xrite ICC default profiles' perceptual and relative rendering tags do in fact exaggerate the color errors of Ctein's Apollo Soyuz source image because Xrite's default profile recipes favor colorfulness at the expense of tonality (i.e. the typical Xrite-built profile favors colorfulness over the preservation of lightness relationships among the various image elements). As such, these profiles were not a good staring point for any subsequent image edits. Because Xrite's default profiling algorithm(s) are so pervasive in the industry, and because Andrew Rodney likely made Ctein a custom profile using said default settingss which Ctein also stated failed to fix his issues, it's fair to say that Ctein does not like ICC profile(s) renderings which favor colorfulness over tonal accuracy.  In my own work, I switch between Logo colorful and logo classic frequently, so I'm sympathetic to anyone who finds precise color accuracy lacking in many typical ICC profiles. Xrite and other profile making vendors really need to emphasize these aspects of "color mapping" more in their literature than is typically found.

5. Once I understood what Ctein's ICC profile problem was, namely that one needs to start with a profile that favors the preservation of tonal relationships at the expense of colorfulness. I reached into my "bag of ICC profile tricks' and applied Xrite's original PM5 "logo classic" to soft proof the Apollo Soyuz image on my calibrated high gamut NEC spectraview II display. The softproof showed me that logo classic was providing a much better starting point for final image edits needed to match or exceed Ctein's printer managed color output because log classic favors tonal (lightness) accuracy at the expense of hue and chroma.

6). I applied just two image edits in my carefully calibrated softproofing mode, one hue/sat layer in PS, and one curve correction. Using a sound softproofing workflow as my guide, I was able to make my corrections in a matter of minutes while working on Ctein's source file image.(note: while the edits were few, they were indeed tricky moves on this file, ones that would be very hard pressed to do without accurate soft proofing guiding the effort). When I had completed those edits, I had on screen a soft proofed image I actually thought was superior to the print I made with Ctein's method. Yah, I know, my word against his as to which was a preferable rendition of the source file, but bear in mind, what I liked better, I was seeing on screen, and I was expecting it to print the way I was seeing it on screen. Moreover, with just one more layer, and a couple of trims on the "fill" percentages of the PS layer sliders,  I knew I could get really really close to Ctein's preferred "printer manages color" rendition of this image if I wanted to, so I wasn't sacrificing any flexibility in color and tone reproduction and in being able to deliver just what anyone would expect in print from that source file, IMHO, by using an ICC profiled workflow.

7). The resultant print matched my edits very accurately, and because I liked this "interpretation" of the image better than what I got when printed with the "printer manages color workflow" on my P600 printer, I ended up with an ICC profiled print off this file that I liked better as well. Moral of the story:  ICC profile managed workflows are not inferior to printer manages color as Ctein seems to assert, but you do have to understand ICC profiling strengths and weaknesses very well.  Indeed, when done properly, soft proofing with the right ICC profile rendering recipe gets you where you want to go and also tells you very closely what to expect in the final print. My faith in my ICC profiling workflow and all those years I have spent mastering it was restored after completing this little exercise :).

All that said, for many folks who don't want to take on all the overhead of a sophisticated ICC custom profile workflow, Ctein has done you a favor, and I admire him for telling us about his personal approach to digital printmaking. Just Use printer manages color with a newer Epson printer, and you probably will get pleasing prints without investing nearly the effort I and others have made in exacting ICC profiled workflows over the years. However, if you are willing to invest the time and take on the steep learning curve of a sophisticated soft proofing environment, you will indeed be rewarded by a richly deserved wysisyg image editing methodology that is extensible to all printers not just Epson printers, and even ones at remote sites not just in your immediate reach.

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

Thank you, Mark. A brilliant post!
Logged
cortlander

Ferp

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 295
Re: Ctein's follow-up article on printer-managed color
« Reply #27 on: October 04, 2015, 09:07:55 am »

EDIT - just realised I'm on windows, so I'm stuck with ICC for the foreseeable future. Doh!

No you're not.  Read the comments on ToP, especially the end.

