Ad
Ad
Ad
Pages: « 1 [2] 3 4 »   Bottom of Page
Print
Author Topic: odd print lab proofing advice  (Read 10117 times)
Mark D Segal
Contributor
Sr. Member
*
Offline Offline

Posts: 6821


WWW
« Reply #20 on: January 07, 2010, 06:15:51 PM »
ReplyReply

Quote from: digitaldog
Rather than sending something bigger or smaller than sRGB, the question is, why not send the data in the output color space of the device using a good profile from whatever original RGB working space you started with?

The discussion/argument isn't that the device is significantly bigger or smaller than sRGB. The point is, this isn't an sRGB output device and the color space you should send is that of the printer, after soft proofing and editing with the profile from that device.

You and I might both have a 30 inch LCD but that doesn't mean I can send you the profile of my display to use and vise versa even if both have very similar sized color gamuts.

Andrew,

If I understand Wayne's description of how these machines are "programmed" to function, there may be a real problem here which defies conventional colour management approaches of the kind you are describing. In order to do what you say should be done, which sounds eminently sensible, it is necessary to be able to profile the device. In order to profile the device you need to be able to characterize its native behaviour with all colour-modification software turned off. Wayne seems to be saying that this is the step which these machines do not allow to be done. So if you can't do that, it would seem that you can't profile it properly, and if you can't profile it properly the rest of the recipe doesn't work. Am I making sense here?

Mark
Logged

Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....." http://www.luminous-landscape.com/reviews/film/scanning_workflows_with_silverfast_8.shtml
jackbingham
Full Member
***
Offline Offline

Posts: 206


WWW
« Reply #21 on: January 07, 2010, 06:18:13 PM »
ReplyReply

This just requires a change of approach on your part. You are trying to do this the right way and the lab has their collective head up their........well you know. I would start soft proofing the way you should, convert the files for print and make them make redo after redo after redo until they begin to get the point. Or if all you want to do is make prints to sell, find another lab and irritate them only part time.
Logged

Jack Bingham
Integrated Color Corp Makers of Coloreyes Display
digitaldog
Sr. Member
****
Offline Offline

Posts: 8628



WWW
« Reply #22 on: January 07, 2010, 06:23:05 PM »
ReplyReply

Quote from: Wayne Fox
However, I believe most labs are offering these profiles as a soft proofing solution only, but the files are still converted to sRGB to be submitted to the printer.

Which is idiotic. You can't know what rendering intent the lab will use, with or without Black Point Compensation or with what CMM. We don't even know if they are using the profile they've supplied for soft proofing to convert the data or if the devices behave as the profile describes the soft proof.

That's my beef with the labs. Its totally marketing based, faith based and brain dead "color management". Either say you don't support color management and force people to send you files in Stupid RGB or do it right. This idea that you can soft proof with some profile and send sRGB to a lab is an attempt to make the slightly more savvy user think the lab has a clue about color management when in fact its a useless "feel good" button.
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
Wayne Fox
Sr. Member
****
Offline Offline

Posts: 2811



WWW
« Reply #23 on: January 07, 2010, 06:24:51 PM »
ReplyReply

Quote from: Czornyj
I did two profiles for the same Kodak Endura VC paper - one was printed on Noritsu (that treats everything as sRGB), second was printed on KIS lab with with Color Management module (that could just print RGB targets without any internal transformations).

The print from Noritsu's gamut volume was only 520.000 dE^3(76), while the print from KIS's gamut volume was 645.000 dE^3(76).

So it seems that the print is potentially loosing about 20% of gamut volume when the machine is locked to sRGB.

Interesting.  Personally I've always believe this, and the fact their gamut is larger certainly implies they are using a more logical approach to color management.  Perhaps soon we can get all silver halide printing out of the dark ages ...
Logged

digitaldog
Sr. Member
****
Offline Offline

Posts: 8628



WWW
« Reply #24 on: January 07, 2010, 06:26:20 PM »
ReplyReply

Quote from: Mark D Segal
In order to profile the device you need to be able to characterize its native behaviour with all colour-modification software turned off.

That's not necessarily so (see example of profiling an Epson using Color Controls). It would be great if the device could be linear and well behaved but that's not a requirement written in stone.

