If you are of the opinion that whatever comes out of your camera is "it", the finished product, than this article is intended to help you to see beyond this limitation and open up possibilities the computer and image modification software can offer. No longer do we have to live with whatever the local drug store gives you for photographic quality. No longer to we have to live with the limitations of our cameras and film. No longer do we have to be satisfied with what we see, suffering the disappointment knowing what we have for a print isn't that wonderful scene we saw in reality.
Carried further, the computer and programs like Photoshop can open up many new possibilities of artistic expression. There are many ways we can alter our images to get the absolute perfect look we want. Since pictorial images is a form of communications, our computers allow us to say things in this exciting graphic language in ways that were never before possible.
Many of us have a need for technical perfection. Print reproduction, for example, places exacting demands on the tonal range of our picture which must be matched to the printing process. For colors, many of the wonderful colors we see on screen simply cannot be printed on a press. We cannot adjust the saturation to our heart's content and expect them to print correctly. In some cases it may even be necessary to desaturate colors a bit to insure they are within the correct range of printable colors.
In this article we explore the concepts of a number of useful and important computer adjustments. Is the list complete? Of course not! Since digital photographs are actually data files, our computer programs manipulate this data in an infinite number of ways. There are many programmers out there striving to find new useful tools that will allow us to produce even better pictures, both technically and artistically.
What is beyond the range of this article are the many interesting and useful local adjustments. This category may include changing the color of a car, removing a telephone pole or other task within an image. The controls discussed in this article are intended for an overall adjustment of the entire image. Now that you've read this limitation, realize it is only a confinement for this article. Each of these controls can and should be used (when necessary or useful) on selected parts of your image.
The digital photo as input
If all conditions at the scene you are photographing were absolutely perfect, your equipment could respond perfectly and you were able to balance conditions to equipment perfectly, we would probably have absolutely perfect pictures every time. But more often than not our pictures are less than perfect.
In the old days with chemical film we learned to balance the quantity of light to the sensitivity and processing of the film. The results would be our image. Upon careful analysis, it becomes apparent that there are many factors involved, we will call them variables, that must be controlled in order to obtain an absolutely perfect image. More often than not this process becomes a balancing trick. What camera settings can be used to compensate for the problems at the scene?
But digital cameras see our scenes slightly differently than chemical film. Worse, we lack the standardization that came with things like exposure times, f-stops and consistent off-the-shelf pre-manufactured film. It may be that each camera brand/model may have its own set of capabilities and limitations that you will have to learn to compensate for before you can get that perfect picture. Further, different cameras of the same model/brand may also be slightly different so advice that holds true for camera A may not be exactly correct for camera B. There is so much to learn about this new way of capturing images.
Only on rare occasions will all conditions at the scene be controlled perfectly and the image be perfect. For the rest of the images, we can turn to the computer for help after the camera exposure has been made. Is this totally new? No, photographers for years have had two opportunities to obtain image perfection. First, they had the opportunity to capture the original scene as best as possible given the variables of the camera and conditions. But they also had the opportunity to work in the darkroom with control over the printing process to compensate for less-then-perfect images. So in the final end, with digital photography we have two opportunities to obtain perfection.
We could take advantage of this two-step process. We could snap away shooting anything that moves (or doesn't) and then depend on the computer to turn each of them into works of art. But no, because real photographers know better that they should try and maximize each step to the best of their abilities and to the extent of the equipment. If an exposure is made that needs almost no computer modifications, all the better. If a great scene needs a lot of computer work to salvage it, then be thankful we have the ability to do so!
Much has been written about the control of photographic variables and in time much will be written about the control of digital imaging. Philosophies will emerge and eventually we will have proven techniques and equipment available. Already we have a lot of digital tools available and we should learn to use them. For one thing is certain, we cannot control our exposures perfectly at the scene and therefore, our input images cannot be considered final artistically perfect exposures but merely the start of the process. And since the variables at the scene will, well, vary, we may be surprised what kind of input challenges we face. We therefore will conclude that we should be prepared that each and every image we photograph will be different and may require different computer modifications to make them perfect. The fact that variables at the scene will produce different images should be our first lesson in computer modifications and also the greatest justification.