That said, I'm sticking with ICCs.  I'm wary of black boxes.  They might contain Schrodinger's cat.
Logged

Jager

  • Full Member
  • ***
  • Offline Offline
  • Posts: 207
    • E vestigio
Re: Ctein's follow-up article on printer-managed color
« Reply #28 on: October 04, 2015, 09:35:26 am »

Like most of us, I began photography many years ago by simply pointing my camera at scenes that I found interesting.  I ended up with some nice pictures, sometimes.  But, more often than not, the prints that came back from the lab didn't translate what I had seen.  I found photography to be laced with disappointment.

Persevering, though, I made my way to photography's first truth:  the camera does not record light the same ways our eyes do. 

Fast forward many years and I now fully appreciate that a good photographer can look at a scene and instantly know how their gear will translate it.  They understand how different film stocks render, how different sensors have varying strengths, how color will or won't be imitated, how a scene might flare through a given lens and how that flare might or might not be useful, how that red flower or that blue dress will map to a tone of gray when shooting black and white, how that red filter will shift those tonal relationships, how a large aperture will capture a very different scene than a small aperture, and on and on.

Photography is a narrowing process, taking the inherently wide gamut of human vision and pushing it through the limiting confines of sensor and electronic processing (or emulsion and chemistry) and optics.  What results is a smaller gamut version of what we saw with our eyes.

What distinguishes every serious photographer in the world from the multitudes wielding their ubiquitous cell phone cameras is their knowledge of those limitations, of that narrowing process.  Every good photographer I've ever known can predict how an image will be rendered, by whatever equipment they have at hand, long before they press the shutter.

Printmaking is a further narrowing, into an even smaller gamut.

A master printmaker in the darkroom didn't print by rote.  When they chose a particular paper, or a particular contrast, when they dodged this, or burned that, when they decided to leave a print in the enlarger for an extra fifteen seconds, they didn't do it blindly.  They did it because of a particular look they sought.  They did it knowingly, predicting the result they would achieve.

We, most of us, seek the same thing in digital printing.  We make changes in our editing software knowingly, fully expecting - predicting - how those changes will be rendered on paper.  Soft proofing with ICC-managed profiles, done properly - which is to say with the requisite knowledge, process, and equipment - allows us to pre-visualize what we're trying to achieve, then provides a straightforward path to doing so.

Which isn't to say there aren't black-box approaches that don't provide good results.  But just as I don't know any great photographers who don't pre-visualize the images they are making, I don't know any master printmakers who throw darts at the wall, hoping for a nice print.

Rhossydd

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3369
    • http://www.paulholman.com
Re: Ctein's follow-up article on printer-managed color
« Reply #29 on: October 04, 2015, 10:07:05 am »

One of the fantastic benefits of inkjet printing is the tremendous choice of papers available. However, to really see what they're capable of, you need to get a custom profile for each one.
Not sure I'd agree with that. Using default or OEM profiles will usually give you very good idea of what a paper is capable of and what's it's physical characteristics are.
Once you decided you like a surface, base colour and weight of a paper, which IMHO is a major part of paper choice, then you can get a custom profile done to extract the final best performance from the paper.
Quote
having to wait a week and spend £30 or more on a custom-made profile from someone
My business (www.colourprofiles.com) delivers normally profiles within 24hrs of receipt of the charts, so delivery time is down to postage........... and is half that cost.
If you'd like to try out the possibilities of different profile gamut mappings, as discussed above by Mark, we'd be happy to advise and offer extra options too.
Quote
it only works with a great deal of money laid out (for very expensive monitors, very expensive profiling hardware). Many of us can't afford that. I currently use the ICC approach, I have a cheaper monitor
It always puzzles me why people spend huge amounts on camera kit and printers then skimp on a monitor. Without a good monitor you'll never know just what your images really look like.
Logged

robgo2

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 388
    • Robert Goldstein Photography
Re: Ctein's follow-up article on printer-managed color
« Reply #30 on: October 04, 2015, 11:27:50 am »

Hi Rob,

I wrote " ... So, his conclusions should be taken seriously, not received wisdom  but an important point that can further our understanding of how to optimize print quality..."  What part of "not received wisdom" did you not understand? How is that inconsistent with "must be verified independently?"  The idea that we are doing science by exploring the function of the existing options in a commercially available printer is preposterous and aggrandizing. Ctein is an authority on prints, if not printing. Maybe we should open our minds to the possibility that there are other ways to do things and not criticize the guy for telling us about his observations.

