Pages: 1 [2] 3 4 5   Go Down

Author Topic: No color management  (Read 16264 times)

digitaldog

  • Sr. Member
  • ****
  • Online Online
  • Posts: 20645
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: No color management
« Reply #20 on: February 11, 2018, 05:07:03 pm »

And to confirm, the printer does not respond to gamma curve information nor LAB number triplets? Only RGB numbers?
For this discussion, soft of. Just examine the measurement data. It's Lab.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

texshooter

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 575
Re: No color management
« Reply #21 on: February 11, 2018, 05:17:52 pm »


The fog is starting to lift.  I had presumed that LAB values were profile-dependent and, thus, cannot exist outside of (or in the absence of) the profile. I will conduct the experiment I discussed in reply #7 to bring it all home.
« Last Edit: February 11, 2018, 06:26:45 pm by texshooter »
Logged

GWGill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 608
  • Author of ArgyllCMS & ArgyllPRO ColorMeter
    • ArgyllCMS
Re: No color management
« Reply #22 on: February 11, 2018, 06:50:21 pm »

I had presumed that LAB values were profile-dependent and, thus, cannot exist outside of (or in the absence of) the profile.
Quite the opposite. LAB are device independent. They correspond to how we see color.

The printer responds to device dependent values like RGB, CMYK etc.

So with a test chart we sent RGB values to the printer, and when we measure them with a color instrument we get the corresponding LAB values. We now have the basic data to make a Device Profile.
Logged

Mark D Segal

  • Contributor
  • Sr. Member
  • *
  • Offline Offline
  • Posts: 12512
    • http://www.markdsegal.com
Re: No color management
« Reply #23 on: February 11, 2018, 07:17:46 pm »

The fog is starting to lift.  I had presumed that LAB values were profile-dependent and, thus, cannot exist outside of (or in the absence of) the profile. I will conduct the experiment I discussed in reply #7 to bring it all home.

What exactly are you trying to "bring home" and how will what you described in Reply 7 do that? Grateful if you could explain, because I'm not "getting it".
Logged
Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....."

Stephen Ray

  • Full Member
  • ***
  • Offline Offline
  • Posts: 218
Re: No color management
« Reply #24 on: February 11, 2018, 09:41:37 pm »

As an example, there is about a 20 point difference in L* printing (128,128,128) on my Epson and Canon.

Meaning their calibration is not the same until the ensuing ICC output profile(s) actually corrects the machines to match one another, which is essentially using output profiles as a calibration process.

Agree?

Logged

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: No color management
« Reply #25 on: February 11, 2018, 10:02:02 pm »

Meaning their calibration is not the same until the ensuing ICC output profile(s) actually corrects the machines to match one another, which is essentially using output profiles as a calibration process.

Agree?

Exactly.  THe RAW rgb values are used to map to the entire printable gamut. This varies a lot with printers and so the RAW rgb values don't conform to any standard colorspace. They are sui generis, or unique to each printer and paper combination. By printing a target with RAW rgb values the printer can print whatever it is capable of. This is then used to create a printer profile.  Then the various working spaces are first converted to Lab. Those Lab values go through a lookup process using the profile to determine the RAW rgb colors that produce the requisite colors the printer is capable of.
Logged

Stephen Ray

  • Full Member
  • ***
  • Offline Offline
  • Posts: 218
Re: No color management
« Reply #26 on: February 11, 2018, 10:29:39 pm »

... a target with RAW rgb values...

So, back to the OP's numbers of 128, 128, 128;  a "middle" gray in the three common RGB spaces, untagged, because they are in fact in the "middle," should print the same with all profiles off?
Logged

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: No color management
« Reply #27 on: February 11, 2018, 10:49:47 pm »

So, back to the OP's numbers of 128, 128, 128;  a "middle" gray in the three common RGB spaces, untagged, because they are in fact in the "middle," should print the same with all profiles off?
They will print the same color when printed without color management if the printer settings and paper are unchanged. It will typically be close to a neutral color. Profiles are not used when printing w/o color management. Typically, if the RGB values are the same (128,128,128) one can expect the printed color to be close to neutral but it often isn't. When printing a neutral L*=50 with color management the actual RGB values needed to produce a neutral color are used as the profile will determine the correct RGB values for the particular paper in use.
Logged

Stephen Ray

  • Full Member
  • ***
  • Offline Offline
  • Posts: 218
Re: No color management
« Reply #28 on: February 11, 2018, 11:10:45 pm »

They will print the same color when printed without color management

I hope this info brings it home for the OP from the perspective of his referenced video.
Logged

texshooter

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 575
Re: No color management
« Reply #29 on: February 12, 2018, 12:12:32 am »

I hope this info brings it home for the OP from the perspective of his referenced video.