Corrections & modifications
All the techniques and tools can be used for either corrections or modifications. Let's set some basic definitions so we can keep track of our terminology. Although not necessarily universally accepted, they will serve our purpose here. Corrections are adjustments necessary to fix problems introduced by the digital camera such as color shifts or over-underexposure. These types of problems ruin generally your pictures but computer magic may "correct" them.
Modifications are new techniques, available to us thanks to computer magic, that allow us to extend our images beyond a level of basic acceptability. Although they may be the same tool, such as sharpen or curve, the perspective is that we are basically changing the nature of the digital picture, hopefully for the better. Modifications do not have to be done whereas corrections must be made or else the image is unusable. Therefore, what separates the two is nothing more than your artistic expression and technical requirements. It is even possible to do both at about the same time. Move your curve and you are making a tonal correction, but move it a bit more and you are making a modification enhancing the contrast more than the original scene. But the results is a more pleasing picture and we are satisfied.
But not everybody is pleased. Although corrections can make an image mirror what the scene actually was, modifications obviously allow you to portray the scene as it probably wasn't. Is this dishonest? Or is this creativity? Can our modifications actually "fix" the original scene to what it should have been? For example, contrast adjustment can make a cloudy picture look as if it were taken in sunlight. Saturation is also relative to the amount of sunlight in the original scene. If it were cloudy the colors will appear desaturated. A quick Photoshop fix is like waiting for a bright sunny day. Of course the extreme can be pushed where the colors and contrast are, well, a bit on the unrealistic side. It is at this point we may have trouble for the image may be too good to be believed. On the other hand, maybe you want it that way.
As for corrections and modifications, the issue is largely philosophical dependent on your knowledge of the technical requirements of your output process, the conditions at the original scene and your artistic ability to improve a picture without making it unbelievable. As you experiment you will obviously make some mistakes but you should strive to learn by them. Push the limits of the tools and see what they can do- and what you shouldn't do!
Tone range control
One of the most important adjustment task is to adjust the tone range, setting the lightest and darkest points of your picture. This is very useful for pictures that lack the full range of tones or are generally very dark or very light. A useful technique is to expand the tone range stretching the range of tones. Another need for tone range control is to maximize printability of your pictures. The printing process requires a range of tones from white to black. We can stretch or compress the tone range as is required by our printing presses or even ink jet or dye sub printers.
The difficulty of setting highlight (lightest part of your picture) and shadow (darkest part) is that not all pictures can and should have whites and blacks. We cannot always assume that something should be perfectly white while something else should be perfectly black. Everything else between the two extremes must be adjusted and that is the contrast of the picture. Now if our picture contains a lot of black shadow areas, we may wish to lighten some of them to improve near shadow details. If there are a lot of near whites, we may wish to darken some of them to bring out near shadow details. But by darkening near highlights and lightening near shadows we reduce the overall contrast of the picture making it appear dull and flat. But maybe that's what the original scene looked like (or should be made to look like).
So where do you draw the line? When do you let some shadows go solid black and some highlights go pure white? Or, for that matter, should you let the range of tones extend right to the end of the scale or stop slightly short? A careful study of the information the picture is trying to communicate will help.
Photoshop Curves is the best tool for setting the highlight and shadow and can be done at the same time as doing the contrast modification (next section) allowing you to balance all the adjustments at the same time. Find the lightest highlight and match it up with the required highlight density. Do the same for the shadow and your tone range is set.
Photoshop Levels is another useful tool developed from the time old tested method and concept of 3-point control. With this you can set the highlight, shadow and midtone. The main problem with levels is that you don't have control over specific midtone densities as you can with curves. Levels is a great tool for making technical corrections and a general modification to the midtones. Highlights and shadows need to be adjusted perfectly for correct printing and levels will do this by setting the triangles in the dialog box to the requirements of your printing process. You also have the advantage of viewing the histogram as a guide for making your adjustments.
