Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Color profile verses working space
#1
When I open a photograph stored in .tif format a dialogue box (see attached file) opens asking if I want to convert to RBG working space.  The message says the file has an embedded color profile identified as "sRBG v1.31 (Canon)".  It seems to be asking if I want to convert to something called "sRGB IEC61966-2.1".

It sort of looks like the choice is between 2 versions of sRGB where I have no idea what the difference might be but do think that the embedded profile might be one associated with my camera by its' manufacturer (i.e., Canon).  I also own a photo printer which is also manufactured by Canon.  Insofar as one of the uses of the the results of any editing done by GIMP is to print, I can't think of any reason to do this conversion.  What am I missing?

BTW I did experiment with both options (Convert & Keep) and output files exported to .tif format seem to retain the choosen option.  In that, there is no conversion done when the file is exported.  The differences on my display which is a moderately good (27" 1920x1080) one (but has not been calibrated) are imperceptible.
Reply
#2
A search around comes up with some old information but indicates that there is little or no difference between the two.

If sending to a printing service usually sRGB IEC61966-2.1 is safe, although some printing companies have their own icc profiles specific to the printer/paper they use.
Reply
#3
Another point of confusion pertains to how one determines the color (?ICC) profile of an image file.  For example, is there data that is part of the image data that defines the color profile or is it something that gets identified only by metadata.  Something I notice is that when an image file, like a photograph, is opened by GIMP and subsequently a file is exported in a format such as jpg the metadata associated with the original image file is mostly gone.  Certainly there is no color profile identified in the metadata of the exported file.  Even when GIMP is allowed to convert the opened file to its' preference for sRGB you can't tell that the exported file is sRGB by looking at the metadata.

I am aware that GIMP 2.8 admits to deficient metadata handling and that may explain this problem.  I've been hoping that GIMP 2.10 will be better.  Am I dreaming?

I'm no expert but it seems like this is important when GIMP is used to develop photographs.
Reply
#4
Quote: how one determines the color (?ICC) profile of an image file. For example, is there data that is part of the image data that defines the color profile or is it something that gets identified only by metadata. Something I notice is that when an image file, like a photograph, is opened by GIMP and subsequently a file is exported in a format such as jpg the metadata associated with the original image file is mostly gone

The data is held in the EXIF 'header' There are tools to save and then restore this after Gimp has stripped it. Usually this one https://sno.phy.queensu.ca/~phil/exiftool/ although there are others.

Quote:I am aware that GIMP 2.8 admits to deficient metadata handling and that may explain this problem. I've been hoping that GIMP 2.10 will be better. Am I dreaming?

There are improvements. This is Gimp 2.9.9 Basically same as Gimp 2.8.22 with extra information/options

1. open image 2. image properties 3. assign a color profile. https://i.imgur.com/0yeWX8g.jpg

There is a Gimp 2.9 metadata editor. Looks like this: https://i.imgur.com/4fnbxNu.jpg

Need confirmation of profile? For something with a GUI try XnViewMP or command line, Imagemagick and identify -verbose filename.ext
Reply
#5
I have been using both ExifTool & XnView (both windows & multi-platform versions).  My level of sophistication is such that I rely exclusively on the use of ExiftoolGUI and haven’t yet begun to exploit all of the capabilities.  I also believe that ExiftoolGUI is aging and no longer being maintained.  As good as it is this creates lots of concern for use in workflow procedures being developed with durability as an objective.  However, my basis (confidence if you will) for saying that the metadata is mostly gone is based on what these tools report.  My lack of sophistication is also why I’ve resisted any temptations to install GIMP 2.9 development versions.

The idea of exporting an ICC profile from a source/input file and then importing that profile back into image files exported from GIMP would suggest to me that what GIMP calls “converting the working space" ought be avoided.  In that, be sure to tell GIMP not to do conversion so as to preserve the color profile that you have (exported).  Doesn’t it?  In the event that GIMP is allowed to perform this conversion, which includes essentially all of my work prior to now, is it possible to find something to import in those cases?

