Several times I stumbled upon a tutorial that states that it is simple to remove color cast by the numbers (see for instance here on this forum (http://www.retouchpro.com/forums/rp-tutorials/17238-better-printing-all.html)).
The method in short: use treshold to determine darkest and lightest point; use 50% gray layer in difference blend mode to determine a 50 % grey point; use these points to set the three eyedroppers in a levels adjustment and voila: color cast gone!

But it never worked for me.
And I think it can't work like this because this method translates a 50 % gray to print as ... a 50 % grey! And precisely because of the color cast that point shouldn't print to 50% grey at all, it should print to the opposite of the (unknown) color cast!
I think you can't really do without a subjective decision as to which point should be neutral grey...

Anybody there with a brilliant method to prove me wrong ??

Actually I hope so, because I shurely would like to have a simple tool to make the decisions for me...

Hi jojaspi,
This video tutorial will help you. www.Layersmagazine.com/color-correcting-images.html (http://www.Layersmagazine.com/color-correcting-images.html)
Ray

Hi jojaspi.....are you sure you're getting the tutorial right? it always works well for me :) and Dave Cross (one of the "photoshop guys") did a demonstration of it on Photoshop TV some time ago. Here is the same method from a different site, http://www.onlyphotoshop.com/photoshop-tutorials/curves-color-adjustment.php ....Oh and welcome to RTP.
Regards....John

If we're talking color cast here I swear by a Katrin Eismann method. I've never found anything that works better or as fast. Make two duplicate layers of your image. In the first duplicate do a blur/average. Take your little eyedropper tool from the toolbox and click on the resulting blur mess in your photo. Go up to the next layer and open either a curves or levels adjustment. Using the middle eyedropper ind the dialog box - the set gray point one - click on the averaged color in your over in your tool box. The color cast will vanish instantly.

If we're talking color cast here I swear by a Katrin Eismann method.

Charlene,

I will try this method, I can imagine this one works well and fast. Doesn't this in essention do the same as autocolor (with more flexibility because of the curves)?

John,

I think the flaw in my "proof" that the method can't work is that I focused on the Grey point only, without thinking about the effect of the black and white point.
The reason my try with this method in practice didn't work out very well lies - I guess - in the fact that my black and white point were BOTH blown out. In other words neither of those two points had a color cast!
I think the method only works if both the black point as the white point show signs of the color cast.

But still, the only thing a 'method by the numbers' can do, is make the image neutral.

Correcting a close-up of a red rose will - by whichever method - deliver disappointing results. Unless you like a grey rose of course.:grin:

So removing a color cast will always remain a subjective decision...

I think the flaw in my "proof" ...

No, No, NO!

Theoretically it CAN'T be right after all.

If we use the 50%grey-difference-blending method to determine the grey-point, we will map a point that is grey in the color-casted original to grey in the resulting print. That's not right.

Example: if we have a heavy blue cast, the "grey-point" we will find is a point in the image that's yellow in the reality; yellow-in-reality plus blue cast = neutrally grey in the image. Color-cast-correction should bring this back to yellow-in-the-print. Not to neutrally grey.

Do I make sense, or am I wrong??:surprise:

The "gray" point is not necessarily equal to the neutral point and is not necessarily gray. In real images, things whose color are gray are often tinted by reflections from surrounding light sources, objects, and reflections. Moreover, it is rare that a cast is linear across the spectrum of highlight to shadow.
Examination of image 1 below clearly indicates a blue cast. If you create a gray difference layer and set the neutral points, you find that they are not yellow but actually blue as seen by the eye dropper samplers 3 and 4.
If you proceed to make a levels correction and just set the shadow and highlight (samplers 1 and 2), the image changes contrast a little but still sports a heavy blue cast. It isn't until you set the gray point 3 that the green of the trees returns. The water becomes gray-ish in the shadows and bluish where is has a lot of reflected sky on it.
Is this now correct? Well it looks more natural, although it is probably not perfect. The reality is that images which have heavy color casts have the relationships between the pixels of the various channels shifted non linearly. Three correction points in a Levels adjustment can not restore the entire balance. Finding the neutral point (at leastin luminosity) in conjunction with B&W point adjustment brings the image back into a range from which it can be fine tuned to what is pleasing to the eyes of the beholder.
Regards, Murray