Bluegrass Bus Museum


You are here

Digital Color Management From Input to Output

Printer-friendly versionPrinter-friendly version
Cybergrass LearningWith the ever growing popularity of the Internet and the World-Wide-Web, many people are now incorporating digital camera and scanned images into their web pages. They are also being widely used for album covers and other promotional material. This is an interesting area as no two monitors or viewing environments look the same. That is, the appearance of an image on one computer doesn't look the same on another computer. Being that this is the case, how does one know that their image looks correct? Many people just scan in an image and assume that everything is OK. This is usually not the case. In this third Cybergrass part of the Cybergrass Education Series, we address how to get close to accurate color management with your digital color devices.

This article is a follow up to the earlier article on Photographing Bands and Artists. In that article I discussed some of the problems associated with creating a good original photo. In this article, I will discuss what to do with it to prepare it for presentation through the electronic media. I am assuming that you are starting with a quality photograph and that you desire to maintain that quality on the Internet's World-Wide-Web.

In order to create better images, it is necessary to first understand some of the techie stuff about imaging and images. You have probably already seen the effects of miscalibration and color errors but didn't know what was at play. Have you ever used outdoor film to take a photograph indoors? When you received your nice pictures back from the photo lab, you probably noticed that they all had an orange tint to them. This is especially true if you took the photograph in the evening or at night.

Just as it is impossible for any person to acoustically equalize a room by ear, it is also impossible for a person to visually equalize an image by eye. With sound or audio, one uses an instrument called a spectrum analyzer so that one sense (sight) can measure what the other sense (hearing) is actually doing. We need a similar method to deal with images. The problem is that with our 5 senses (sight, touch, hearing, taste, smell), we can't really use another sense for sight. To illustrate this, think about how you would describe a color, red for example, to a blind person. With no frame of reference, this would be impossible. Yet, with sound, one can show both the volume and the frequency on a spectrum analyzer. With photographs and images, this just isn't feasible so, we must use sight to measure sight.

When scanning an image it is extremely important to scan at a reasonably high resolution and with as many bits-per-pixel as your scanner will allow. Although the final image won't contain all this data, it is essential to have it in order to make some of the corrections which will be described later. I realize that this may mean your scanned image may be hundreds or thousands of kilobytes but, the resultant image produced for the web won't be this complex or large. We can reduce the file size later but we must first have as much information as possible to work with. I scan at 32 bits per pixel and 300 spots per inch which, I find to be very satisfactory for achieving the desired web image results which I desire.

There are many factors which come into play with both the original image and, the scanned resultant image. Skin-tones are probably the hardest to deal with. You surely don't want green or red faces yet, you also need to maintain some other qualities in the image you plan to present. What are the primary factors? Here is a list which defines some of them.

The brightness of an image is the overall light to dark ratio of the image. The brighter the image, the lighter the overall tones of the image. In extremely bright images, there may not be a solid black but, rather, a dark gray.
The contrast of an image is the amount of extreme range between light and dark. For example, at infinite contrast, everything is either white or black with no gray. With no contrast at all, everything would be a neutral gray. What one wants to achieve is an even progression of black to white with smooth grays in between.
Hue is the color balance of an image. Just as on a television set where you can adjust your color, the hue ranges from a form of red to a sense of green. Somewhere in the middle, the image becomes neutral.
Saturation and Intensity are closely related but have slightly different meanings. Saturation is the overall amount of color being applied. By turning up the red saturation of a pink image, the image may actually become red.
Intensity doesn't change the color's appearance as it is increased. that is, pinks do not become red. What they may do is become white or black but the foundation color remains pink.
This is a techie term to describe what is called a function. We use this because the correction this performs is not equal to all colors or shades of gray within an image. This function is a curve and not a straight line as the others mentioned above. The gamma function adjusts the variances between the two extremes (white and black) to achieve a non-linear balance and to correct for conditions which have ultimately affected the image.
Color Temperature
This term has to do with artificial lighting. Sunlight has a different color temperature than does a regular light bulb. Florescent lights have yet a different color temperature. This is why photographs taken indoors with out door film look orange. Because the color temperature from regular incandescent light bulbs is redder than is that generated by the sun. Florescent lights are more green as a rule. Computer monitors also have a color temperature. Humans don't notice this much because they are used to it and they correct for this automatically. Machines, on the other hand, do not.

Photographers who must use outdoor film indoors, employ a series of filters to compensate for the color temperature of the indoor lighting. They may use various intensities of blue filters on their camera lenses to correct for this difference and thus, the photos they get back from the photo lab, no longer have the orange tint to them. With digital imaging, we have the same ability to apply corrections after the fact. Both the digital camera and the image editing software have the ability to make the corrections -- the camera adjusts while taking the image and the software for after the image is taken. We can scan in an orange image and apply the blue filtering electronically to the digital image. We can also do many other adjustments.

Anybody may scan in an photo or drawing and make a digital image of the result. Rarely do the two look the same. If the photo and what appears on you computer monitor look the same in all respects, it probably still won't print properly. Thus, if one was to scan a photo and then print it on a color printer, the original photo and the print would not be the same.

