9

COLOR CORRECTION: HOW TO GET IT RIGHT

How often do you need to color-correct your video? Chances are, every edit session. The same issues that plague video also affect still photography (especially digital photography). The bad news is that you won’t find a waveform monitor inside Photoshop. The good news is that all of your knowledge about fixing video signals will translate to Photoshop (with a little practice). If you think you’re SOL because you don’t have much experience “onlining” an edit, don’t worry, you’ll be up to speed quickly.

The Approach

When working in Photoshop, the most common type of color correction you will need to perform is scene-to-scene correction. That is to say you will need to bring a variety of shots, from a var iety of sources, closer together. They will need to look as if the same photographer shot them under similar lighting conditions. This is an extremely challenging task if you consider the likelihood that you will pull images from several different stock libraries, client-provided sources, and video frame grabs.

The key when starting out is to work on a copy of the image. This way you always have a copy to return to if something goes wrong. Open the image in question, then choose File>Save As…, and give the corrected version a new name. In fact, this is always a good idea. Image correction is often destructive editing, meaning that you cannot revert to the original state at a later date. Once the modified file is closed and saved, you lose the ability for multiple undos. By preserving an original version, or employing adjustment layers, nondestructive editing is possible.

Color Correction allows you to make important improvements to color and exposure within an image. The top image is the original image; the bottom image has been enhanced.

Choosing a Bit Depth for Color Correction

When fixing images, bit depth is an important consideration. While most Photoshop users are most comfortable working with 8-bit images, there is a strong case to switch to a 16-bit workflow. The extra bit depth gives you more image details (especially in areas such as shadows and highlights). The extra bit depth can also reduce banding and posterization in gradient areas (such as skies). Since most digital cameras and scanners now offer 16-bit modes, be sure to explore this option.

Mode	Pros	Cons
8-bit	More filters.Faster image processing.Greater compatibility with older video and motion graphic tools.	Banding in gradients.Image can show “wear” more quickly after application of filters or adjustments.
16-bit	Greater range for color and exposure.Allows for more processing of image without visible pixelization.	Larger file sizes.Not as many filters or adjustments available.Overkill for users working with 8-bit video formats.

Using the Histogram Palette

While Photoshop doesn’t offer the same scopes that you are used to, there is a useful palette that you can leave open. The Histogram palette shows you the data within your image (including the ability to view per-channel data).

How to Spot a Problem

Color correction, like editing, is a never-ending process. It’s done when the budget runs out, all the time is used up, or the client is satisfied. The end product could always be better if you only had more time/money/processing power. So understand that it is important to be able to spot problems quickly. What sort of problems? There are several. We will analyze the most important tools in Photoshop and evaluate which situations they might help you in. This evaluation will give you a better understanding of color correction.

Levels

There’s a very good reason this image adjustment comes first in the menu. You will need to make a Levels adjustment on every image. The Levels command allows you to correct tonal ranges and color-balance issues. That is to say, that you can fix poor exposure and adjust your white and black point. If you understand the need to white-balance a video camera, the Levels command will soon make sense.

Working with a Histogram

The key to understanding the Levels adjustment is the histogram. If you can learn to read this graph, it can serve as a visual guide for adjusting the image. To illustrate this powerful command, let’s try it on an image.

What about all those other buttons?

Curves

Curves aren’t easy, but they are powerful. The Levels adjustment just gives you three control points (highlights, midtones, and shadows). The Curves dialog gives you up to 16 control points, opening significantly more possibilities. One option is to add a single control point and pull it up to lighten the image or down to darken. The adjustment is applied evenly throughout the entire image. Multiple points can be employed for contrast adjustments based on tonal range.

Let’s take a look at Curves in action. I’ve chosen a grayscale file for the exercise because it is easier to work only on contrast issues when first using Curves. When you feel comfortable with grayscale images, switch to color ones, being sure to keep an eye on color shift.

What differentiates Curves is that you have precise control over what points get mapped, whereas in Levels, you do not. Additionally, a Curves adjustment uses a curved line (rather than Levels’ linear) to make adjustments that are eased through the image. It is possible to create adjustments that are easier to blend in with existing data, as opposed to Levels’ frequent problem of hard clipping.

