RetouchPRO

Go Back   RetouchPRO > Technique > Photo Retouching
Register Blogs FAQ Site Nav Search Today's Posts Mark Forums Read


Photo Retouching "Improving" photos, post-production, correction, etc.

ICC false profiles for color grading/correction

Reply
 
Thread Tools
  #61  
Old 02-17-2017, 10:46 PM
John Wheeler's Avatar
John Wheeler John Wheeler is offline
Senior Member
 
Join Date: May 2011
Location: Colorado
Posts: 580
Re: ICC false profiles for color grading/correctio

I am late to the game yet thought it would be of interest to see Dan Margulis changed opinion in 2008 vs 2001 re false profiles:

Quote:
Re: false profiles
Posted by: Dan Margulis
Sun Dec 7, 2008 5:59 am (PST)

George Machen writes,

Consequently, I conclude that the Exposure
dialog's Gamma setting may be a useful
alternative to lightening or darkening with a
false profile because:
- It's "close enough," which is all that's needed
with, for example, the broad strokes employed by
the Picture Postcard workflow.
- It displays a dynamically changing preview
while moving the Gamma slider, all the better to
visually gauge how much to apply in real-time.

That's correct. Within reason, messing around with the gamma setting in Exposure is the same as applying a false profile followed by Convert to Profile>Current RGB workspace.

For the past year or so, I've been using Exposure rather than false profile in my lectures to basic ACT classes and at Photoshop World. My reasoning is that when time is limited I don't want to waste any of it by explaining what a false profile is and does.

For people who *do* know what false profiles are and have some premade, I don't think it's a big deal either way, but I still prefer false profiles in some but not all cases:

*The usual purpose of either move is to lighten the image prior to going into LAB. If so, the Exposure method involves an extra conversion of data. I am not unduly buffaloed by this, but others might be.

*If, OTOH, the file is not going to LAB but will remain in RGB, I prefer Exposure on the grounds that I might be stupid enough to forget later in the process that the file is being seen through a false profile.

*Applying a premade false profile is marginally quicker than using Exposure.

Basically, however, it's not a big deal either way. The question that may be of interest is under what circumstances would it pay to use Exposure on a layer set to Luminosity mode. This normally produces a more colorful result than either straight Exposure or a false profile. Most of the time, that's not desirable because the file is going into LAB where everything is going to be intensified beyond recognition anyway. I can visualize instances, however, where it might be helpful. But I don't have any at hand just yet.

Dan Margulis
And just for reference, I personally never ever ever use false profiles. I have found other workflows to be much more productive, accurate, explainable, and more predictable in their results. Not surprising since are talking about a 15 year old article about techniques from at least 6 years earlier than the article.

As far as data loss, if you are already in 16 bit to begin with it is more of a moot point. If you are in 8 bit with wider gamut colorspaces, you have already posterized your color data since it used a much smaller portion of the 0-255 RGB color data range.

Sorry for jumping back to the original topic yet hope the info was of some value. Just my opinion of course.
Reply With Quote top
  #62  
Old 02-17-2017, 11:13 PM
klev klev is offline
Senior Member
 
Join Date: Jun 2010
Posts: 1,109
Re: ICC false profiles for color grading/correctio

Quote:
Originally Posted by marameo View Post
I see. I think the only way would be to apply that lut on a look up layer adjustment on this color checker image (Adobe RGB) and return it so that I can read the lab values and see what it's all about..

Thanks
I don't know if a color checker will be sufficient, but I ran it. Here's the result. I also think kodachrome looks neat.
Reply With Quote top
  #63  
Old 02-18-2017, 08:32 AM
marameo's Avatar
marameo marameo is offline
Senior Member
 
Join Date: Oct 2014
Location: The Eternal City
Posts: 155
Re: ICC false profiles for color grading/correctio

Quote:
Originally Posted by klev View Post
I don't know if a color checker will be sufficient, but I ran it.
Thanks!

I think a ColorChecker Digital SG won't sufficient neither. Anyway, I managed to generate an abstract profile (download here) by using ArgyllCMS. The idea is to compare LAB values of test chart before/after color grading.