The problem is that every step in the process, the human eye is the final judge and, unlike perfect pitch for music tuning, color isn't that easy. What is red or green or beige? Based on what? There are all kinds of red, green and beige. With music, you can say Concert Pitch A is 440 Hz and see it on a chromatic tuner. With color and light, you can do the same however, the instruments to do this are expensive and not really of much use outside the laboratory.

This is a problem just as with recording music. The band sounds great but once they have been recorded and then put on some media (cassette, CD, vinyl disk, etc.), they don't quite sound the same as before. The reason has to do with proper calibration of every step of the process. With audio, engineers and technicians have calibrated their equipment, applied proper equalization, and adjusted monitors to achieve an extremely close environment to what the source material sounds like. Because human ear and hearing is the judge, what you get depends on who's ear was making the judgement call. The same holds true with photographs.

With imaging however, this same care to detail is rarely applied outside of a graphics art shop of some sort. Thus, digital images generally do not have proper balance, gamma, color, etc. The primary reason is that many people don't know they can do something about it. Also, what is the "perfect" skin tone hue? A person's skin generally doesn't change color quickly but just walking from outside to inside, the hue changes. Turn on a florescent light and it changes again. The skin color didn't change however, the perception of the color is not the same.

In order to adjust digital images properly, some basic calibrations must first be established. Two of these are very easy. Solid black always has a digital value of 0 (zero) and white always has the maximum value allowed (255 for 8-bit or 65,535 for 16 bit). These are the number of steps between black and white. There is also a step for Red, Green and Blue which is where RGB gets its name. Obviously the more steps you have, the greater control you have in getting the desired result. Its what's in the middle that always gets messed up. Getting a particular color such as a skin tone requires precise adjustments of all 3 colors and the full range of steps for each.

If you ever watch a television during sign-off or before sign-on, you may often see a screen with various colors on it. Some may also have various levels of white to black. What you are seeing is a color chart or a calibration chart. The studio tries to get black as black as possible and white as white as possible and to have a linear transition from one end to the other. The same with each color.

If you look at an image on a screen, you may realize that there is something wrong with it but often, you may not be able to tell just what. The reason is that it is usually multiple elements which are off at the same time. Any or all of the above variables may be out of adjustment. What is needed is a method to put them all back where they belong. To do this, we use a series of standard color charts and calibration images.

Many printers and scanners and even some cameras have what is called an ICC profile. These are a file with the pre-defined sets of data that represent the characteristics of the device for either input or output. They are unique to the specific device. The standard is based on the International Color Consortium (ICC). These ICC profiles describe the color attributes of a your device to your computer or editing software to try and approach a proper balance.

Before we go off and start making random adjustments to images and equipment, it is absolutely necessary to first get our own environment calibrated as close as possible. To do this, we must employ some more techie stuff. The idea is to get what you see on your monitor to represent something close to "normal." You will probably never get perfect but you should be able to get close. Before proceeding, make a note of where everything is set before you start so that you can easily get back to where you started from.

Calibrating your Imaging System and Software

Color Temperature First, it is essential that we adjust our color temperature variable. Various monitors come with different phosphors in the picture tube which are used to display color. There are different tube manufacturers which incorporate different phosphor techniques and composition. Some of the more common are the Trinitron, Hitachi, and regular TV tubes called NTSC. There are others however most computer screens are either Hitachi or Trinitron varieties.

Even the new Plasma, LED and LCD monitors have options for color temperature. Generally, the factory defaults are good but it doesn't hurt to check and make sure you are properly set for your application.

Some monitors such as Samsung and Sony allow you to change the color temperature of your screen. Others are fixed at some value between 5000°K and 9300°K. These numbers represent a temperature on the Kelvin scale where absolute zero is the point where motion of molecules stops and has no energy at all. Think of these as the surface temperature of the Sun or the filament of a turned on light bulb. These are very hot temperatures. Most monitors usually have a value of either 6500°K or 7500°K. Your book on the monitor will usually indicate this. If your software has a place to select this value, you should do so. Many packages do not allow any correction for the color temperature so, if yours does not, you will have to live with what you have.

gamma To get us in the basic ball park of imaging, we must now calibrate both our scanner and our image editing tool with the appropriate corrective gamma function. Luckily, this is possible with carefully made calibration images made especially for this purpose. The images are presented below.

You may save this image to your disk by having your browser save the image. You do not want to ever edit these actual caqlibration images but, rather use them to determine the proper settings for your system and adjust your system or software tools. Make a note of these settings so that you can adjust your system whenever you need to. Most web browsers do not have the ability to adjust this imaging feature so you will need to correct your image prior to putting it in your web page. You will probably not be able to correct your display to show these images properly. Remember also that once you get the image looking proper on your screen, another screen may not present it properly.

gamma calibration image

If, in this image, the gray box in the middle of the frame next to 1.0 isn't the same, your gamma is off. Find the value where both the frame and the little box inside of it are the closest and you will have a very close approximation of your correction factor. Adjust both your scanning software and your image editing software so that inside box for the 1.0 value is the same level of gray as the frame around it. This corrects your gamma for all images. If you can save this permanently you will probably want to do so. If you cannot save it, write the corrective value down and always make this correction before any others. This achieves the proper gray balance for your image.

