recover the treatment applied by the white balance ?

(lux) #1


I would like to recover the treatment of the color applied on a photo, to apply it on another photo …

For that, I would like to try a white balance on the original photo and apply the settings of this white balance on the second reversed photo (passage in negative)
I reverse the image once again, the treatment applied …

Not sure that this treatment works very well, but maybe it can give me a first track.
How to recover the setting that made the white balance?

(Morgan Hardwood) #2


I don’t follow why you need to reverse and reverse. Could you explain with pictures? Once I understand, maybe I can offer a better solution.

(lux) #3

Thank you for your reply.

My goal is to understand the type of curve applied to a photo by an artist.

For that, I had the idea to apply a white balance on the concerned photo.
This white balance (in my opinion ?!) would allow roughly to find the original photo colors, before the curves treatment.

The data of this white balance would then (perhaps) correspond to the treatment given by the artist in his photo.

So it would be enough for me to apply a white balance treatment (with the data previously recovered) by inverting it on the target photo-test.

So :

Either I reverse the data (difficult) or I apply the treatment on the inverted photo.

Am I clear ? it’s hard to explain a crazy idea:grinning:

I have not tried anything with “ImageJ” … maybe I have to explore this way?

(Morgan Hardwood) #4

No, curves and white balance are two different things.

I can’t help you with finding what was done to a photo without seeing it.

Reversing treatment of an 8-bit image will lead to poor results, and will be impossible if any lossy operations were performed, such as using curves to cut off colors.

If you have access to the person/program/device which performed the treatment, you can feed an identity HaldCLUT image through them/it, then you can use this HaldCLUT to repeat the effect on any photo. See

(Alberto) #5


I’m not sure I fully understand what you want to do, but you might take a look at histogram matching.

(lux) #6

Eurêka !

thank you, you gave me the solution!
why did not I think of it?

(lux) #7

Thanks Alberto.

(Hermann-Josef) #8

@agriggio Can histogram matching be done in RT?


(Alberto) #9

that’s what the auto-matched tone curve does, on luminance only though. if you want to match the RGB channels you need to do it with some other way

(Hermann-Josef) #10

@agriggio Thank’s Alberto. I was just wondering, where it takes the reference image from :slight_smile: . But the context help tells me that the auto-matched tone curve needs an embedded JPG.

However, in the book “Digital image processing” by Burger & Burge one of the plugins for imageJ calculates matched histograms. The version as supplied only works for 8bit grey level images. I will have to look into it to expand it to 16bit RGB-images.

Thank you for pointing this out. It should be of great help if one wants a series of images be developed in a consistent way, which is hard doing it manually.