In order to profile a device, one thing is certain. It has to be consistent or you have to have calibration processes in line to keep the device in a state the profile describes.
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
Wayne Fox
Sr. Member
****
Offline Offline

Posts: 2811



WWW
« Reply #25 on: January 07, 2010, 06:30:34 PM »
ReplyReply

Quote from: digitaldog
Which is idiotic. You can't know what rendering intent the lab will use, with or without Black Point Compensation or with what CMM. We don't even know if they are using the profile they've supplied for soft proofing to convert the data or if the devices behave as the profile describes the soft proof.

That's my beef with the labs. Its totally marketing based, faith based and brain dead "color management". Either say you don't support color management and force people to send you files in Stupid RGB or do it right. This idea that you can soft proof with some profile and send sRGB to a lab is an attempt to make the slightly more savvy user think the lab has a clue about color management when in fact its a useless "feel good" button.

Agreed.  Hopefully they are working on a correct approach.

One issue is that 99% of the customers of these large labs are clueless themselves about color management, don't have calibrated monitors and simply want something that gives them pretty good results, which can be had without that much work (at least for the expectations of those customers).  Those that frequent this forum are not typical of their customers.
Logged

Mark D Segal
Contributor
Sr. Member
*
Offline Offline

Posts: 6821


WWW
« Reply #26 on: January 07, 2010, 06:36:03 PM »
ReplyReply

Quote from: digitaldog
That's not necessarily so (see example of profiling an Epson using Color Controls). It would be great if the device could be linear and well behaved but that's not a requirement written in stone.

In order to profile a device, one thing is certain. It has to be consistent or you have to have calibration processes in line to keep the device in a state the profile describes.

OK, so let's explore that. You have this Noritsu printer which predates colour management and it's equipped with some "smart programming" which makes images come out like how the manufacturer thinks people would like to see them. You feed it a target to print patches for creating a profile. It's "brain" analyzes the "image" and comes up with something which it prints. You take that page of patches, measure them, generate a profile and you think you've characterized the printer. Maybe so. But from what Wayne seems to be describing, the question is "have you really"? Does this printer software necessarily respect the "steady state" repeatable behaviour needed for the profile to work properly with real photographs? I don't know, I'm asking. Will the printer's firmware do a completely different KIND of reading of a real photograph, contrasted to how it interpreted the page of patches, because with each different kind of image it gets, it's trying to automatically simulate a consumer-pleasing outcome, so perhaps its internal behaviour is intentionally not predictable from target to photo, or from one photo to another?
Logged

Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....." http://www.luminous-landscape.com/reviews/film/scanning_workflows_with_silverfast_8.shtml
Mark D Segal
Contributor
Sr. Member
*
Offline Offline

Posts: 6821


WWW
« Reply #27 on: January 07, 2010, 06:39:32 PM »
ReplyReply

Quote from: Wayne Fox
Agreed.  Hopefully they are working on a correct approach.

One issue is that 99% of the customers of these large labs are clueless themselves about color management, don't have calibrated monitors and simply want something that gives them pretty good results, which can be had without that much work (at least for the expectations of those customers).  Those that frequent this forum are not typical of their customers.

Yes, this must be true, otherwise they would be responding to another set of consumer requirements. My first stimulus ten years ago to get into a digital printing workflow of my own arose from dissatisfaction with what I was getting back from labs. So that's at least one customer in a sea of millions they lost! (Epson got me instead!)
Logged

Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....." http://www.luminous-landscape.com/reviews/film/scanning_workflows_with_silverfast_8.shtml
digitaldog
Sr. Member
****
Offline Offline

Posts: 8628



WWW
« Reply #28 on: January 07, 2010, 06:42:07 PM »
ReplyReply

Quote from: Mark D Segal
OK, so let's explore that. You have this Noritsu printer which predates colour management and it's equipped with some "smart programming" which makes images come out like how the manufacturer thinks people would like to see them. You feed it a target to print patches for creating a profile. It's "brain" analyzes the "image" and comes up with something which it prints.