Brightness/Contrast should be avoided at all costs! This very old tool is simply too primitive for serious work. Think of it as curves where the line is always straight or levels where there are no useful triangles to match to a histogram. Very primitive and a perfect example of how much progress we have made in the past decade or two.
At far left is part of the Levels dialog box showing the histogram of the flower picture (shown at the beginning of this article). It shows that there really isn't much of a highlight or shadow. And why not? There are no white or black parts in the entire picture. But with adjustment we could stretch out the scale a bit making the highlights even lighter and the shadows even darker.
To set the highlight and shadow simply move the black and white triangles to the edge of the "mountain" as shown above. All the densities to the right of the white triangle will be pure white and all the densities to the left of the black triangle will be solid black. What about all the pixels that are beyond the mountain? Nothing, because there were no pixels of these densities in the picture they cannot show up as perfectly white or solid black. Note that if there were pixels of these densities we will see them as black or white. Not very good!
At right, above, is the picture after clicking OK in the Levels dialog box. Compare this with the illustration at the start of this article and notice that there are lighter and darker shades giving us a much greater range of tones.
At left we see the histogram after the levels adjustment. It seems we could have moved the triangles closer to the mountain and get an even better range. But is it really necessary? After all, there aren't any white or black areas in our original scene so should we put them in? For technical printing reasons we may want an expanded range, but artistically we do have an improved image. Further, note the differences between the input and output histograms. Notice that the mountain is stretched out redistributing the tones along a wider tonal range. This is what this adjustment is all about.
Contrast modification
A very important task is to adjust the contrast range of your input image. This can make a flat picture look more lifelike and bright. This can improve any image taken on a dull sunny day or even with a cover of clouds. Where tone range adjusts the highlight and shadow, contrast modification affects the midtones. Should they be made lighter or darker? It is the midtones that hold the key to the overall contrast of the picture.
You should first study the Histogram before deciding what to do about your contrast. Should you strive to fix the profile of the histogram? No, use the information it presents to help guide your decisions. Because each picture is different (even the same scene) each histogram will be different. You really can't tell what the picture is by looking at the histogram. Ideally, a "bell curve" might be the most ideal mathematical representation but that rarely happens to real-world photographic images. A picture is a communications media and is intended to say something with pixels instead of words. You must learn to "read" the media value of the picture and identify with what it is trying to say. Your modifications should help bring the message out.
Photoshop Curves is the best tool for setting controlling your contrast. The curve works by allowing you to set the ratio between input and output densities merely by moving the curve line. For example, you could set the curve so that 40% input equals 30% output. This will make your midtones a bit lighter. At the same time, 5% input could be made to output at 12% slightly darkening the near highlights. For the shadows you might like to reduce 95% to 85% so that the near shadows can show more detail.
However you choose to set your curve, it must look like a curve. Pictures seem to look better when the curve has a gentle sweep. Odd looking curves may accent parts of your picture but other parts that contain the same densities may make it difficult for the viewer to recognize what the picture is! Not carefully done, curves can really mess up your pictures.
At far left is the histogram of the kid picture shown in the second section of this article. Here the tone range already stretches from white to black so we don't need to make highlight or shadow adjustments. But we could work with curves and improve the contrast.
At right is a curve in the Curves dialog box. The densities along the bottom are the output densities while the densities going upward on the left are the output densities. What we have done is made the lighter tones even lighter (the curve goes below a 45 degree line you start with (can't see it because it's been adjusted!). We also make the shadow areas a bit darker, shown by the top/right part of the curve is higher. For the midtones, the area in the center of the chart, we lower the curve a bit making the midtones a bit lighter. This improves the contrast a bit.
The result is shown at far left. At near right is the histogram after the adjustment. Notice that the tones are stretched out along the tonal range. But more of them seem to be moving toward the highlight. This is the result of the midtone part of the curve being lowered. Densities that would have otherwise been in the middle of the histogram are now pushed toward the right.