So how do you know what points to pick? You click on the image itself (actually, +click on the Mac and +click for PCs). In the Curves dialog box, you can then move these new points up to lighten or down to darken.

You will notice as you drag on a point that the others react as well. Be careful, because as you add contrast in one area, it is removed from another. Radical adjustments will leave you with an undesirable posterization effect.

Because Curves are so complex, most people use a Curves adjustment layer instead of applying the adjustment directly to the source layer. An adjustment layer affects everything beneath it and has the added ability to support masks and blending modes. The results are just as effective as a trad itional adjustment, but significantly more flexible. We’ll look at adjustment layers in greater detail later in this chapter.

Color Balance

This is a simple yet useful tool. By using the Color Balance command, you can change the overall mixture of colors in a particular tonal range. This is quite useful for generalized color correction. When you think an image is too blue or red, a quick adjustment allows you to tone down specific colors. The adjustment is constrained to the shadows, midtones, or highlights as specified in the dialog box.

Hue/Saturation and Desaturate

These two commands are incredibly important for video graphics. If colors are too saturated, you will need to tone them down. Other wise, you will get “bleeding” onscreen with color data spreading from its original areas into other parts of your graphic. This can severely impact readability and is distracting to the viewer.

Oversaturation Test

How do you know an image is oversaturated? The best way is to import the graphics into your NLE and use features such as Final Cut Pro’s Range Check option. Better yet, put the image onto a vectorscope and see if the image exceeds the target values. If hardware scopes aren’t available, you can purchase software alternatives such as Scopo Gigio from http://www.metadma.com/.

These rich, saturated colors may cause us problems on a video monitor (Photo by James Ball.)

But if you are looking for a built-in option for Photoshop, you’ll need to settle for the NTSC Colors filter. While this filter is designed to fix saturation problems, it generally produces posterization and banding that make it an unacceptable fix. However, it can be used to check whether the image has issues.

Notice color shift in the red and yellow?

Isolate for Saturation Adjustment

You now need to create a selection matte to isolate the saturation adjustment. You can create this matte by using existing layer data.

Equalize will force the matte.

With experience, you will learn to spot saturation problems visually. You can, however, always use the NTSC Colors filter as a way to spot potential issues.

Exposure

If you are working with 32-bit images, you will turn to the Expos ure command for fixing your images. Generally referred to as High Dynamic Range (HDR), 32-bit images offer great flexibility in exposure. These images are well suited to re-creating the wide range of exposures found in outdoor scenes or intense lighting conditions. This image function usually exists in 32-bit space, and is said to be 32-bit floating point (often shortened to “float”).

HDR images are generally created in one of two ways. First, they can be generated with 3D software tools. The second method requires that the camera be secured firmly to a tripod and that you are careful when triggering or adjusting the camera to not move it. Several photos at various exposures are taken of the same scene (a minimum of three, but usually five to seven is adequate). The camera should have its auto-bracket and ISO features disabled. Each shot should be about two f-stops apart. The user then harnesses the Merge to HDR command (File>Automate>Merge to HDR…) to create the 32-bit image.

We created a HDR image in Chapter 2, so you can use that one. If you skipped that exercise, then open the sample Ch09_HDR1. psd from the chapter’s folder on the DVD-ROM.

Useful Commands in Brief

Channel Mixer

By using the Channel Mixer, current channels can be combined to form a new channel. This has two distinct purposes:

Let’s try the second scenario out and manually create a new grayscale image.

Gradient Map

The Gradient Map is an extremely useful effect for stylizing images. With that said, you should always choose to use the Gradient Map adjustment layer instead of the Image Adjustment command.

What does the Gradient Map do? The name says it all; it takes a gradient and then maps it based on luminance values. The effect is particularly useful for colorizing backgrounds and textures. When used on photos, however, it can create some dynamic looks, especially when used with blending modes.

After Effects users have an extremely similar effect too; in AE, it’s called Colorama. To see Gradient Maps in action, open the file Ch09_Gradient_Map.psd and turn on the individual gradient maps.

When Gradient Maps are used as adjustment layers, they can be masked and blended. All of the effects shown used a single Gradient Map. There is no rendering time involved, and the original color data is preserved. (Photo by James Ball.)

