Tuesday, April 16, 2013

Contrast and White Balance problems in Photo Retouching

By Marvin Olson


When editing photos one should first consider white balance and contrast. White balance is normally the thing one ought to fix first, then contrast.The reason for correcting white balance first of all is that you can't correct color contrast if the image has a color cast.

White balance concerns the hue or tone of the illumination within the picture and sets white as a goal. White balance software attempts to normalize the tint of the light to neutral and in order to do that, the application usually needs some whites or grays in the picture to calibrate the suitable correction tint from. The whites can for example be a white wall or a sheet of paper or a dedicated white card. Gray cards are manufactured for the purpose of adding a neutral gray to the photo.

White balance software comes in two varieties: automatic and manual. The manual mode usually consist of a single temperature slider for adjusting the light cool or warm. This is OK for incandescent light, but not for fluorescent light or mixed light. When opening RAW images, one usually has a temperature slider. One can also have three color sliders for red, green and blue. Color sliders can somewhat correct fluorescent light and mixed light, but the problem with using color sliders is that the black and the whites usually get a bad tone. Automatic white balance correction usually needs neutrals in the image, like a gray card or white card or both. Some software can dispense with that, but usually neutrals are needed.

Contrast comes in three varieties: contrast of hue, brightness and saturation. Software usually has a single slider for contrast adjustment, which addresses all three at once. However, a single slider for all three is unsatisfactory since the result is usually over saturated and gaudy. Two sliders should be the minimum: one for luminance contrast and one for color contrast.

The normal way to manipulate contrast is simply by altering the difference between the individual red, green and blue values and the middle value (128); like this: R= (R-128) * contrast + 128; and likewise for green and blue. This method is not suitable for very dark or very pale images. What about very dark or very pale images? In that case you change the algorithm to: R=(R-RAverage)*contrast+RAverage where RAverage is the average red channel value of the image. And so on for G and B. The algorithms are essentially the same since a full brightness range image will have 128 as an average value.

Another problem with contrast adjustment is that not only may the average value not be 128, but the darkest and brightest areas may not be black and white. In that case one should be able to expand the brightest range to reach black and white. Levels adjustment is meant for this type of correction. If one's software does not offer the option to expand brightness range, one can do it with Photoshop's levels adjustment like this: First convert the image to Lab mode, select the L channel only and run auto levels on that. Then convert back to RGB mode.




About the Author:



No comments:

Post a Comment