Is the picture better? Compare this with the version shown above and decide for yourself. Unlike highlight and shadow points for which their placement may be technical requirements, midtone adjustments are largely based on your artistic abilities. Should the midtones have been a little darker? Are the highlights too light or dark? are the shadows dark enough? It's up to you!
Color range (saturation)
Hue/Saturation/Lightness, HSL, is useful for working with your colors. Hue shifts the color wheel changing the tones. Not a difficult mathematical process, it is a bit more difficult to use. With careful use it can shift colors changing them. The best use for hue is to make a slight color shift to adjust for the color temperature of light. The entire color range can be made a bit warmer or cooler with the move of a simple triangle on the HSL dialog box.
Saturation is very strange but extremely useful for making color adjustments based on process or primary colors. Saturate allows to either increase or decrease the saturation because the starting point is in the middle of the scale. By using saturate with individual colors it is possible to increase saturation, for example, on the greens while decreasing the reds. This can allow for fine tuning the saturation levels of individual colors giving a great deal of control.
Lightness, at this time, has not proven to be very useful yet. Tonal ranges can be adjusted better with curves or levels.
At far left is the original picture. In the center is the Hue set to a setting of plus 20. At right is about the same area adjusted to minus 20. Zero is the input with no adjustment shown at far left and each scale extends to 180 (for degrees of a circle). Notice that the tones are shifted. In the center the grass and ground colors looks interesting but the face is not. The jeans also look unrealistically clean. At right the face looks too red and the grass a bit yellow. The jeans don't even look like jeans!
Shifting the hue is not something you would apply a lot of. Only a few numbers plus or minus might show an improvement. Otherwise, it should be left alone. The problem is that the color shift is uniform throughout the entire image. All colors will shift equally and although this may help some colors, it's probably going to ruin some other colors.
Saturation is a useful adjustment that makes modifications to the saturation level of your colors. At far left the saturation is set to plus 50; quite a bit! In this case we could easily present unrealistic color levels to our viewer. In the center is the saturation set to minus 50 which actually desaturates your colors.
Use saturation sparingly with a great deal of discretion. Each image will require its own level of saturation and some may not even need an adjustment at all. The results of your saturation adjustments should improve the colors. At right is the same approximate area where the saturation levels of individual colors are adjusted separately. It takes a lot longer but you can tailor the colors to suit the requirements of your image. In this image the reds and yellows were increased making the image warmer. Cyan and blue were desaturated slightly and green was increased slightly to fine-tune the grass color. Compare this with the original image, above. Not that much of a difference but hopefully an improvement. When you work with the saturation, you really can't over-saturate the colors making them unrealistic.
Sharpening
Let's get one thing straight here and now. Sharpening is actually localized contrast adjustment where adjacent pixels are adjusted making them darker and lighter to increase the contrast. Do not think of sharpening as a kind of focus enhancement for this is not what it does or can do. For that matter, input images that are out of focus are most difficult to sharpen because they lack natural contrast breaks for the computer to detect and enhance!
Many years ago, in film-based color separation techniques, a process was developed that used a kind of fuzzy contact image of the picture to enhance the contrast locally. This fuzzy image was called an Unsharp mask. The digital programmers merely developed a tool that synthesizes this effect. The result of using "Unsharp Mask" is an image that looks sharper because the contrast has been locally adjusted.
Confusing, no? Just believe that the computer can actually look into the digital file (nothing more than numbers) and figure out where colors change suddenly. Then it merely alters the densities and the result is an illusion of a sharper looking picture. Again, focus has nothing to do with it.
But how much do your sharpen? This is the question because it is all too easy to oversharpen your image. This is when the lighter side of the contrast break becomes pure white and the darker side is solid black. The result is two adjacent lines, black and white, called halos. But we're working in color, no? Oversharpening will darken and lighten the individual colors until they reach solid or white (absence of color).