I believe this is how the Kodak ColorFlow ICC plug-in worked. Yet, instead of a 24-patch color chart more paches were used to provide higher depth (something similar to hald clut in a way).
Attached Images
File Type: png comparing.PNG (31.6 KB, 6 views)
Reply With Quote top
  #64  
Old 02-18-2017, 09:46 AM
klev klev is offline
Senior Member
 
Join Date: Jun 2010
Posts: 1,109
Re: ICC false profiles for color grading/correctio

Quote:
Originally Posted by marameo View Post
I believe this is how the Kodak ColorFlow ICC plug-in worked. Yet, instead of a 24-patch color chart more paches were used to provide higher depth (something similar to hald clut in a way).
That's entirely possible. There's also a 300 or so patch version, but I'm still not sure I would trust that. These things appear different under different lighting. I would imagine that it takes a considerable amount of work to get one close to a desirable state. I don't have the link handy, but VSCO discussed how they did theirs at one point. They were processing older film types and scanning the results. I would imagine that it required both technical references like color patches and images shot on those film types under different lighting conditions. Otherwise it would be difficult to get it right.

Oh and more patches doesn't really mean higher bit depth. The bits basically come from the types. If you have more patches, you are likely to have less interpolation, but if you have too many, it may overfit. That's because the chart is just a sample of what can be reproduced, not a snapshot of the entire space. I'm assuming that it will try to be close to each patch. Aside from that, I like Argyll. It's a decent CMS.
Reply With Quote top
  #65  
Old 02-18-2017, 01:08 PM
marameo's Avatar
marameo marameo is offline
Senior Member
 
Join Date: Oct 2014
Location: The Eternal City
Posts: 155
Re: ICC false profiles for color grading/correctio

Quote:
Originally Posted by klev View Post
Oh and more patches doesn't really mean higher bit depth. The bits basically come from the types. If you have more patches, you are likely to have less interpolation, but if you have too many, it may overfit.
This is a Hald CLUT Identity image (untagged rgb). Zoom in at 3200% to see each pixel going from 0,0,0 to 255,255,255 and that is a ~16-million-patch chart.

http://www.quelsolaar.com/technology/clut.html

«A Hald CLUT is an image that has a specific color pattern on it. In this pattern all colors in the color space are represented. An application that uses the HALD CLUT image to color correct an image, takes a source image color and looks it up in the color pattern of the HALD CLUT, and the color it finds in that place is the corrected color that should replace the source color in the destination image. If the color doesn't exist in the CLUT, one can look up several colors and interpolate between them.»

I wonder if there's a way to convert a target like this (tiff/png) to a CGATS text file..
Reply With Quote top
  #66  
Old 02-18-2017, 05:04 PM
klev klev is offline
Senior Member
 
Join Date: Jun 2010
Posts: 1,109
Re: ICC false profiles for color grading/correctio

Quote:
Originally Posted by marameo View Post
I wonder if there's a way to convert a target like this (tiff/png) to a CGATS text file..
I can guess what that thing does. It may be trying to account for angle or other variations. They might also take some kind of weighted average over the patches. As I mentioned before though, it's dependent on the spectrum of the light illuminating the target. These targets are generally pretty well tuned for metamerism though, or the appropriate illuminant source for a match is specified.

You can technically convert an image to an ascii text file. You need something that can load a tiff or png. This would be my suggestion if you're comfortable writing a few lines of python. You wouldn't need to display the image. You load it into a numpy array. You then have python open a text file and write to it. You would need the appropriate specification for CGATs, which I assume would be under developer documentation somewhere. Those things aren't always intuitive.

If you're on a Windows machine, the easiest way to get an appropriate python installation would be via anaconda, but it's a pretty large download if installed that way. I think it's a few GB. If you're on a mac, you could always use homebrew, again assuming you are comfortable trying this. It took me an embarrassing amount of time to become comfortable with a lot of this stuff.

