I’m having some color management problems and there are some curious findings. The pictures in question are photographs that have been edited using software such as GIMP and Rawtherapee. I think I may have done something, unknown to me, that makes a problem that might otherwise not be to obvious very conspicuous. A specific example involves using GIMP to create a border with sufficient width to add descriptive text to a photograph. The border color is picked from the picture but ends up being solid as well as the most concentrated occurrence of said color.
I have 2 inkjet printers. One an HP Officejet the other a Canon Pixma Pro-100. I bought the Canon to use for photo printing. In this case the pictures were taken with Canon cameras. My display has been calibrated. The paper is HP Premium Plus Photo Paper (Glossy Glace).
In some cases when using the Canon printer the border color actually changes from what is displayed on screen. For example, from a grayish brown to a grayish green. If I then use the HP to print the same image file it comes out much closer to what was expected.
Another problem seems to be that the printed pictures end up being darker (i.e., less brightness) than what I’m seeing on the display. Sometimes to get a good printed picture I have to increase the brightness to the point that the picture looks overexposed on the display.
Insofar as the goal is to produce images that look good both when printed and viewed on screen this is troubling.
The printer drivers allow for selecting paper that corresponds to what the respective printer manufacturer sells. It looks like there are ICC profiles that correspond to each paper selection, which causes me to start thinking that this problem might have to do with differences in paper and the corresponding color management. While I would have expected such considerations to be appropriate for trying to produce very high precision results, I also expected that anyone with a decent printer and paper made for photo printing would get a result that reasonably approximates what is seen on a monitor. Of course, I also realize that everybody’s monitor is slightly different to say nothing of eyes which sort of says that everybody sees something that is slightly different. At the same time I thought it might be a sufficiently reasonable approximation that differences would be unnoticed by casual observers.
Is it possible that differences in paper can cause this kind of problem? Is there some other explanation?
Paper can cause this problem, yes. So can the light under which you are viewing your print. I have the latter problem in my apt; the light is so dim every print looks like trash. Open the blinds and everything is good.
It can also have a lot to do with driver settings… And color spaces.
My printer supports Adobe RBG, so I work in that space all the way through my process. It makes things simple.
I’ve done some experimenting. For example, GIMP seems to want to convert everything to what it calls sRGB which I think corresponds to to the ICC standard (? sRGB IEC61966-2.1). However, Canon (both my camera and printer) have something called “Canon sRBG 1.31”. I’ve tried it where I both allow GIMP to convert and keep the Canon sRGB 1.31. Whatever, difference is imperceptible to me. Furthermore, even though I’ve calibrated my display it isn’t clear to me how that relates to these profiles. Also, unlike GIMP Rawtherapee doesn’t ask any questions about conversion but it does have a bunch of choices the default being sRGB IEC61966-2.1.
Can you (or me) use Adobe RGB in GIMP? If so how?
Your point about ambient light is valid. These prints do look better outdoors and of course my monitor would be lousy looking under such conditions.
The other experiment I have planned is to buy some Canon paper and see what difference that might make. Unfortunately, I haven’t been able to find the Canon paper locally but have some on order. We’ll see.
Subtle differences are no surprise to me based on having done some reading about color management. However, brown becoming green is a surprise to me.
To me that doesn’t sound like a paper problem but a problem with profiling the monitor. Either your profile is bad, or you are not setting it up correctly.
About the different flavours of sRGB, in practice those should all be interchangeable.
I just got a Pro-100 and with the Canon luster paper, and the matching driver setting, I get results that exactly match what I see on my monitor, including gamma. I used to think that you always had to lighten an image for printing relative to display on a screen, but I guess that’s not true if you use the right paper profile.
I don’t have Canon paper but speculate that having a profile that matches the paper is what I need. The paper I have is mostly HP. The color match is better when I print the same picture on my HP printer where I’m thinking the profiles may match the paper I’m using.
I have some Canon paper on order which will allow test of this speculation.
I’m not sure if this will receive any notice but I have now received the Canon photo paper previously referenced and have completed a few experiments.
The basic finding is that paper matters but it doesn’t seem to have much to do with brand. For example, I have 2 kinds of Canon paper. Paper 1 (CP1) is called “Photo Paper Plus Glossy II”. Paper 2 (CP2) is called “Photo Paper Plus Semi-gloss”. When I print the same picture file on my Canon Pro-100, CP1 comes out pretty good but not what I’d call a match with the monitor. However, the same thing using CP2 gets a result pretty close to the terrible print that triggered this post in the first place. In that, about the same as when using HP Premium Plus Photo Paper (Glossy Glace), which I’ll now call HP1, on my Canon Pro-100 printer. I should note that I went to a lot of trouble to make sure that I got the correct ICC profiles specified when doing this experiment on the Pro-100 with Canon paper.
Interestingly, I didn’t know how to, precisely, do that (i.e., specify ICC profiles) when I first posted this topic. I naively thought that selecting what Canon calls “Media Type” (i.e., paper) would do that but apparently this amounts to only a good guess rather than the actual profile specification. I have since discovered that when I specified a Media Type of “Other Glossy Photo Paper” to the Canon driver that appears to have resulted in the use of very generic ICC profile that likely does NOT correspond to any specific paper. This caused me to think that I should try telling the driver that I was using CP1 when in fact the paper was HP1. In that, use the profiles for Canon paper on the HP paper. This resulted in prints on both CP1 & HP1 that were very nearly the same. In that, a barely perceptible difference.