Once this adjustment has been made, the color gamma chart below should now be very close. It is exact when the bars inside each of the colored frames completely disappear. You may adjust each of Red, Green and Blue to fine tune your gamma settings and achieve the proper balance. By looking at this chart, you may notice that some colors are much closer than others. This is normal. Picture tubes usually operate with the three colors shown. This calibration chart is designed to be used with computer monitors.

Color gamma chart

At this point, the hard part is done and your system should be extremely close for correcting variances in your scanner, scanner software, and image editing software. You may want to bring up an image which you have scanned prior to making these adjustments and apply the proper corrective gamma to it and I believe you will see a noticeable improvement over the original image. You are now on your way to creating excellent images for the web!

Other Adjustments

Now that your system is adjusted properly, we can begin to make other adjustments to photos and scanned images. In order to get the most out of this part of this article, it is important that you first familiarize yourself with your image editing software. Almost all programs have the same functionality and features but they are not called the same thing nor are they in the same place. I will attempt to use generic industry terms for this article.

Each of the following editing functions may be applied to the entire image, a single color, or just the black & white portion. You will need to experiment with these to get the feel of what they do and when you will want to use each.

Brightness and Intensity have almost the same visual impact to your digital image. Increasing either of these will make your image appear brighter. Which or the combination you use depends upon the problem you are attempting to correct. If you have a photograph of a person and the skin tones appear correct but the color of their shirt appears faded, you may want to reduce the brightness and increase the intensity of the entire image. If the faces appear to be washed out but the clothes appear correct, you may want to decrease the intensity and increase the brightness. You will need to play with this to achieve the desired result.

Think of the brightness applying to the image if it is black and white only and the intensity as applying to colors only. In reality, white is comprised of all the colors so this isn't really a true statement but should help you decide which function to use and when.

Contrast and Saturation achieve a different role in image processing. Just as mentioned above where brightness is primarily a black and white function, contrast is also. Saturation is a color function just as intensity was above. What is generally desired is a smooth transition from black to white through each individual color as well as the grays. These are the functions used to adjust images which may require either an increase or decrease in the graduations. The gamma curves above did most of the work already however, there are times when an image needs a little help in this area.

If a photograph was shot in bright light and is over exposed, the ability to resolve features in faces or textures in fabrics may be lost. In other cases, the photo was taken in a shadow of a building or other object and the detail is also lost due to sharp transitions from light to dark. To some extent, we can recover from these errors by adjusting the contrast and/or saturation of the image.

Skin isn't solid white. With black and white images, especially with fair skinned people, it is easy to create an image where the skin tones are actually almost pure white. When this happens, the contrast which defines the fine features in the face are lost or washed out. It is possible to darken the image and then increase the contrast to attempt to bring these features back.

Here, the higher the number of bits per pixel, the easier this is to achieve as there is more digital information (number of levels and/or colors) to work with. As was mentioned in the beginning of this article. you need to scan with as many bits-per-pixel as possible. Here is why. With 256 colors, only about 8 levels of white are possible. The information required to increase contrast is lost and cannot be recovered. When scanning at 32 bits you end up with millions of colors and generally have over 65,000 levels of white. Thus, even though you may not be able to perceive the differences on your screen, they are within the digital data of the image and can be used to our advantage. We can increase the contrast to bring out these subtle differences in white levels and recover the data to the screen.

Unless your software will allow you to select a region to work with, you will probably need to adjust the entire image's contrast which may result in other parts of the image being altered in an undesirable way.

Colors may be individually adjusted by modifying the saturation level of the color or the entire image. When we increase the saturation, we are actually pouring more electronic ink into the picture. The result is richer and deeper color. By reducing the saturation, we electronically remove ink from the image. Thus, the color saturation impacts the amount of color and not the brightness of the color. It is with saturation that you can make faded colors come alive again. Just as was the case in bringing out subtle changes in texture or skin tones with contrast, we may also bring out subtle changes in colors by adjusting the saturation. Adjusting the blue may make it possible to return a sky to a bright blue from a gray blue yet retain the other qualities of the overall image.

Tone Maps, Equalization, and Histograms offer yet other ways to adjust an image. Sometimes you want to eliminate shadows, increase background detail or only alter a small region within the photos range of graduations without altering the entire image. You may want to apply a bit of digital Visine to get the red out of indoor photos. You need to shift the black or white a bit. There are numerous functions available for this type of control over your digital image.

The tool used is a function of what it is you're trying to achieve. Not all the tools are linear and many allow the creator to adjust just a portion of the Tonal Reproduction Curve (TRC) without modifying other aspects of the image. Changing of thresholds and color balance are all available with most commercial digital imaging editors today. Many allow you to save a color pallet or TRC file for use with other images.

The best way to improve your images is to play with them. Make a copy of one and play with it. There is a lot of flexibility in digital imaging to allow even the worst photograph to be made exciting and bright. This article just scratches the surface but should help get you on your way to making great images for the web.

Theme by Danetsoft and Danang Probo Sayekti inspired by Maksimer