Need Help With Details Of PS & Color Management Workflow

[bonoboprobo]

Lots of questions under one post, hope it's not too muddled.

So I've been doing a lot of reading on Roger Cavanaugh's, Fred Miranda sites and while I understand the basis for the actions in the work flows, I am not quite clear on some of the details of the use of color spaces and color profiles.

Here is the gross structure of a good workflow as I see it:

-Take picture in .CRW format
-Download and linear convert with BB
-Futz1 in PS to get a good presentable image

At some later date...
-Futz2 further to prep for presentation in a particular format (.jpg for web, print to local inkjet, send to photo printer, etc)

Clearly Futz1 and Futz2 are where the action is, and my questions are mostly about Futz1. There has been enough discussion of actions and scripts that I probably understand what needs to be done where, but it's the details of the color spaces and color management that still don't make sense.

Questions:
Since .CRW files don't come from the camera with and assigned color space or profile what is the most natural space to assign and where? I sort of have the suspicion that many say the answer is Adobe RGB when the TIFF is opened.

But what about Wide Gamut space, where is it appropriate?

Why is it that the camera ICC profile is not applied during or right after the conversion from .CRW? I get this impression from some the discussion of work flows at Rodger and Fred's sites. For instance BB can embed a profie during conversion.

Would embeding a profile during conversion change the image created or just the display of that image?

Why is it that issue of Magenta Overload http://www.rogercavanagh.com/scripts/magenta/pages/magfix01.htm out is not cleaned up by applying the camera's ICC profile? I thought that was the point of a profile, to say that this color from this device is really this other color?

In any case how is it that camera, monitor and eventually printer profiles get stacked on each other in PS? In PS I can see that you can turn on and off seeing the picture in proof colors and select which profile prism to see the picture through, but don't quite get it. BTW my system is NT4.0, which I understand can be made to manage color, but maybe I'm missing some details at the system level. But it would make sense to be able to do it all through PS.

How would I go about using a linear workflow and LinearSharpen if I have a Canon PowerShot S30. Look I know it's not a D30, but I gotta start someplace.

I understand that going from a given RGB space to LAB does not entail much if any color conversion problems, but there is some discussion that going back to a RGB space could really mess things up if (something about working color space not matching...but I don't understand the details.)

Thanks

[redbutt]

I originally posted this in the "Talk about photography" thread....


Since there have been enough questions about this on the forum, I thought I would share one more link. Popular Photography magazing published an article in May 2001, that was VERY good. It covers all the important things you need to know including profiles, color measuring devices, etc. (edit) and gamut.

http://www.popphoto.com/HowTo/ArticleDisplay.asp?ArticleID=23

(If you are using an Epson printer, be sure you click the "STEP BY STEP DIGITAL COLOR GUIDE" link on that page)

I found this to be very helpful and used the advice pretty mush step by step to set up my new Epson Photo 2200 and the results that I get are amazing. I have to tell people that they are looking at an inkjet otherwise they would never know.

Happy calibrating.

[Roger_Cavanagh]

I'm just back from a cumulative 19 hours flight, so I hope this answer doesn't suffer from too much jet lag. :)

bonoboprobo wrote:
Lots of questions under one post, hope it's not too muddled.

So I've been doing a lot of reading on Roger Cavanaugh's, Fred Miranda sites and while I understand the basis for the actions in the work flows, I am not quite clear on some of the details of the use of color spaces and color profiles.

There is a significant difference between the approach Pekka has propounded in the LinearSharpen method (which I have employed in LinearSharpenMenu) and Fred's approach in Linear Pro Batch.

Both methods operate on the linear version of the raw file. Personally, I have never seen an detailed definition of the linear file. It seems to be regarded as the purest form of image data that can be used, i.e., the minimum amount of processing applied to the image data as captured by the sensor. If anyone has seen a good explanation, I'd like to know.

LP uses the more usual approach of applying an ICC profile for the input device (the camera) with some minor curve adjustments for brightness before converting to a standard colour space such as Adobe RGB or sRGB.

(I'm not going to expand on what ICC profiles are for. There are good links on this page http://www.rogercavanagh.com/library/t18_colour_management.htm.)

LS does not use a specific camera profile, but applies much more aggressive channel and level adjustment. This effectively simulates a custom profile, but at the expense of speed, but with the benefit of extra control.

