For this exercise, you should find two or three images that have what you judge to be a significant colour cast, as the main purpose of the exercise is to ‘correct’ it. For most photography, all that is important is that the overall colour looks reasonable and expected. Make sure that at least one image contains a surface that is ‘known’ (that is, expected) to be grey.
Use your preferred processing software. First examine the image and make a judgement on the colour cast. As in the previous exercise, there are two approaches, depending on whether or not you shoot RAW.
As in the previous exercise, I will use the same three photos for both the JPEG and RAW processing; this is so I can once again see how Photoshop treats these different file formats.
TIFF or JPEG
Unlike the RAW workflow, an actual white balance control is not available, but there are ways of simulating it. These vary with the software, but all have the equivalent of a ‘grey dropper’, tool which you click on an area of the image and which then shifts the entire colour of the image so that that small area becomes neutral grey. Try this first on the image that contains an expected grey. In Photoshop, the grey dropper is available in both the Levels window and the Curves window, and once you have used it, you will be able to see that each channel, R, G and B, have automatically shifted. If the correction is too extreme for your taste, you can go into each channel in, say, levels, and drag the central slider back towards its central starting position. This will reduce the correction.
In an image that does not have any identifiable neutral grey (most do, by the way, although there may be tiny patches that you have to hunt out), make the correction by dragging the central sliders of what you judge to be the opposite colour (R, G and B) in Levels or Curves or whatever is the equivalent control in your processing software.
If you feel that the image could benefit from this, then attempt a local correction. Again, the methods vary between software. The Photoshop approach is to make a selection (such as by painting in Quick Mask) and then use any of a number of controls such as Curves, Levels, Photo Filter or Hue/Saturation.
I have chosen three different photographs to carry out all of the tasks mentioned above. Two of these images have deliberate changes to their white balance and one does not.
My first image, taken for the Digital Image Qualities white balance exercise, shows the grey structure of a bridge in Ulsan, South Korea. In this image, ‘shade’ has been selected for the white balance (my shooting conditions at the time), and as the camera is trying to import warmth into the shot, an orange cast has been produced, which is incorrect.
For comparison later, a copy of the images histogram is also included.
First, I used the grey dropper to change the overall colour cast to a neutral grey, I achieved this by selecting one of the aluminium bike stands at the bottom of the image.
This did a really good job of bring the colour cast nearer to the actual colour of the scene, although I was sill a little unhappy with the effect. I therefore went into the Levels application and played around with each of the colour channels, enabling me to bring the colours cast more in line with how the image looked at the time of shooting.
At the time I did not think that I was making huge changes to the image, but when comparing the two histograms, you can really see a difference, in not only the shape of the graph, but also where the combing has appeared, which is obviously due to the loss of pixels in the image.
Below is a copy of the final adjusted image.
In order to confirm how the image should have looked, below is an identical photo, taken with the white balance set to Auto. This shows the best representation of the scene.
You can see that I have almost matched the colour cast in my manipulated image, and with more tweaking, I could probably bring this completely in line, but if I had played around even more, more degeneration of data would have occurred with the possibility of making the image completely unusable.
I took my second image during a recent morning walk, and although the correct white balance was selected in camera, the petals of the dying flower have an orange cast, which is not quiet right.
I did not feel the need to make major changes to this image, just a subtle lightening in the actual plant, so, once again I turned to the Levels application, but this time used the white dropper. My point of reference was the corn, sitting just above the flower head, and this selection has made a subtle difference to the overall colour cast and has brought the flower petals in line with what I saw whilst out on my walk.
When comparing the histogram from these two images, you can see that there is little change to the actual shape and line of the graph, but once again there is pixilation present, confirming that some alteration has been made to this image.
Moreover, to confirm that the procedures carried out in Photoshop has brought the colour cast back in line, here is a different image showing the correct colour cast within the plant and its surroundings.
The final image in this section has been chosen as there is no neutral grey present in the shot, so to regain the colour cast I need to work solely with the sliders in either the Levels or Curves application.
This image is of an old battlement at a fort local to my home in the UK. Whilst taking photos, I tried using the Shade white balance setting to see what the image would look like; as you can see the final product has an orange colour cast.
