Didn’t know which category to put this under so put it here.
I like to do a bit of (digital) photo restoring/retouching. Usually to old photos. Recently I have come across a couple of colour photos with an orange all over or in a strip across the photo.
My main tool is gimp.
How can I quickly remove (I assume there is a quick fix) If not, Can it be done using gimp and G’MIC?
Just a guess here since I’ve not done this specifically, but it seems like you could use the color select tool with the proper threshold to select the orangy areas. Perhaps feather the edges of the selection region. Then go into the curves or hue tools and pull down the oranges until the color matches the rest of the photo… It would only affect the selected region.
This is done in about a minute or so. You can definitely do a better job of it by spending more time. I am just giving you a proof of concept here…
Also, if you really want colour, you can colourize individual parts after doing a BW conversion.
The problem is that the orange color really, really throws the image pretty badly. I tried quickly just to bring the colors back into line and it was not happy.
You could get the Green channel (which doesn’t show the banding very much), and then use that as a base for painting the color back in to the image (colorization of old B&W images basically). On the good side, you have the good portions of the image to pull reference colors with…
@patdavid I was just laughing to myself about that
@sallyanne I know of no simple solution to remove the cast while preserving colors.
The easiest fix to this photo I would suggest is as @shreedhar wrote - convert it to black-and-white, then apply color toning if desired. The reason why what worked well for @paperdigits doesn’t work well here is that the affected area is huge, it contains important information (faces), and it affects all channels (lightness, chroma and hue).
Here is a step by step using RawTherapee:
Scan your image in the highest quality possible, preferably to a 16-bit TIFF file. Unfortunately the uploaded image we get to work with is a JPEG so there will be artifacts in some areas - they wouldn’t be there if it was a 16-bit TIFF.
Convert it to black-and-white using a parameter combination which neutralizes as much of the blob as possible. In this example, “Infrared” worked well out of the box, though the shadow of the blob still remains.
I used the “Relative ROYGCBPM” preset to be able to control by color, then I moved the yellow slider down until the shadow of the blob disappeared. At this point you’re basically done.
Now for the other stuff. First we see that there is a gap in the right side of the histogram, so I used a “Before” curve to stretch the histogram to touch both ends without clipping. It’s a simple linear curve.
Why “Before”? You could use the “After” curve, it makes no difference in this case, but using a “Before” curve which “fixes” the histogram leaves the “After” curve free for you to apply some aesthetic effect.
What I would do is generate two copies of the image as layers and place one on top of the other. One to focus on restoring the unaffected area and the other to focus on the orange area. That way you can toggle the visibility of each as you modify the respective areas. Then you can blend them and keep the good parts. A tool that could help you with masking is Filters → G’MIC → Colors → HSV Select.
I don’t use GIMP, but I have some ideas. In Krita, you can set brushes to color blend mode. That will only change color and does not affect saturation or luminosity. I have used that to repair photos in Krita. You can do that with Krita clone brush.