I'm hoping the test results will show that the printer responds to the LAB triplet numbers (per digitaldog) instead of the RGB triplet numbers (per Doug Gray).  I'm hoping my printer can tell the difference between these two targets, because my display surely can.

Logged

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: No color management
« Reply #30 on: February 12, 2018, 01:49:08 am »

I'm hoping the test results will show that the printer responds to the LAB triplet numbers (per digitaldog) instead of the RGB triplet numbers (per Doug Gray).  I'm hoping my printer can tell the difference between these two targets, because my display surely can.
Printing, with color management disabled, only takes RGB values, not Lab. Lab values do not go from 0-255.

Further, Lab IS a defined color space. The RGB triplets used to print profile targets are not in a defined colorspace. They are interpreted directly by the printer driver in such a way as to map to the range of colors a printer/paper combo can print. They have little to no relationship with any defined colorspace such as Lab, sRGB, ProPhoto, etc.
Logged

kirkt

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 604
Re: No color management
« Reply #31 on: February 12, 2018, 08:16:33 am »

Pretend you have a printer with a keypad attached to it - via the keypad you can enter an RGB triplet directly in the printer's native color space and the printer will print a color patch as a result.  So, you enter a triplet (say 128, 128, 128) and it prints a patch that is sort of middle gray.  Then you measure that patch with a spectrophotometer and you get a spectral response, characterized as a L*a*b* value.  You compare that result to what you want, or know, that RGB triplet to represent: say the measured result is L*55, a*1, b*1 but the desired result is L*50, a*0, b*0.  You can devise a relationship between the printer's NATIVE response (the patch it printed in response to your input) and the desired response so that when the printer receives an RGB triplet of 128, 128, 128 from the keypad, it will print a patch that measures L*50, a*0, b*0 instead of its native L*55, a*1, b*1 - that relationship is the ICC profile.  Say you could load your mapping relationship (profile) on a hypothetical SD card and plug that card into the hypothetical keypad - then, when you enter 128, 128, 128 on the keypad, the keypad would read the profile from the card, beep boop bop and, via the profile, convert that input triplet in the printer's native space (128, 128, 128) into a modified triplet (say, 123, 124, 125) so that it would result in a printed patch of L*50, a*0, b*0 instead of the native L*55, a*1, b*1.

In color management, the input color space (the RGB triplet) is converted to a device-independent value first (the PCS), so that an RGB triplet in one input space that represents the same color in another space (but with different RGB values) is sent to the printer in its native color space as the same native value.

This is an oversimplification, but the point is, to establish the relationship between the native output and the desired output, you need to know the native output.  This is what you get when you print a target with no color management. 

kirk
« Last Edit: February 12, 2018, 08:48:01 am by kirkt »
Logged

Mark D Segal

  • Contributor
  • Sr. Member
  • *
  • Offline Offline
  • Posts: 12512
    • http://www.markdsegal.com
Re: No color management
« Reply #32 on: February 12, 2018, 10:05:13 am »

Nice way of putting it.
Logged
Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....."

nirpat89

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 661
    • Photography by Niranjan Patel
Re: No color management
« Reply #33 on: February 12, 2018, 11:37:17 am »

Love the sound effects!
Logged

arobinson7547

  • Full Member
  • ***
  • Offline Offline
  • Posts: 187
Re: No color management
« Reply #34 on: February 13, 2018, 04:50:34 pm »

I did not read the whole thread [yet]

but a long time ago the story was told, "The Color of Toast".

See if this helps your conversation, at all.
Logged

texshooter

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 575
Re: No color management
« Reply #35 on: February 13, 2018, 05:14:55 pm »

The results are in.

I created two step-wedge test targets from Photoshop, one in the ProphotoRGB space and the other in the AdobeRGB space, and then printed each one with color management turned off. Then I measured the zone V middle gray patch with the ColorMunki.

Target #1 =  ProphotoRGB color space.  Notice that Zone V has RGB = 128,128,128 and *L = 61.  And notice how it is visually brighter than the other. This is to be expected because ProPhotoRGB has a 1.8 gamma curve.

prophoto by texshooter, on Flickr


Target #2 = AdobeRGB color space.  Notice that zone V has RGB = 128,128,128 and *L = 54.  And notice how it is visually darker than the other. This is to be expected because AdobeRGB has a 2.2 gamma curve.

aRGB by texshooter, on Flickr

So the question is:  How does the printer respond to each of these targets? Does the printer read (respond to) the RGB triplets or the LAB triplets? Will these two target print differently or will they print the same?

The results:  They print the same.  The ProphotoRGB target printed at *L = 32.6 and the AdobeRGB target printed at *L = 32.3, which I will treat as close enough to be the same.

So Doug Gray is right. However, I'm still left confused about one thing. (And by the way, I appreciate all the refreshers on how color management works in general, but my question is more granular than that and is focused to just one step of the process, namely the step where the calibration system tells the printer how the user (me) wants middle gray to appear. Everything else I get.)