I've now experimented with the suggestion about export and import of ICC profiles using ExiftoolGUI.  I used the menu item indicated as "Export/Import>Copy metadata from single file ...".  The source file was .tif format and the target (updated) file was jpg format.  The source file was equivalent to a camera developed image but was created without any editing/revision of a raw file using the software supplied by the camera maker (Canon).  Curiously there are entries for RedTRC, BlueTRC, and GreenTRC within the ICC profile which all say "(Binary data 2060 bytes, use -b option to extract)".  It isn't clear to me whether that binary data is part of the metadata or image data but I'm suspecting that the binary data is an important part of the profile and that it was NOT transferred.  Any idea what to do now?
Reply
#6
Getting a little off the original what is the difference between sRBG v1.31 (Canon)" and "sRGB IEC61966-2.1".

Quote:The idea of exporting an ICC profile from a source/input file and then importing that profile back into image files exported from GIMP would suggest to me that what GIMP calls “converting the working space" ought be avoided.

That is one view, however over the years there are certain industry standards such as AdobeRGB1998.icc suitable for the majority of purposes. A useful comparison here: https://www.cambridgeincolour.com/tutori...gb1998.htm

Send your photo off to a printer and they will probably ignore the embedded profile and use stock sRGB. I mentioned earlier that a commercial printer might specify a profile for use with a specific printer/paper. This a quote from my local printer

"These are the ICC Colour Profiles that have been specifically created for Digitalab’s Lightjet 5000 printer. These are for photographers with a knowledge of Colour Management and are by no means essential." - note no means essential.

Quote:In the event that GIMP is allowed to perform this conversion, which includes essentially all of my work prior to now, is it possible to find something to import in those cases?

Any number of icc profiles free to download if you search. I will start you off, top of the list http://www.color.org/profiles2.xalter

Gimp 2.8 will change the profile Image -> Mode -> Assign -> Colour Profile. As a note Gimp 2.9 can extract a profile as something.icc Using Windows? Partha's Gimp 2.9 co-exists with Gimp 2.8 (using linux, there are portable versions of Gimp 2.9 that might work depending on distro)

Going back to your first post "the differences on my display which is a moderately good (27" 1920x1080) one (but has not been calibrated) are imperceptible."
There is information here: https://www.cambridgeincolour.com/color-...inting.htm where even an adjustment in monitor brightness can get a better match between what you see and what you print.

Your posts do read like you should really be asking advice on a more photographic biased forum, such as https://discuss.pixls.us/
Reply
#7
Yes, I am coming at this from a photography perspective and it's pretty evident that you know a lot more about it than I do. The Cambridge Color articles are excellent. I've dabbled with Rawtherapee and the forums you reference are the only thing available with it. I'm very grateful for the feedback herein even if the discussion strayed a bit from my original question. However, that question did result from a very real problem that I experienced with GIMP for which I may be guilty of attempting to identify a plausible explanation myself. The problem that triggered the post comes from trying to do something with GIMP that one might argue has little to do with photography. If the question is whether I'm more ignorant about photography or GIMP, the answer is I don't know enough about either for it to matter.

One of the things that I've learned to do with GIMP is to frame a photograph in a somewhat decorative way to which I can add text which explicitly describes that picture and becomes part of (i.e., difficult to separate from) the picture. In that, better than what people did in the old days when they wrote on the back of a picture. This idea was motivated by the idea of producing long lasting printed pictures. I'm afraid when it comes to assembling such things as family albums intended to be passed down to future generations I have more confidence in the durability of paper & pigment than I do of digital media which is useless by itself (i.e., other technology and, oh yes, knowledge needed to use).

A recent experience involved doing this where the color of the frame is picked for complimentary purposes from the photograph. In that, digital image from camera pasted onto background frame sized for photo paper using GIMP. When the jpg picture exported from GIMP is printed on my Canon Pixma Photo Printer the color of the frame changes from I'd describe as grayish brown when viewed on my monitor with some aging eyeballs using GIMP, or any other software, to what I'd describe as grayish green on the photo paper viewed with those same eyeballs. What is very curious is that the object in the picture from which the color was picked appears the same on both paper and monitor. Also, my workstation is positioned near a window that delivers lots of natural sunlight during the daytime. I just last night came to notice that the colors in question look much more like I've expected when viewed by an incandescent lamp. Is this a real life example/lesson in white balance?

The good news is that these frames are not intended to be focal points of attention but rather unobtrusive. The color deviation did not change this and the picture is still good even if the color of the frame is not faithful to what I was trying to create.

Finally, the summary of the referenced article at Cambridge Color says something that makes all kind of sense to me which is "My advice is to know which colors your image uses, ...". Might this mean I need to learn how to understand histograms?

Sorry about the verbosity but THANKS AGAIN.
Reply


Forum Jump: