In a recent question, Steve Ross points out that Adobe Photoshop has a useful feature for matching color between photographs. Of course, color adjustment can be accomplished in Gimp with careful use of the curves tool, that requires both expertise and a good eye, while Photoshop's Match Color tool seems easy to use with little manual input required.

Is there an equivalently-easy way to accomplish this in Gimp? Or even a moderately-intermediate approach?

I've found a tutorial for A "Match Colour" Method for the Gimp, and a Colour Match Plugin, but these seem more concerned with special effects (or with matching skin tones) than with the more general yet simultaneously easier tool provided in Photoshop.

Here's an example of the results of using the "Match Colour" method as suggested above (or in @dpollitt's answer).

I took a single (boring; sorry) photo of some plants my porch and saved as jpeg in camera, once with auto WB and once with WB set to tungsten:



Then I ran the Gimp script, and got this:

match colour gimp script result

This is kind of cool, but not what I'm going for. It's my impression (maybe I'm wrong — please correct me and this answer if so) that the Photoshop tool can, with the right settings, basically do this magically.

Here's Steve Ross's sample image run through the same script:

green landscape

This does not compare favorably to the result Steve gets with Photoshop. Is this just a matter of the script not having enough knobs to twiddle, or is it fundamentally insufficient?


5 Answers 5


Gimp does not have an automatic way to do it with the built in commands. What you can do is extend its functionality using scripts. Here is a definition of what the script does:

This script matches the colours in one single layer image to the colours in another image

The actual script is here, and you can also find a thread full of info and example on it here. You also will probably need to follow the script fu info here.

  • 1
    \$\begingroup\$ Note that these are the things I linked to in the question. :) \$\endgroup\$
    – mattdm
    Commented Jun 18, 2011 at 22:07
  • \$\begingroup\$ Yes, you did. But the script definition might add to the fact that this is the only option available, so it may be worth noting. I see you have now added examples. \$\endgroup\$
    – dpollitt
    Commented Jun 22, 2011 at 0:30
  • \$\begingroup\$ So, yeah, the answer seems to be that this is a missing useful tool. Some work could be done to fill the gap, but no one had really done that yet. If you wanna edit your answer to emphasize that, I'll mark the answe accepted. (I'm not the one that downvoted, by the way.) \$\endgroup\$
    – mattdm
    Commented Jun 22, 2011 at 2:59

I don't know an automatic way to adjust colors in Gimp, buy you can do it manually with curves. Open the good image, use color picker to see the colors you want to match (activate its info window).

For example, I noticed that in the good image:

  • chair purple is approximately R=140, G=110, B=140
  • leaves are approximately R=110, G=150, B=110

And in the bad image:

  • chair is R=55, G=90, B=190
  • leaves are R=20, G=155, B=180

So there is too much blue, almost correct amount of green, and way to little red.

As a simple correction, I open the curves tool, select Red, channel, and let the curve pass through points (20,110) and (55,140) (because Red=20 should become R=110, and Red=55 should become Red=140). The image already looks much better now...

The next step, I select Blue channel, and let its curve pass through point (180,110) (Blue=180 is reduced to Blue=110) and through the second correction point (190,140):

curve transform in Blue channel

Finally, the green curve can pass through (90,110) and (155, 150), but I don't like the result, and didn't apply the green curve. Look at what you are doing, also don't hesitate to move curve points a little and look at preview (I'd move red curve down a little, in fact, try it yourself).

Corrected image (blind two-point curve adjustments in Red and Blue channels):

corrected image

It is not exactly the same as the target image, but much closer than initially. Target image was:

target image

Consider also taking more color samples and more shades of the same color into considerations (in this example: the wall, dish, ceramics, shadows in the leaves). They may help to construct better curves

Update: being curious, how far this method can go, I wrote a small Python script, which takes colors from the first image, calculates best curves by averaging target values and adjusts the second image.

Corrected image:

corrected image

Target image:

target image


You might also find the get curves plugin helpful.

The basic usage of Get Curves is that you have an "original" and a "modified" version of an image. Open both as layers (modified on top), then use Get Curves to extract the color curve which changes the original into the modified. Now you can apply that curve to another image to get the same 'effect'. Of course, this assumes that the changes between the orignal and modified are possible to represent as a curve, so it works best when the changes were in fact color curve changes… I'm not sure how well it'll work if you don't have an "original" vs "modified" pair.

  • \$\begingroup\$ Could you explain what the plugin does, to help readers who view this answer? \$\endgroup\$
    – jrista
    Commented Jul 13, 2011 at 22:32
  • \$\begingroup\$ explained a bit now, hope this helps :) \$\endgroup\$
    – unhammer
    Commented Jul 16, 2011 at 11:46
  • \$\begingroup\$ I had to go back and read the docs about how it saves the extract curve to a file in the background with no apparent result. It does do a great job with my from-the-same-RAW-file test image — but that's contrived. I'll have to try it for the case where I have one RAW file where I can replicate the awb->tungsten curve and a second JPEG from the same lighting but of a different scene. I'll post back the results of that once I get a chance. I think that'll still be a "here's a way to solve part of the same problem" answer, not "yes, there's a match color function". \$\endgroup\$
    – mattdm
    Commented Jul 16, 2011 at 23:37
  • 1
    \$\begingroup\$ On Ubuntu I had to downgrade to use gimptool (from libgimp-2.0-dev package) to compile the 'elsamuko-get-curves.c' plugin. The plugin is then accessed from the Colors menu, output goes to ~/.gimp-2.0/curves/ directory for me. Results for me on JPEG images were like dithered GIFs. \$\endgroup\$
    – pbhj
    Commented May 19, 2015 at 14:10

Oh yeah a quick and easy way that is kind of a cheat way is to go to Colors, Colorize then try to match it with the others layer's colour


I'm not using Gimp. I've only unpacked it a couple of times, but since I've had a Photoshop license forever, it doesn't make sense for me to use Gimp. There may be reasons that I'm not aware of, but at present, I use Capture One, Lightroom, Aperture, and Photoshop, each of which can help me achieve natural color balancing.

It's worth mentioning that if you have the luxury of shooting targets, you can alleviate a lot of the problems right from the get-go. Match Color in Photoshop was (as I understand it) intended to do an after-the-fact match of whatever colors seemed similar. That implies that you don't have a known neutral or calibrated color set.

If, when shooting, you want to get close to a "correct" color temperature, shoot an 18% gray card or an Expo Disc. That will get your tint and color temperature pretty darn close to right.

If you are concerned that your colors are straying outside the bounds of the straightforward neutral balance I've described, look into something like the x-rite color checker. These are expensive, and require that you shoot a frame of the color target before firing away at the subject. Best suited for still life, fashion, catalog, product, and other setup shoots -- not nature, landscape, street, etc.

I recognize that everything I've mentioned is a new way to spend money, and the cost/benefit may not work out well. These are suggestions based on what you could do if you wanted to get your color dialed in as close as possible to what was actually in front of your lens. They aren't recommendations about what you should do.


  • \$\begingroup\$ Note that the original question pertained to grabbing colors out of a good photo and applying them to one that was inadvertently captured incorrectly. That question is here: photo.stackexchange.com/questions/13177/…. My response to this question (which was downvoted why?) has more to do with making digital images that are resiliant in the first place than with a specific tool. Just because I don't use Gimp doesn't mean the suggestions I made are irrelevant. \$\endgroup\$
    – Steve Ross
    Commented Jul 19, 2011 at 18:51
  • 1
    \$\begingroup\$ I think your suggestions are irrelevant because the question really is about how to do this after the fact. Advice for getting it right beforehand is useful in the absolute sense but not as an answer to the question. Let's go ahead and assume I know what I should have done, but for whatever reason the opportunity is lost. \$\endgroup\$
    – mattdm
    Commented Nov 25, 2012 at 16:46
  • \$\begingroup\$ Also: pointing to the color target is interesting, but not complete without a description of how one might use it. \$\endgroup\$
    – mattdm
    Commented Nov 25, 2012 at 16:47

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.