Sharpening is an illusion and you have to apply enough to make the illusion of your image more convincing without giving the viewer something else to look at. Halos, resulting from oversharpening, will ruin the image because the viewer sees them and not the picture.
On a final note, it has been noted that the Sony FD-91 will apply sharpening to an image as it is being exposed. Often times, there is evidence of oversharpening right out of the camera! This is very distressing and hoped that this is not common with all digital cameras. Oversharpening, by the way, has also be detected on sample images from other digital cameras as well.
At left is a comparison of several amounts of sharpening applied to our flower image. From left to right are settings of 0%, 100%, 250% and 500%. How far can you go before the image starts falling apart showing evidence of oversharpening? The 500% clearly shows a lot of oversharpening and the 100% may not be enough for this subject. But the 250%? Is it a bit much? When you work with sharpening (Photoshop's Unsharp Mask filter, in this case), study the effects on all parts of your image, not just the main subject. As we add more sharpening we are enhancing out detail in other parts of the image as well, such as the leaves. If you concentrate on only one part of your image you may be ruining other parts!
Look at the illustrations above. Notice how the effect of sharpening can be easily confused as a focus enhancement. It does look like the results of a sharpen focus. But it really doesn't affect focus, just local contrast. Notice how the flower is enhanced too much showing oversharpening. But even at 500% the leaves in the background aren't all that noticeably oversharpened.
Color correction
Remember the old days when you used outdoor film indoors? Notice a color shift? Well, digital cameras have the same type of problem. They have to deal with the color temperature of the incoming light. If you're wrong, there will be severe color shifts! Instead of stocking different types of film, digital cameras allow you to set a "white point" which adjusts the camera for whatever temperature light you are currently in. This is a lot better than rewinding a half roll of film because you decide to move outdoors. If you forget to set the white point you can tell it wasn't done because you can review the exposure and this is obviously one of the things to study. If it is set wrong, delete the picture, set the white point and take the picture again.
But what if there is only a slight shift in color? It is possible to shift the entire spectrum a bit using Hue in Photoshop. Think of the color wheel and that Hue will simply rotate the wheel a few degrees. The result appears to correct minor shifts in color that you can use to compensate for this problem or even enhance your picture a bit if you like.
A more sophisticated method of color correcting is to use the individual channels of the Curves tool. In this way each of the channels can be adjusted independently giving a greater range of control. Further, the color model can be changed from RGB to CMYK allowing for color correction for print; a necessity because of serious technical issues.
After considering the color of this picture, it was discovered that it would look a bit better by reducing the blue. At near left the curve of the Blue Channel is raised up decreasing the amount of blue a bit. Also, on the RGB Channel the overall densities were adjusted to improve overall contrast. No other adjustments were made to this picture, especially hue or saturation.
Color correction can be rather difficult because it usually is done with technical objectives instead of artistic taste. But as a creative tool, working with the individual channels in the Curves dialog box can make uniform color adjustments such as this one. The ratios of input densities to output densities produces an even alteration that is pleasing to the eye.
Image size and cropping
Although these tasks are not tone or color adjustments, they are worth considering as adjustments. With cropping, we can select only a part of the image we want for our final picture eliminating unwanted image. When you crop, your picture will get smaller. Sure, smaller in size but more important, smaller in the number of image pixels. Let's say you crop only to the main subject in the center eliminating 3/4ths of all the original pixels, you really can't expect the final print to be the same size as the original. We then have to make the 1/4th remaining pixels enlarge to fill the print area. This is not good. One characteristic of digital photography is the lack of almost unlimited resolution to play with. With 35mm we could crop a bit and not see much of a difference. With digital, the slightest cropping may be noticed.
So how do you crop? One solution is to buy a digital camera that gives you higher resolutions than you need for your output system. In this way when using the entire image you always throw excess resolution away. But when you want only a selected area you can afford to throw some pixels away and still be at the required resolution. It will be your task to know how much resolution you can afford to throw away and how much you can crop before you are below the threshold of minimum resolution. The answers to these questions must be determined by studying the output system(s) and your input resolutions.
For example, the Sony FD91 allows for 640x480 and 1024x768 resolution. If the output resolution is for web, and for this webzine we generally use about 300x200 images or smaller for these pages, we obviously have a lot of excess resolution to work with. If, in the field, we cannot zoom in enough to capture the desired scene within the entire 640x480 frame the resolution can be changed to 1024x768. Then there are more pixels for the same optical view and that may be enough to crop to 300x200 pulling out only the desired image. The rest can be thrown away through cropping. From your perspective as viewer, you will never know what was outside this nice image!
But what if you don't want to throw image away by cropping? The 1024x768 image size is way too big for your screen and will take too long to download (which is why the roughly 300x200 was decided on for illustrations: a reasonable compromise). The solution is to use Image Size to reduce the resolution by combining tiny pixels into bigger pixels. The result will be less pixels, or, for that matter, 300 wide instead of 1024.
When you use image size you are obviously throwing resolution away. Not by cropping but by averaging small pixels into bigger ones. The resulting larger pixels contain less detail per image area. No surprise, there is less detail to the 300x200 image than in the 1024x768 image. That's OK because we only need 300x200. From our perspective we see a dramatic reduction in overall quality. Lots of fine details that were in the original image are averaged out of existence. This is very distressing but a necessary technical requirement. From your point of view, you really don't know what you're missing so this cannot be labeled an error. It is actually a required part of preparing an image for output: resolve the picture to the required output resolution.
But what if you want to increase resolution? Suppose we wanted to crop our 1024x768 image and take only the nicest 200x100 pixels? Cropping will obviously let us do this. But now we face the output requirements of 300x200 and realize our cropped image has too little resolution. We can use Image Size to increase the resolution as well by simply inputting 300x200 as the output resolution. This will force Photoshop (or other programs) to recalculate the 200 pixels wide and divide them into 300. The result? The required image is now at the proper output resolution.
But what about the image itself? Realize that we cannot put detail into our image, we can only mathematically average our pixels making them smaller. The same amount of detail remains after image size because we cannot invent image, only manipulate it. In fact, the resulting image will be fuzzier than the original because of the averaging process that may not follow the lines of image in our picture but instead a strict mathematical averaging. Our image quality will be less than when we started! It is in your better interest to avoid trying to "manufacture" image using image size. It doesn't work. You are better off going back to the original 1024x768 image and taking 300x200 original pixels and live with a bit of unnecessary image around your subject. Of course, another approach might be to alter your layout to accommodate a 200x100 instead of a 300x200.
At far left is a cropped 250x250 pixel section from the original 1024x768 image. The result is an image of unmodified pixels. At near left this image was cropped from the reduced image shown near the beginning of the article. Since this image was already reduced in resolution there were very few original pixels. By using Photoshop's Image Size the resolution was increased to match the cropped version at far left. But there is a difference! You really can't invent image with the Image Size dialog box. You can only mathematically manipulate their relative size. The results? Not very good. There is no substitute for real pixels from the camera with as little modifications as possible.
Conclusion
If, after reading this, you didn't find any answers, you're on the right track. There are no answers for what you should do with your images because each image is different. Each requires its own treatment to bring out the very best. Furthermore, your artistic and technical objectives will be different for each image. The result is that there can be no advice for what to do with your images. A careful study of the input image and being aware of the output requirements is essential for each and every image. This is really the fun part of working with digital images. Look at the picture and decide what tool or tools you will have to use to get the results you want and need.
Think about the concept that each and every image can and should be somewhat different. Your input is only as good as your equipment and is dependent on the conditions at the scene. When you open an image consider it the starting point and study it carefully. Could it use a sharpening? Is the tone range acceptable? Should the contrast be adjusted? Are the colors saturated enough or maybe too saturated? Does the image need an overall color correction? The list of possibilities goes on and on. Each image, being unique, will require its own set of adjustments. The next picture may require something entirely different. Your skill is to determine the needs and make the necessary adjustments to produce perfect pictures every time.