Perhaps you missed the point that I was agreeing with you about not accepting Ctein's conclusions blindly.  But I do not agree with your dismissing the scientific method when it is applied to mundane subjects.  This is how we attempt to approach true knowledge about practically everything:  1. Pose a hypothesis, 2. test the hypothesis either by experimentation or observation, and 3. test its ability to predict future outcomes.  If the hypothesis passes these tests, then it is taken as provisionally true.  The scientific method applies just as well to color management as it does to "real" science.  The only hooker is the subjective element as to what constitutes the "best" outcome, but that, too, is present in much scholarly research.  I suspect that most people making photographic prints could not care less about the subject, but this is how it can be properly studied.  There's nothing grandiose about it.

Rob
« Last Edit: October 04, 2015, 12:17:18 pm by robgo2 »
Logged

MHMG

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1285
Re: Ctein's follow-up article on printer-managed color
« Reply #31 on: October 04, 2015, 11:43:21 am »

Mark, thanks for all that, but the take-away for me is your last paragraph. While the full ICC approach may work, your suggesting it only works with a great deal of money laid out (for very expensive monitors, very expensive profiling hardware). Many of us can't afford that.

I find it hard in rapidly evolving forum posts to always "calibrate" my replies perfectly. We've been discussing an author, photographer, and experienced printmaker, Ctein, telling us softproofing is only approximate for him, and so he'd rather just iteratively print to a closed print system (Epson color controls and select gloss/luster papers). In the context of high end mastery of fine art printmaking I was simply trying to say that precision soft proofing and mastery of an ICC profiled workflow is definitely available at higher performance level than Ctein has apparently ever investigated, and like many folks, he may simply not want to invest more money in a technique he's happy to live without.

So yes, you can up your game and get a really high end monitor, pursue 16 bit workflows, wide gamut working spaces, and buy a higher end pro printer, etc. etc., to get more than Ctein has gotten out of an ICC profiled workflow to date, but I didn't mean to imply one can't get very reasonable results using a more budget based approach to an ICC color managed workflow. I softproofed with very good results on CRT monitors back in the day (with the earliest monitor calibration software) and pretty much prepped every image file in sRGB working spaces, then graduated to an Apple Cinema display and improved third party calibrators several years ago (still primarily using srGB working space in PS to edit my images for print). I only recently upped my game to a NEC Spectraview II, and moved to proPhoto RGB or sometimes aRGB as the image editing workspace in PS. At each step along the way, I was improving my accuracy and precision (and paying more to do it). Cutting right to the chase, critical use of an accurate soft proofing methodology allows me to make very intricate image edits that I simply couldn't make if I was flying blind, just hoping to score the right moves with each successive print iteration. This may be the biggest factor that's being overlooked in this whole discussion. Soft proofing not only allows me to make informed edits, it allows me to make very sophisiticated edits. This is a part of any color management tutorial that goes decidedly under reported! Could I tell you about my intricate image editing moves in a basic article or tutorial?  No. But I could teach these more involved image editing techniques in a series of lectures that would more closely parallel a college level semester course on image editing and color management. The key to said course would be the ability to implement a viable and trustworthy soft proofing technique. How can one do that without ICC profiles in today's digital imaging arena?

Like any endeavor, it's about setting reasonable expecations for time, and money. As a further example to the opposite of what I just previously described, i.e., vastly simplified and very cost effective closed loop image reproduction methods similar to what Ctein is advocating, my daughter recently asked me to print some photos from her smartphone. Rather than try to go all out, I decided that she'd probably be thrilled with a much more basic way to accomplish her request. I had her download an app "Canon Easy Photo print". After a minor glitch of about twenty minutes getting her android phone to recognise my wifi network, she opened up the Canon app, picked from a limited number of choices "4x6 photo", "bordered" (she didn't want borderless), and then I told her what media setting (Canon Photo Paper Plus Glossy II) to use. The app let her quickly choose those options, but very little else!  She hit "print", and with that Canon paper loaded in my Pro-100 printer, the resulting print(s) popped out a few minutes later, and they were really quite good.  She was very pleased. This approach is yet another another example of a closed loop system (i.e printer, paper, ink, all designed to work directly together with simple driver settings). It completely met her needs and expectations.  But don't try to use that approach to do serious image edits or print on thicker fine art media. It would really choke :D

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

« Last Edit: October 04, 2015, 07:39:54 pm by MHMG »
Logged

Ernst Dinkla

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4005
Re: Ctein's follow-up article on printer-managed color
« Reply #32 on: October 05, 2015, 08:08:05 am »