As it will be relevant later, I have also included is a copy of the histogram produced by this image.
I had previously seen good results when using the Levels application, so decided to use this again for this exercise. Using the central slider, I began to make changes to each individual colour channel, and found that the blue channel gave the best result, although I also made small changes to the red and green channels. The histogram produced by this manipulation is nothing like I have seen before, but it does make sense as the combing is more prominent in the blue channel, where most of the changes were made, and the small spikes and troughs represent the other two channels.
Of all the images I have altered during this exercise, this image is the one I am most pleased with as I did not think I would be able to achieve my objective completely and I did not think it would be so easy to do. The final manipulated image is blow, along with a photographic comparison or what the colours should have looked like.
Because RAW shooting keeps settings separate from image capture, any RAW converter software valuably allows you direct access to the white balance and hue settings. The first useful choice is the list of already worked out white balance settings, which should more or less match the descriptions on your camera’s menu (e.g., daylight, shade, tungsten, etc.). alternatively, or as an additional way of refining the effect, adjust the white balance slider and the hue slider, this latter is between red and green, and so works on a different axis to the bluish-yellowish scale of the white balance.
As mentioned above, for the exercises I have used the RAW file equivalent of the images used previously.
As before, the first image is of the bridge in Ulsan, taken during the white balance exercises in the Digital Image Qualities module and using the shade option in-camera in order to see the results of the different uses of white balance in our images.
This time, as I am using RAW data, when opening the image it first appeared in the RAW image converter, allowing me to make changes to the white balance. Changing this to AWB (the best choice available), I did not think that any further manipulation would be needed, but I felt that the image was still a little to red, so decided to make additional changes in the Levels application.
These changes were minor, in fact I kept the red channel at 1.00, but changed the green channel to 1.03 and the blue channel to 1.06. This slight enhancement has balanced out the red’s producing an image identical to the original AWB photo taken during the exercise.
As there is far more information contained in RAW files over JPEG’s, there is little change to the histogram and although slight combing is present in the channels I have altered, when refreshing this, the original graph returned to its original state.
I took my second image, of the flower, using the correct white balance setting for the conditions, but find the colour cast within the petals a little too orange, not matching any of the other images taken on the day.
As I had needed to make additional alterations to my previous shot, I expected that once I had selected a new white balance setting for this image, additional manipulation in Photoshop would be called for. However, to my surprise, when changing the WB to Auto, everything aligned and the image looked as good as, or if not better than my comparison (seen earlier).
The other difference here is the foliage in the background of the shot, as this has lightened too, making this a very pleasing image to look at. There was no difference in the histogram at all.
Finally, I looked at the battlements shot, and as I had carried out so much work on the JPEG version of this image, expected the same to apply with the RAW file.
As with the other RAW files, my first task was to change the white balance, which I did, converting it to auto – again, this was the best representation for the image, although the fluorescent setting came a very close second. I then recovered the highlights and opened the image in Photoshop as again I was not happy with the colour cast and wanted to make a few final tweaks.
In the Levels application, I made the following adjustments to each individual colour channel; the red channel was changed to 1.04 and the blue channel to 0.88, I did not feel that changing the green channel made any difference to the shot. So with all of this completed, I was left with the image below.
As with the previous histogram’s, no change was seen in the graph, apart from slight combing, which was once again rectified by refreshing the information.
This turned out to be a very intense exercise, taking me much longer than I anticipated, but I feel that I have learnt some valuable lessons.
Of the two JPEG processes I used, arranging the RGB sliders in the Levels application produced the best results, although I did find that too much manipulation produced heavily combed histograms, indicating degeneration of pixels. I was also surprised at how easy it was to manipulate JPEG images, as I always thought this would be nearing impossible, but with a little knowledge and the right skills, it confirms that anything is possible. Although, we need to be careful, and aware of the use of our manipulated images, as too much manipulation means the degeneration of data resulting in not being able to use the shots.
Changing the RAW data was much easier in this exercise as it was only really the white balance that needed altering, with minor additional tweaks, just to get things right. RAW, uncompressed data is far more flexible than compressed JPEG files, and if necessary, we can manipulate this data much further to achieve our results.
Although this project was intense, I have really enjoyed doing the work and I have learnt much in the world of Photoshop.