You're assuming this is some kind of auto fix, alter numbers without a way to turn it off device. I know of no such device, at least one that can't simply send the data to the output device without mucking around with it. I know some people said this was the case with the Fuji Frontier but Fuji has (had?) a front end that fully allowed data to be sent through the device with the use of profiles and without alteration of the data.

And we're told "send sRGB" so we are now to assume that this RGB data is somehow magically altered based on what? So I send the same RGB data (numbers) to the device and every time it produces differing results? That would make reprints a bitch.

Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
Czornyj
Sr. Member
****
Online Online

Posts: 1371



WWW
« Reply #29 on: January 07, 2010, 06:47:21 PM »
ReplyReply

Quote from: Mark D Segal
OK, so let's explore that. You have this Noritsu printer which predates colour management and it's equipped with some "smart programming" which makes images come out like how the manufacturer thinks people would like to see them. You feed it a target to print patches for creating a profile. It's "brain" analyzes the "image" and comes up with something which it prints. You take that page of patches, measure them, generate a profile and you think you've characterized the printer. Maybe so. But from what Wayne seems to be describing, the question is "have you really"? Does this printer software necessarily respect the "steady state" repeatable behaviour needed for the profile to work properly with real photographs? I don't know, I'm asking. Will the printer's firmware do a completely different KIND of reading of a real photograph, contrasted to how it interpreted the page of patches, because with each different kind of image it gets, it's trying to automatically simulate a consumer-pleasing outcome, so perhaps its internal behaviour is intentionally not predictable from target to photo, or from one photo to another?

From my observations, Noritsu just makes some kind of simple, internal covertion from sRGB to it's own color space. It gives tolerable and repeatable results - as long as the image is rendered to sRGB, of course.
« Last Edit: January 07, 2010, 06:47:51 PM by Czornyj » Logged

Mark D Segal
Contributor
Sr. Member
*
Offline Offline

Posts: 6821


WWW
« Reply #30 on: January 07, 2010, 06:48:21 PM »
ReplyReply

Quote from: digitaldog
You're assuming this is some kind of auto fix, alter numbers without a way to turn it off device. I know of no such device, at least one that can't simply send the data to the output device without mucking around with it. I know some people said this was the case with the Fuji Frontier but Fuji has (had?) a front end that fully allowed data to be sent through the device with the use of profiles and without alteration of the data.

And we're told "send sRGB" so we are now to assume that this RGB data is somehow magically altered based on what? So I send the same RGB data (numbers) to the device and every time it produces differing results? That would make reprints a bitch.

No, not quite what I'm saying. I would expect it to treat multiple passes of the same numbers identically. What I'm wondering about is whether it behaves in a "proportionately systematic manner" (if I can put it that way) when different sets of numbers of are sent to it. Is it possible that the machines are programmed to produce outcomes which respect certain output parameters regardless of the differing nature of the information sent into them, so that they "bend" the data differently depending on the relationship between a varying set of inputs and an output with a comparatively fixed set of constraints.
Logged

Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....." http://www.luminous-landscape.com/reviews/film/scanning_workflows_with_silverfast_8.shtml
digitaldog
Sr. Member
****
Offline Offline

Posts: 8628



WWW
« Reply #31 on: January 07, 2010, 06:53:52 PM »
ReplyReply

Quote from: Czornyj
From my observations, Noritsu just makes some kind of simple, internal covertion from sRGB to it's own color space. It gives tolerable and repeatable results - as long as the image is rendered to sRGB, of course.

That syncs up with my understanding of the process. The front end assumes sRGB and converts to the native color space (it has to, the native color space simply isn't sRGB as you've illustrated).
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
Schewe
Sr. Member
****
Offline Offline

Posts: 5422


WWW
« Reply #32 on: January 07, 2010, 10:05:26 PM »
ReplyReply

Quote from: Wayne Fox
(I might add I have my point person contacting Dry Creek Color to try and understand their process.  I'm hopeful things have progressed, but I"m still a little skeptical based on all the frustrations I've had with this over the past 10 years or so).


I don't think you have to go too far to contact the guy from Dry Creek since he's a regular poster here on LuLa (not often but often enough). Try pinging Ethan Hansen...he's a good guy!