3). I then noted that the source file itself has unwanted encoded color errors in the hue constancy of the colors in the night time beams of light, such that some corrective action must be taken to get the print he wants (where the beams of light should trail off uniformly in hue throughout the image). That means the LUT which Epson hardwired into its Epson Color control workflow is doing some error correction in the blue sector of the color space to fix the classic "blue turns purple" problems common to many digital image color reproduction methods. The Epson "fix" just happens to help this particular image in a good way, but may cause errors in others. And on a wide gamut monitor, the aRGB source image definitely does not come close to matching the printer manages color prints. So, let's clear up that misconception right now. Ctein was editing an image in aRGB color space, but looking at his edits on a monitor generically factory calibrated to mimic a native sRGB-like gamut.  As such, should it surprise us that Ctein believes soft proofing is not very accurate? Printer manages color did not render a perfect reproduction of the source file, perfectly displayed on his monitor. It was just a method that happened to "play nicely" with this particular image that Ctein printed along with others to test his printer manages color hypothesis.

4). In step 2, I had verified that Xrite ICC default profiles' perceptual and relative rendering tags do in fact exaggerate the color errors of Ctein's Apollo Soyuz source image because Xrite's default profile recipes favor colorfulness at the expense of tonality (i.e. the typical Xrite-built profile favors colorfulness over the preservation of lightness relationships among the various image elements). As such, these profiles were not a good staring point for any subsequent image edits. Because Xrite's default profiling algorithm(s) are so pervasive in the industry, and because Andrew Rodney likely made Ctein a custom profile using said default settingss which Ctein also stated failed to fix his issues, it's fair to say that Ctein does not like ICC profile(s) renderings which favor colorfulness over tonal accuracy.  In my own work, I switch between Logo colorful and logo classic frequently, so I'm sympathetic to anyone who finds precise color accuracy lacking in many typical ICC profiles. Xrite and other profile making vendors really need to emphasize these aspects of "color mapping" more in their literature than is typically found.

5. Once I understood what Ctein's ICC profile problem was, namely that one needs to start with a profile that favors the preservation of tonal relationships at the expense of colorfulness. I reached into my "bag of ICC profile tricks' and applied Xrite's original PM5 "logo classic" to soft proof the Apollo Soyuz image on my calibrated high gamut NEC spectraview II display. The softproof showed me that logo classic was providing a much better starting point for final image edits needed to match or exceed Ctein's printer managed color output because log classic favors tonal (lightness) accuracy at the expense of hue and chroma.

6). I applied just two image edits in my carefully calibrated softproofing mode, one hue/sat layer in PS, and one curve correction. Using a sound softproofing workflow as my guide, I was able to make my corrections in a matter of minutes while working on Ctein's source file image.(note: while the edits were few, they were indeed tricky moves on this file, ones that would be very hard pressed to do without accurate soft proofing guiding the effort). When I had completed those edits, I had on screen a soft proofed image I actually thought was superior to the print I made with Ctein's method. Yah, I know, my word against his as to which was a preferable rendition of the source file, but bear in mind, what I liked better, I was seeing on screen, and I was expecting it to print the way I was seeing it on screen. Moreover, with just one more layer, and a couple of trims on the "fill" percentages of the PS layer sliders,  I knew I could get really really close to Ctein's preferred "printer manages color" rendition of this image if I wanted to, so I wasn't sacrificing any flexibility in color and tone reproduction and in being able to deliver just what anyone would expect in print from that source file, IMHO, by using an ICC profiled workflow.

7). The resultant print matched my edits very accurately, and because I liked this "interpretation" of the image better than what I got when printed with the "printer manages color workflow" on my P600 printer, I ended up with an ICC profiled print off this file that I liked better as well. Moral of the story:  ICC profile managed workflows are not inferior to printer manages color as Ctein seems to assert, but you do have to understand ICC profiling strengths and weaknesses very well.  Indeed, when done properly, soft proofing with the right ICC profile rendering recipe gets you where you want to go and also tells you very closely what to expect in the final print. My faith in my ICC profiling workflow and all those years I have spent mastering it was restored after completing this little exercise :).

