Image submitted by Ekaterine Shapatava From Atlanta, Georgia
Published: June 19th, 2008
Duration: 09:09
Image submitted by Ekaterine Shapatava From Atlanta, Georgia
Published: June 19th, 2008
Duration: 09:09
This tutorial can be accessed by purchasing an Extended Access window.
Very cool workbench episode. Apart from these new technique for neutralizing color casts, I learned several new things about Photoshop....
Hi Mark,
This technique appears to work okay in the case of this image although a slight magenta cast is introduced. I think it would be valuable to explain what is actually going on.
So what is going on behind the scenes here, is you are taking a point in the image which is tonally most similar to 50% grey (i.e. if you desaturated the image you are taking the area that is 50% grey). Then you are assuming that this will give you the best point from which to take a colour balance reading.
The logic behind this is presumably the assumption that 50% grey is tonally the best representation of the image or something along those lines.
In reality I think any assumption made in getting to this technique is very wrong. Take this shot. The area which turned out to be closest to neutral grey tonally was an area which appears green with the initial white balance. Unfortunately if you white balance a green area, it essentially introduces magenta/purple, in the case of this image, that was too strong and hence you are left with a magenta cast at the end.
Take an image that is white balanced correctly, consisting of several strong colours. All of these with a desaturated tone of 50% grey. Clicking on the different colours will introduce completely different colour casts. Clearly this follows through to similar scenarios with colour casts. There is a good probability that the 50% grey area (when desaturated) will have a colour of its own and hence introduce a colour cast when you apply this method.
There will be circumstances in which this technique works but that will really be luck rather than good practice. It would be just as valuable to click randomly on any point in the image. This is in fact what I do in this circumstance, I click all over the place until I have a point I am happy with, then fine tune it. This method is faster and more likely to be accurate because I am not mechanically applying a method that in all likelihood will not work.
The first image I tried this technique on (a similar scene with a window and tungsten lighting) the mid grey point happened to be the edge of the window frame, which had a blue tint. Using the technique added a red colour cast to the already noticeable yellow cast.
It would be much more valuable to perhaps suggest ways of recognising colour casts (which clearly you can spot with relative ease as this video demonstrates) and then explaining a manual method with curves for correcting the colour issue, or perhaps even identifying points in the image which are most like to produce good results.
I find so many of thee workbenches incredibly helpful. In fact I would even say I learnt a majority of processing techniques from this site. However a lot of beginners will come here and apply this technique and it simply does not work. I recognise you say “this does not work on every image” that is merely a reflection of the fact that the technique is random. If you applied a tonal adjustment to the images that didn’t work, and then tried again I am sure it would be possible to correct the cast.
I really respect what you do here, and I by no means intend this to be a slight on what you do here, although I do appreciate that this comes across as rather blunt to say the least!
All the best,
Alex
I tend to agree with Alex, but for beginners, who have no idea how to curve in channels, this is not a bad method to try - although when I tried it on 'normal' images it completely failed.
There is an average method (with CS2+) which works pretty well.
But the real answer to this problem is to use CurveMeister with Hue clocks and pins - it really is worth spending all of $10 (?) on their 4 week course - you will never look back!
Thank you for the tutorial and keep them rolling.
Chris
Very cool technique. Thanks again.
Joseph
Mark
Neat trick/effect/tool/workflow
When I am trying to learn something, I find I learn best by understanding what is going on, rather than doing it by rote. Understanding should allow me to adapt for other situations.
While ALEXNAIL explains his understanding of what is going on when he said:
"So what is going on behind the scenes here, is you are taking a point in the image which is tonally most similar to 50% grey (i.e. if you desaturated the image you are taking the area that is 50% grey). Then you are assuming that this will give you the best point from which to take a colour balance reading.
The logic behind this is presumably the assumption that 50% grey is tonally the best representation of the image or something along those lines."
That does really explain to lesser mortals how your choice of technique created this situation, nor why 50% grey is the route to go.
I realize the full explanation might take a much longer time to explain, but personally I would learn more.
Keep up the great, helpful work with the workbenches !!!!
Mark,
That's a pretty cool trick. Thanks for another tool. I am learning so much from these tutorials. Love your book, too.
Best,
Jim
This is a great tip, Mark. I tried it on an image of a room interior that had a strong color cast [all incandescent lighting], and while the color correction attained was not perfect, it gave a far better image than the original.
I do not have PSCS2 or PSCS3, and use PS Elements 6.0. There is no color sampler eyedropper in Elements 6.0. So, I hope you don't mind me adding an addition to your process for Elements 6.0.
Here's what I did in Elements 6.0. I used your method to create the two layers, and determined where on the background image the 50% neutral gray area was located. Then I deleted both layers. With the background image open, I opened levels. The levels pallete, in addtition to the sliders, has three eyedroppers - white point, neutral gray, and black point. I clicked on the neutral gray eyedropper, and then with the eyedropper, clicked on the neutral gray place in the image found through your technique. Then I duplicated the corrected image and saved it. No corrections were saved on the original image, so I still have that. And the new duplicated image is the starting point for any further corrections or adjustments I need to make. It's not perfect, as I said, but I got a pretty good result.
Thanks.
Marie
To leave a comment you must login or become a registered user.
Discussion Open - 7 Comments