And I'm pretty darn sure there ARE ways of sending stuff through relatively recent Noritsu and Fuji printers where the front end does NOT step on the color. It DOES require an active effort by the operator and it's not quite as productive in terms of prints/hour–which is one of the reasons I think lab owners try to discourage it. But maybe that's just my suspicious nature...

:~)
Logged
Czornyj
Sr. Member
****
Online Online

Posts: 1371



WWW
« Reply #33 on: January 08, 2010, 02:23:08 AM »
ReplyReply

Quote from: Mark D Segal
No, not quite what I'm saying. I would expect it to treat multiple passes of the same numbers identically. What I'm wondering about is whether it behaves in a "proportionately systematic manner" (if I can put it that way) when different sets of numbers of are sent to it. Is it possible that the machines are programmed to produce outcomes which respect certain output parameters regardless of the differing nature of the information sent into them, so that they "bend" the data differently depending on the relationship between a varying set of inputs and an output with a comparatively fixed set of constraints.

It behaves in a systematic manner. To illustrate this behavior, here's the profile I did:

The softproofed image was in sRGB. As you can see, the difference between converted and unconverted image is not that large - the machine renders the colors in slightly different way and there's yellow cast, but it definetly looks like it makes some transform from sRGB to its native color space. The problem is that you can't switch it off...
« Last Edit: January 08, 2010, 02:46:06 AM by Czornyj » Logged

digitaldog
Sr. Member
****
Offline Offline

Posts: 8628



WWW
« Reply #34 on: January 08, 2010, 06:54:31 AM »
ReplyReply

If you have something like ColorThink, plot the gamuts in wire frame and pay attention to the Lstar axis down around Lstar 0 to oh, about Lstar 10-15. I can post a shot when I get a chance (off to the airport). In the Noritsu_2006 profile I have, there's a difference one can attribute to the dissimilarities in emissive based working space profiles and printer profiles (one using light, the other a colorant to build color so to speak). In addition, at least with this profile, the sRGB gamut in greens and blues is significantly larger. IOW, there's a lot of difference in the two that a soft proof doesn't show.
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
Slobodan Blagojevic
Sr. Member
****
Online Online

Posts: 5536



WWW
« Reply #35 on: January 08, 2010, 12:00:51 PM »
ReplyReply

Quote from: Czornyj
It behaves in a systematic manner. To illustrate this behavior, here's the profile I did:
...
The softproofed image was in sRGB. As you can see, the difference between converted and unconverted image is not that large - the machine renders the colors in slightly different way and there's yellow cast, but it definetly looks like it makes some transform from sRGB to its native color space. The problem is that you can't switch it off...
Pardon my ignorance, but I am confused as to what exactly caused the difference between the left and right image. Would you care to detail your steps?
Logged

Slobodan

Flickr
500px
Wayne Fox
Sr. Member
****
Offline Offline

Posts: 2811



WWW
« Reply #36 on: January 08, 2010, 01:10:32 PM »
ReplyReply

Quote from: digitaldog
And we're told "send sRGB" so we are now to assume that this RGB data is somehow magically altered based on what? So I send the same RGB data (numbers) to the device and every time it produces differing results? That would make reprints a bitch.


Actually this is a problem ... these machines have large amounts of drift.  

The sRGB data is modified through a lookup table that is constantly being modified to compensate for this drift.  In the early days with the Kodak CRT and LED printer, we usually had to tweak the lookup table by hand.  I'm sure these have been improved, but I don't think this LUT is as sophisticated as an actual profile, and at this point that process is assuming an sRGB file coming in.  It could be the very latest hardware has improved and has found someway to do this via profiles, but that is also challenging.  It's hard for a lab to justify spending 100k to 200k to upgrade a printer for this capability, and these printers run for a long time.

We have a test image that is printed at startup of each machine after it is calibrated daily, which is sent to our central lab for monitoring.  The drift is apparent and the variation from machine to machine is obvious.  When we see one that is too far off we send in the tech. Of course, this drift is nothing new, just goes with the territory of the chemical process, and it usually isn't enough to reject the prints.  But it can be a problem if customers compare reprints side by side.