I’m not sure where that leaves us, other than it is a lot more complicated than I had imagined when I bought a printer specifically for photo printing. With that said, I have to imagine that it doesn’t get any easier to rely on someone else for printing.
You may be able to try ‘Soft Proofing’ with Gimp.
I proceed as follows:
I adjust my screen with a probe and make a first editing of my photos.
To print I visualize and modify the photos for printing (enter an ICC profile, edit, remove the ICC profile).
So I have two results, images to view on screen and images to print.
There may be interesting information by mixing the term ‘Soft Proofing’ in a search engine.
:o)
There are services available where you can have a profile made for your printer. They send you an ICC file. Unfortunately, Gimp has no means at this point to convert you image to this ICC profile.
I used to use CinePaint to convert images to a printer profile I downloaded for CostCo photo printing for each store I used.
believe Krita is able to do this conversion also.
Without being able to do this conversion, it’s a hit or miss process as you have discovered.
So, a quick difference between screen and print that no one’s mentioned yet, specifically about how the images end up looking darker in print. Its not necessarily a solution, just something to keep in mind:
Screens produce color in RGB color space with light. Printers use CMYK color space and use pigments to produce the colors. That’s already a huge translation that needs to happen. I personally found that sRGB will always end up looking darker when printed out without calibration on both end. Without the option to fully convert GIMP files to CMYK profiles, you’re gonna be fighting with hardware.
I am not a developer and I don’t know any software language, but have you tried color calibrating your printer to your monitor to your camera and back again? Something something drivers?
FWIW, I work for a commercial printer and I just hand my native files to the production team with a ‘plzkthx’, but I’ve had to repeatedly ask them to lighten images with sRGB profiles for print after proof rounds, no matter how they look on any monitor. Often I’ve had to specifically reduce yellows too, which could account for your green problem if your browns are cool toned. (which would include the cyans during printing)
Adobe RGB does a bit better, but its not the same as being able to work in CMYK.
Additionally, paper brands dont matter as much as paper coating. Gloss vs semi gloss vs matte vs lustre vs uncoated will all produce slightly different colors, but overall it comes down to how your printer reads the color profiles being sent to it.
I don’t mean to comment spam, so I’m just gonna edit this one. I had two beers and then another questions.
What file type are you sending to your printer? What settings are you exporting it with?
I have calibrated my monitor but I haven’t made a custom profile for the printer/paper.
I’ve pretty well figured out that printers are quite different from monitors. It would also be fair to say that both printers and monitors can be quite different from other devices of the same type.
Given that so much picture viewing is done on monitors that very wildly in characteristic it seems that there exists some futility when it comes to trying to develop an image for such an audience. At least prints end up in a final form that is the same for everyone. Of course it would also be fair to say that human vision varies a fair amount as well but as best I can tell nobody is yet trying to factor this into the formula.
I’m still struggling with workflow but presently I’m creating both a tif and a jpg for finished product. Intention being that the jpg is good enough for those using unknown electronic devices to view an image whereas the tif is intended preserve as much quality as possible for archiving. The tif would also be the one selected for printing.
Your comment about coating being what matters fits well with my own findings. With some experimentation, I’ve been able to identify Canon supplied profiles that work well with my HP paper but this entailed using the profiles designated for Canon paper and not the generic ones.
Seriously, what I wonder is how to go about it when producing for display somewhere you don’t have control over it. I recently had to produce both digital and scanned photographs for my son to construct a video slide show to be displayed at our local church. I studiously made sure all my exports were converted to sRGB with a corresponding embedded ICC profile (although i was pretty sure that wouldn’t be used), but thinking sRGB gamut would be handled okay by the church projectors. Alas, it was not to be, the vivid colors, particularly reds, were posterized and crushed at the vivid extremes.
I’ve considered playing around with xyY values to make a sort of ‘subset’ sRGB profile that I could use for export in these situations, but I’m not sure how well that’d work without multiple trips to church to play my latest iterations. i wrote a C program to read a text file profile definition and use LittleCMS to construct the ICC file, so I can crank 'em out, but it may really require a LUT profile to be able to use the better rendering intents. Or something else my little brain can’t comprehend…
Son and I went to a seminar recently conducted by a well-known landscape photographer. He brought his own projector, thank you, images displayed quite nicely. I can’t afford that…
David, are you still struggling with this problem?
I just noticed this thread, and I believe that I will be quite able to complicate matters even more – this really is a quagmire!
a) Have you google’d for “why are my prints too dark”?
b) Paper qualities matter a great deal. There is an enormous difference between printing on a “photo glossy paper” and printing on a piece of newsprint (but the latter can be done!).
c) Are you on Linux, Win or what? What print manager are you using? What applications do you print from?
d) Do you own (or can borrow) a spectrophotometer?
e) What are your color management settings in Firefox?
Oh man - this is an entirely new can of worms (do you target fluorescent? tungsten? LED? incandescent? direct? diffuse? or, God forbid, have it hanging near a window and have a bulb focused on it?).
– this really is a quagmire!