It's also possible that these things do not involve a simple straightforward translation. I guess in spite of what I wrote above, I would first check whether there's a tool that takes in an image, either as a file or a set of pixel arrays, and outputs the appropriate values for a CGAT. What I wrote before assumes a straightforward translation, like if you wanted to just output those number in order to a text file rather than an image file.
Reply With Quote top
  #67  
Old 02-18-2017, 05:58 PM
andrewrodney's Avatar
andrewrodney andrewrodney is offline
Senior Member
 
Join Date: May 2010
Location: Santa Fe
Posts: 1,079
Re: ICC false profiles for color grading/correctio

Quote:
Originally Posted by marameo View Post
I wonder if there's a way to convert a target like this (tiff/png) to a CGATS text file..
ColorThink Pro can extract the colors into a ColorList (Lab values with reference values too) then this can be accomplished. Not sure why but doable.
Reply With Quote top
  #68  
Old 02-19-2017, 05:03 AM
marameo's Avatar
marameo marameo is offline
Senior Member
 
Join Date: Oct 2014
Location: The Eternal City
Posts: 155
Re: ICC false profiles for color grading/correctio

I had to play a bit more with ColorLab to figure out that it can actually convert a TIFF LAB target (1 pixel = 1 patch) to a CGATS file with XYZ/LAB values!

(Filter -- Layout and Format -- Testchart Define -- Eye one)

Now, ArgyllCMS can be used to generate an abstract profile by comparing XYZ/LAB measurement values (CGATS) from two test charts (original -> color corrected).
Attached Images
File Type: png chart_format3.PNG (76.6 KB, 8 views)
File Type: png chart_format2.PNG (28.9 KB, 7 views)
File Type: png chart_format.PNG (23.4 KB, 5 views)
Reply With Quote top
  #69  
Old 02-19-2017, 11:19 AM
andrewrodney's Avatar
andrewrodney andrewrodney is offline
Senior Member
 
Join Date: May 2010
Location: Santa Fe
Posts: 1,079
Re: ICC false profiles for color grading/correctio

Quote:
Basically, however, it's not a big deal either way. The question that may be of interest is under what circumstances would it pay to use Exposure on a layer set to Luminosity mode. This normally produces a more colorful result than either straight Exposure or a false profile. Most of the time, that's not desirable because the file is going into LAB where everything is going to be intensified beyond recognition anyway. I can visualize instances, however, where it might be helpful. But I don't have any at hand just yet.
That bold text doesn't make any sense to me. But that wouldn’t be the first time such prior text and after text seems to confirm and deny the same concept and then produce a result that doesn't make sense from the either. Can anyone explain? The conclusions appear to be based on the very last sentence too; no proof or experience by admission.
Reply With Quote top
  #70  
Old 02-19-2017, 04:19 PM
klev klev is offline
Senior Member
 
Join Date: Jun 2010
Posts: 1,109
Re: ICC false profiles for color grading/correctio

Quote:
Originally Posted by marameo View Post
I had to play a bit more with ColorLab to figure out that it can actually convert a TIFF LAB target (1 pixel = 1 patch) to a CGATS file with XYZ/LAB values!

(Filter -- Layout and Format -- Testchart Define -- Eye one)

Now, ArgyllCMS can be used to generate an abstract profile by comparing XYZ/LAB measurement values (CGATS) from two test charts (original -> color corrected).
I'm glad you found a way to do that. I've never used ColorLab, which is why I would have used scikit-image to extract them. Remember LAB differs from XYZ in the way it models things. Their common features are that they are both device independent and usable as profile connection spaces.
Reply With Quote top
Reply

  RetouchPRO > Technique > Photo Retouching


Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is Off
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
where are icc. profiles stored on mac OS 10.8.3? Caravaggio Photo Retouching 10 05-05-2013 10:42 AM
Custom DNG Profiles MikeyPee Software 0 11-20-2011 02:08 PM
NEC SpectraView and ICC profiles shizam1 Hardware 2 04-05-2011 10:09 AM
Installing Ilford icc profiles on a Mac Ken Fournelle Image Help 2 07-31-2004 11:20 AM
new custom ICC profiles black and white, color christophe Input/Output/Workflow 5 05-26-2004 08:26 PM


All times are GMT -6. The time now is 08:09 PM.


Powered by vBulletin® Version 3.8.4
Copyright ©2000 - 2017, Jelsoft Enterprises Ltd.
SEO by vBSEO ©2011, Crawlability, Inc.
Copyright © 2016 Doug Nelson. All Rights Reserved