The printer is trying to reproduce the same color and all of this was designed because of the limitations of the silver halide process.  It works pretty good, but I've always wondered if because of the large amount of drifting if a color managed workflow as we know it would work, since it would almost require new profiles to be made daily.  A few years ago I spent a great deal of time trying to come up with a way to standardize color across all output devices, as we wanted a solution that would allow customer to pick up prints at other locations.  Because all of the images are portraits it is very challenging because the expectation is higher than consumer snapshots being produced at places like Costco and Walgreens.

I finally gave up since I sold the company, but never was very successful.  As I mentioned, I'm now researching it pretty extensively because of some new projects I'm working on.  Hopefully there has been some progress, but even if it has it will take a while to trickle down so it becomes as standard as current inkjet workflows.
Logged

Scott Martin
Sr. Member
****
Offline Offline

Posts: 1311


WWW
« Reply #37 on: January 08, 2010, 01:21:44 PM »
ReplyReply

I just started a new thread that hopefully addresses a number of the issues brought up here about color managing silver halide processes:

http://luminous-landscape.com/forum/index....showtopic=40713

Color Managing silver halide workflows can be done very well but there are some unique challenges. I hope you find it worthwhile reading.
Logged

Wayne Fox
Sr. Member
****
Offline Offline

Posts: 2811



WWW
« Reply #38 on: January 08, 2010, 01:35:27 PM »
ReplyReply

Quote from: Schewe
I don't think you have to go too far to contact the guy from Dry Creek since he's a regular poster here on LuLa (not often but often enough). Try pinging Ethan Hansen...he's a good guy!
Thanks .. I appreciate this.

Quote
And I'm pretty darn sure there ARE ways of sending stuff through relatively recent Noritsu and Fuji printers where the front end does NOT step on the color. It DOES require an active effort by the operator and it's not quite as productive in terms of prints/hour–which is one of the reasons I think lab owners try to discourage it. But maybe that's just my suspicious nature...

:~)

I hope you're right, but machines run for years, are very expensive so used machines are very attractive, and upgrade paths are usually not easy.

As far as lab operators, probably some truth.  If it ain't broke don't fix it.

It's not like the system is horrible, for the most part it works pretty well.  It just doesn't have the fine control we've come to expect from our inkjet workflows, but good results can be had relatively easily.  Just requires a different workflow mentality to which most don't have a problem with.  
Logged

mmurph
Sr. Member
****
Offline Offline

Posts: 507


WWW
« Reply #39 on: January 08, 2010, 06:48:31 PM »
ReplyReply

I haven't read this full thread, so apologies if this is way OT, but ....

WHCC is a "pro" lab in the US that has a pretty strong quality focus.  They require clients to print 5, 8x10 inch prints (provided free of charge) when they open an account to check color accuracy and calibration on the client end, to help ensure customer satisfaction.  

They have pretty decent documentation on their CM process. They also provide profiles for soft proofing that seem pretty acurate and work well for me in in PS. I haven't looked at the size of the color space.

A couple of thoughts:

1) You should be able to use their documentation, profiles, etc. to establish a workflow that should work with other Noritsu devices (of the same model/family of course - I'll let you track down the specifications)    , and

2) They may be able to advise you on the machines capabilities, and help you establish a process that will work with your provider (ie: translate/fill in for what your lab is unable to articulate.)


Here are a couple of snippets from their site:


"Q.  What kind of printers do you use?
A.  Depending on the print sizes requested your order could be printed on one or two different pieces of equipment. All prints 12x18 and smaller are printed on one of our Noritsu printers."


http://www.whcc.com/resources/faq/#photographic prints


"Calibrating your monitor is the first step in receiving the best monitor to print match. The second step is to embed a valid ICC profile into each file you send to WHCC. Without this profile, we do not know what colorspace your files are in. Most photographers use either sRGB or Adobe RGB (1998) as a working colorspace."

http://www.whcc.com/resources/ordering-pro...ion-management/


Hope that helps a bit? I can log in to their site on another machine (with my accoiunt info on it) and quote from their detailed instructions for the 5 quality control "test prints", if that info isn't around on the public site (I don't see it after a quick look ...)

I can happily recommend WHCC, but it seems like print fulfilment might be a bit awkward for you!    

Good luck!

Best,
Michael
Logged
Pages: « 1 [2] 3 4 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad