In the good old days, back in the darkrooms with photo paper, there was a black and white system called multigrade. Multigrade paper was available in six grades and each grade produced images with more or less contrast. Today software provides contrast adjustment, but not all contrast adjustments are equal or equally good.
What is contrast anyway? One (noun) definition of contrast could be: "The state of being strikingly different from something else, typically something in juxtaposition or close association." It can also be defined as (verb): "To set in opposition in order to show or emphasize differences." This means we in photos have three different kinds of contrast: contrast of luminance, contrast of color and contrast of saturation. Most software offers a single contrast slider that edits all three at once.
Software usually alters contrast by simply altering the difference between the middle value (128) and the individual red, green and blue values; like this: R= (R-128) * contrast + 128; and likewise for green and blue. But what if the image is generally dark or pale, then this method will fail. A better method is to use the average values of the images R,G and B channels, like this: R=(R-RAverage)*contrast + RAverage. And so on for G and B. The drawback of this algorithm, in both cases, is that photos tend to become overly saturated and colorful when raising contrast. Similarly contrast reduction makes the image look like you have placed a semi-transparent gray film over it.
Luminance contrast needs special attention. One can use the above algorithms after converting the image to luminance (L) values. The R, G and B channels are not equally bright, so luminance is not just the average of the three. Green is brighter than blue and red and red is brighter than blue. Therefore one usually uses a weighted conversion like this: L = R*0.299 + G*0.587 + B*0.144. This algorithm is used in the YIQ colorspace. Photoshop has a checkbox in their contrast adjustment panel called: "Use Legacy". The problem with this method is that the image becomes pulled towards the primaries R, G and B if you enhance contrast and pulled towards a uniform mid-gray if you reduce contrast. It is not very useful.
Lumiance contrast, however, can be enhanced in several ways. First you can simply expand the range: if Lmax is 200 and Lmin 50, then you can for example expand the range so Lmax is 255 and Lmin is 0. But what if Lmax and Lmin already are 255 and 0? Then one will have to pull the existing near mid tone L values towards the extremes. This leaves two situations: one will normally lose the mid tones that way and split the photo; in stead one can use dedicated software that can preserve mid tones while pulling L values towards the extremes.
There exists dedicated applications for luminance contrast adjustment. If one wants to do it in Photoshop, you will have to convert the photo to Lab mode, make the L channel alone active and adjust contrast for that. Remember not to use Legacy or the photo will tend towards either black and white or uniform gray. When adjusting contrast, remember not just to look at the extremes and the contrast in the image, but keep an eye out for the mid tones.
What is contrast anyway? One (noun) definition of contrast could be: "The state of being strikingly different from something else, typically something in juxtaposition or close association." It can also be defined as (verb): "To set in opposition in order to show or emphasize differences." This means we in photos have three different kinds of contrast: contrast of luminance, contrast of color and contrast of saturation. Most software offers a single contrast slider that edits all three at once.
Software usually alters contrast by simply altering the difference between the middle value (128) and the individual red, green and blue values; like this: R= (R-128) * contrast + 128; and likewise for green and blue. But what if the image is generally dark or pale, then this method will fail. A better method is to use the average values of the images R,G and B channels, like this: R=(R-RAverage)*contrast + RAverage. And so on for G and B. The drawback of this algorithm, in both cases, is that photos tend to become overly saturated and colorful when raising contrast. Similarly contrast reduction makes the image look like you have placed a semi-transparent gray film over it.
Luminance contrast needs special attention. One can use the above algorithms after converting the image to luminance (L) values. The R, G and B channels are not equally bright, so luminance is not just the average of the three. Green is brighter than blue and red and red is brighter than blue. Therefore one usually uses a weighted conversion like this: L = R*0.299 + G*0.587 + B*0.144. This algorithm is used in the YIQ colorspace. Photoshop has a checkbox in their contrast adjustment panel called: "Use Legacy". The problem with this method is that the image becomes pulled towards the primaries R, G and B if you enhance contrast and pulled towards a uniform mid-gray if you reduce contrast. It is not very useful.
Lumiance contrast, however, can be enhanced in several ways. First you can simply expand the range: if Lmax is 200 and Lmin 50, then you can for example expand the range so Lmax is 255 and Lmin is 0. But what if Lmax and Lmin already are 255 and 0? Then one will have to pull the existing near mid tone L values towards the extremes. This leaves two situations: one will normally lose the mid tones that way and split the photo; in stead one can use dedicated software that can preserve mid tones while pulling L values towards the extremes.
There exists dedicated applications for luminance contrast adjustment. If one wants to do it in Photoshop, you will have to convert the photo to Lab mode, make the L channel alone active and adjust contrast for that. Remember not to use Legacy or the photo will tend towards either black and white or uniform gray. When adjusting contrast, remember not just to look at the extremes and the contrast in the image, but keep an eye out for the mid tones.
About the Author:
This article is based on these two articles: contrast and contrast. Unique version for reprint here: Contrast in Photo Editing.
No comments:
Post a Comment