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

Mark,

From experience I know that profile creators are not made equal, reading on the subject I have learned that is the case for more profile creators I do not have access to. I have always assumed that there should be a (virtual) ideal (say neutral) profile creator that does the description of the printer's gamut best. Mapping color without bias. Tweaks to overcome the flaws in the color model and more of that allowed but aiming in the end to a correct mapping of colors. When profile creators are discussed the shortcomings are mentioned and the results compared but I can not recall a discussion on say color tastes observed in the results and by that sketching the character of a profile creator as giving pleasing color etc. It is not something advertised as such by the manufacturers either. Applications have CM rendering choices, BPC etc. that are not equal to one another due to the absence of rules, developer's different interpretations/tastes and used algorithms. Not ideal but we tend to stick to one application and learn its character. Naive probably but when it has to pleasing it should be made pleasing in the image editor and then transferred as pleasing through the CM conversions in a correct way. Pleasing profiles represented by a corresponding pleasing softproof could work too but is that ideal if we have to pull the pleasing sliders backwards in the image editing if the print has to be less pleasing and just correct?

Where Ctein's printer CM route had a favorable flaw for his workflow/equipment, the application cm with or without custom profiles is not free from manufacturer's taste that can also work along or against a workflow or may work for certain images and not for other images. I still think it is a more sound method than the one proposed by Ctein but we should know the possible deviations to rich color or toned down, etc.

Would it be possible to describe profile creators' characters in a list and see where they stand to one another? To quote your words "Xrite and other profile making vendors really need to emphasize these aspects of "color mapping" more in their literature than is typically found". Enough color experts here that could make a list like that, I would not mind if there are outdated creators added as I understand new was not always better. What is the influence on the profiles of each creator if they start from measurements with different spectrometers + colorimeters, UV enabled and UV cut? The influence of software compensations for UV cut instruments and OBA papers measured? Where does ArgyllCMS stand given the fact that it supports almost all spectro- and colorimeters characteristics and aims at one and the same result?

Met vriendelijke groet, Ernst

http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
December 2014 update, 700+ inkjet media white spectral plots
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20646
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Ctein's follow-up article on printer-managed color
« Reply #33 on: October 05, 2015, 10:47:18 am »

Profiles, with any such Rendering Intent, or some other process like Printer Manages Color know nothing about color in context. Only humans can. Anyone, including Ctein who believes that one size fits all, that all conversions are prefect and you're done, that you do not need to conduct post conversion editing to produce the print rendering you desire is very confused about color management and printing digital images! Some images will print as expected with the selection of the best RI, some will need post conversion editing. I tried to explain this to Ctein in multiple emails, I'm not certain it sunk in:

In terms of all color management, and current color science not based color appearance, all the CMS knows about is a single colored pixel. Color management doesn't know anything about color in context! That's why we need to view millions of solid pixels such they appear like an image. And the reason why LOOKING at something (subjective analysis of pleasing color) is more valid than measuring it (Colorimetry which does provides a metric of accuracy) is because measurement is about comparing solid colors. Color appearance is about evaluating images and color in context which measurement devices cannot do.


Ctein thought (might still believe) those of us with expensive profiling products are using all kinds of sliders and controls to 'fix' the issue he saw with his few image examples and that's simply not the case. But humans, including Ctein can and should edit their output converted images to produce a print rendering the desire, ideally one that matches the soft proof! Something Ctein appears to be struggling with as one major example of a workflow that needs optimization IMHO.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Ernst Dinkla

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4005
Re: Ctein's follow-up article on printer-managed color
« Reply #34 on: October 05, 2015, 04:30:15 pm »

Andrew,

I did not write about image context, I see in Mark's post that Logo Classic and Colorful set a different flavor on the rendering, how subtle that difference may be. And where Mark writes that X-rite should inform the user I understand it is not a flaw but a deliberate choice. X-rite calls Colorful a more appealing perceptual rendering and sees it as an improvement, only keeps the old choices aboard for compatibility with the past (PM discontinued anyway). I see the choices get another purpose now. I am interested in how widespread those flavors are in other profile creators too. I do not argue that no image editing should be needed in application color management, I am well aware that a print can be improved in image editing, to the image context and even to the limitation of the print gamut if the rendering choices are not totally satisfying. An accurate softproof should help then.

Met vriendelijke groet, Ernst

http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
December 2014 update, 700+ inkjet media white spectral plots
Logged
Pages: 1 [2]   Go Up