If the printer (with CM Off) renders middle gray the same way regardless of the color space (and gamma curve) the target was created in (and intended for), then how does the printer calibration software (eg, ColorMunki) tell the printer how the photographer (me) wants middle gray to appear as? When I use the ColorMunki to profile my printer/paper combo, I don't have a choice of what test target gets printed or whether middle gray should be *L=54 instead of *L=61). The ColorMunki does that automatically. It prints a chevron-shaped strip of patches from its own mystery internal library.   I have no idea how the ColorMunki software is instructing the printer to interpret middle gray. Because I have no control over this, I'm left in the dark and don't know what's going on.  All I know is this:

If I'm working in the AdobeRGB color space in photoshop, I want middle gray to print as *L=54 because that is how it appears on the display. And when I am working in the ProphotoRGB color space, I want middle gray to print as *L=61 because that is how it appears on my display.  The ColorMunki program never asks me which target I wold like to print. It only prints what it wants to print, a one-size-fits-all target, which means that the printer is left thinking that middle gray only has one possible meaning. And that is not true.  I don't understand why it doesn't matter what the gamma curve of the test target is when generating an ICC profile. I would think it matters a great deal. The printer needs to know that "middle gray" does not have only one definition. But that's what the ColorMunki software is telling the printer. Doesn't seem right to me.

Sidebar question:

I know that every printer is imperfect and will tend to print either too dark or too light. That's why we calibrate them.  But in my experiment the zone V middle gray patch printed way darker than it should have. I was expecting a rendering of *L=50 to 60, but got *L= 32.   Is that normal?


« Last Edit: February 13, 2018, 05:56:54 pm by texshooter »
Logged

Stephen Ray

  • Full Member
  • ***
  • Offline Offline
  • Posts: 218
Re: No color management
« Reply #36 on: February 13, 2018, 05:53:41 pm »

However, I'm still left confused about one thing.

So, what happens when you make prints with PROFILES ON? Do you see the results you're looking for?

PS:  Yes, your printer is way out of calibration.
Logged

GWGill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 608
  • Author of ArgyllCMS & ArgyllPRO ColorMeter
    • ArgyllCMS
Re: No color management
« Reply #37 on: February 13, 2018, 06:35:18 pm »

The results:  They print the same.  The ProphotoRGB target printed at *L = 32.6 and the AdobeRGB target printed at *L = 32.3, which I will treat as close enough to be the same.
If they don't, then you haven't turned Color Management off.
Quote
If the printer (with CM Off) renders middle gray the same way regardless of the color space (and gamma curve) the target was created in (and intended for), then how does the printer calibration software (eg, ColorMunki) tell the printer how the photographer (me) wants middle gray to appear as?
It doesn't, because Color Management doesn't work that way. Calibration might, but you weren't talking about Calibration up to now.
Quote
I have no idea how the ColorMunki software is instructing the printer to interpret middle gray. Because I have no control over this, I'm left in the dark and don't know what's going on.
Obviously not.

Color Management

Summary - profiling doesn't change the printer. What it does is give the computer system & applications information about how the printer behaves. The applications then use that to transform images from the colorspace they are tagged with into the printers native colorspace. i.e. the printer always behaves the same - it isn't doing the Color Management.

Quote
I know that every printer is imperfect and will tend to print either too dark or too light. That's why we calibrate them.
So now you are talking about Calibration rather than Profiling. They are quite different things.

Calibration vs. Characterization
Logged

texshooter

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 575
Re: No color management
« Reply #38 on: February 13, 2018, 06:48:40 pm »


So now you are talking about Calibration rather than Profiling. They are quite different things.

Calibration vs. Characterization

My ColorMunki only came with one button, not two, so I'd be interested to know what device you use to calibrate with and what device you use to characterize with.
Logged

Stephen Ray

  • Full Member
  • ***
  • Offline Offline
  • Posts: 218
Re: No color management
« Reply #39 on: February 13, 2018, 08:42:27 pm »

My ColorMunki only came with one button, not two, so I'd be interested to know what device you use to calibrate with and what device you use to characterize with.

I'm not sure if you've mentioned what printer you are using but I believe you don't need to venture there. Normally, to control the "black box" brains of a professional printer, investment in a third-party RIP is required. If / when you properly create your ICC output profile, it will correct the color values to your liking to the best ability of your system. However, at least you know your printer is "dark" using this particular printer setting, media, and ink at this time. 

Professional printers are linearized and calibrated to a common standard (in factory) before they leave the factory.

Maybe Mark can explain your dark un-profiled print, but I think in the history of the world, no printer mfr who also sells ink has ever calibrated a printer to make a light print. ;-)

Logged
Pages: 1 [2] 3 4 5   Go Up