Here is the gross structure of a good workflow as I see it:

-Take picture in .CRW format
-Download and linear convert with BB
-Futz1 in PS to get a good presentable image

At some later date...
-Futz2 further to prep for presentation in a particular format (.jpg for web, print to local inkjet, send to photo printer, etc)

Clearly Futz1 and Futz2 are where the action is, and my questions are mostly about Futz1. There has been enough discussion of actions and scripts that I probably understand what needs to be done where, but it's the details of the color spaces and color management that still don't make sense.

Questions:
Since .CRW files don't come from the camera with and assigned color space or profile what is the most natural space to assign and where? I sort of have the suspicion that many say the answer is Adobe RGB when the TIFF is opened.

If you open a file with no embedded profile, then Photoshop will assume the currently defined working space. Depending how the colour management preferences are configured, you may or may not get a warning. With LP a custom camera profile is assigned early in the action. With LSM Wide Gamut RGB is assigned as the first step. All major adjustments in LSM are done in this colour space until the user chooses to convert to Adobe RGB or sRGB.

If you are using LP, LS or LSM then no profile should be assigned as this is handled by the actions.

But what about Wide Gamut space, where is it appropriate?

Why is it that the camera ICC profile is not applied during or right after the conversion from .CRW? I get this impression from some the discussion of work flows at Rodger and Fred's sites. For instance BB can embed a profie during conversion.

With LS/LSM this would not be appropriate. However, if you are using LP (for D30), you could embed the profile during BB conversion, but you would need to adjust the action accordingly. I haven't seen the D60 version of LP, so don't know, if the same thing applies.

Would embeding a profile during conversion change the image created or just the display of that image?

It's my understanding that embedding a profile is the same as assigning in PS. The appearance of the image changes, but the underlying colour number data does not. You can see this operating in PS by choosing Image>Mode>Assign and then picking different profiles. Some will hardly change the image, others make a big difference. In PS Image>Mode>Convert to profile will not change the appearance of the image, but changes the colour numbers to map into the new space. If the new colour space is smaller (or a different shape) then some colours may change to fit into the new space.

Why is it that issue of Magenta Overload http://www.rogercavanagh.com/scripts/magenta/pages/magfix01.htm out is not cleaned up by applying the camera's ICC profile? I thought that was the point of a profile, to say that this color from this device is really this other color?

LS/LSM do not use camera profiles, hence the fix. Bear in mind that no camera - film or digital - can capture all possible colours accurately. No workflow method will work on all kinds of pictures. Sometimes the magenta fix is appropriate, sometimes it is not.

In any case how is it that camera, monitor and eventually printer profiles get stacked on each other in PS? In PS I can see that you can turn on and off seeing the picture in proof colors and select which profile prism to see the picture through, but don't quite get it. BTW my system is NT4.0, which I understand can be made to manage color, but maybe I'm missing some details at the system level. But it would make sense to be able to do it all through PS.

Profiles are not stacked. Profiles are used to describe how a device (input or output) recognises colour. If you use a profile for an input device such as the D30, it is usually appropriate to convert to a standard working space, such as Adobe RGB. Printing in PS does a temporary conversion to a colour space that defines the printer (+paper combination) being used, but the change is only temporary unless you use Image>Mode>Convert. Proofing in PS will show how the image colours will be changed when printed. The gamut warning shows where there will be a significant change. Printer colour spaces are usually smaller than Adobe and so colours will often need "adjustment". There is good stuff in the links I mentioned before.

How would I go about using a linear workflow and LinearSharpen if I have a Canon PowerShot S30. Look I know it's not a D30, but I gotta start someplace.

You gotta suck it and see. :) Take a few test images, run them through LS, LSM or Pekka's new D60 offering. Find out colours aren't right and work out how to adjust the action to fix them.

I understand that going from a given RGB space to LAB does not entail much if any color conversion problems, but there is some discussion that going back to a RGB space could really mess things up if (something about working color space not matching...but I don't understand the details.)

Conversion from RGB to Lab space is supposed to be non-destructive, but it isn't. If you check the histogram of an image, convert to Lab and back to RGB, and then check the histogram, usually there will be small differences. Not big enough to notice, but a difference IS a difference.

I discovered by chance that the colour space is lost when converting to lab. This makes sense as lab is supposed to be a device independent colour space. However, when you convert an image back to RGB, it assumes the default colour space. If this is different from the image's original colour space, your image colours can be seriously messed up. It's not apparent in every case - I spotted it with an image that had very bright yellows that became blown after the conversion in and out of lab.

Regards,