Invert

This image adjustment creates an image that is a direct inverse or negative. This is very useful for reversing grayscale images—for example, reversing the masked areas of a layer mask. It can also be used to make a positive from a scanned black-and-white or color negative. When an image is inverted, the brightness of each pixel is assigned the inverse value from the 256 color-values scale. This means that a 0 value would map to 255, while a 15 value would map to 240.

Equalize

If an image appears too dark or washed out, Equalize is the sledgehammer way to pound it into a usable state. The command attempts to redistribute pixels so that they are equally balanced across the entire range of brightness values. You can tell Photoshop to look at the entire image when equalizing, or sample a smaller area that will drive the overall adjustment. The command will take the lightest area and map it to pure white (255) and the darkest area to pure black (0). Since the equalize command uses the full 0–255 range, you will need to adjust levels to keep the image broadcast safe.

Threshold

This command is used to set the “continental divide.” All pixels on one side of the new midpoint are black; the others are white. This adjustment can be used to clean up a mask made from individual channels.

Not-So-Useful Tools

Brightness/Contrast

These controls are an inferior substitute for Levels and Curves, because the overall brightness or darkness is affected. Most of the time, the problem is in the midtones or gamma. A brightness and contrast adjustment will often leave your image washed out. Skip it. While this command has been improved in Photoshop CS3, it is still inferior to a proper Levels or Curves adjustment.

Replace Color and Selective Color

These are not the most useful commands. Replace Color combines the select color range interface with a fill command. I personally prefer to split this step out and have fine control. The Selective Color command is similar to Color Balance (but not as easy to use) and does not produce the high-quality results that are possible with Levels or Curves. You are better off using the individual Color Range command (Select> Color Range) and then creating a Levels or Curves adjustment layer.

Posterize

Posterization cuts down on the number of colors used, thus producing banding. Reduced color palettes and banding are both extremely poor for video use. Skip the Posterize adjustment.

Variations

Inexperienced users try Variations. This wizard feels a lot like a visit to the eye doctor with its “more-and-less” approach to image adjustments. While it is initially attractive, it will ultimately leave you wanting more. Don’t bother with this command.

Standard Definition Broadcast-Safe Concerns

Have you ever watched a low-budget commercial on late-night TV? Listen closely, as you may actually hear the graphics. If whites are too hot, the video track bleeds over into the audio track of a broadcast signal. This can also cause problems when tapes are duplicated. It is an accepted industry practice to keep video signals broadcast-safe.

Recently, creating broadcast-safe graphics has gotten trickier. Many video editing tools have begun to automatically adjust graphics to make them broadcast-safe during import. The system will automatically remap the white and black points to a broadcast-safe value. For example, both Premiere Pro and Final Cut Pro will properly interpret a zero black as 7.5 IRE on a Waveform monitor. Avid systems continue to offer you a choice and will ask if the graphics have RGB levels (0–255) or 601 Levels (16–235).

While this can take care of graphics where luminance is the only concern, it generally does not solve the problem of too much saturation. Whenever you adjust an image, you have to be concerned about modifying its colors to the point where they are no longer “broadcast-safe.”

It is important to read the documentation that comes with your video software (or see the guides that come in the Appendixes of this book). If you do need to remap your graphics, then it is better to make this adjustment in Photoshop.

If you need to clamp your graphic to 601 levels, then you’ll want to pull things into the 16–235 broadcast-safe range. In doing so, however, you don’t want to overdo it, so you must monitor the adjustment. By placing targets using the Color Sampler tool (stored in the same well as the Eyedropper and Measure tools), you can monitor the values of white and black.

In the example (Ch09_BSC.tif), I have placed targets on the clouds (for white) and the deepest shadows (for black). In the Info palette, it is clear that these colors are out of the safe range. Add a Levels adjustment layer and set the output levels to 16 and 235. The colors are now in the safe range. You can leave the adjustment layer floating and modify it at any time within Photoshop.

High-Definition Color Space

When creating graphics for high-definition video, you must keep in mind that SD and HD define colors differently. For standard definition, the International Telecommunication Union (ITU) 601 color space is used, while HD normally uses ITU 709. The most common problems occur when conversion occurs between SD and HD video sources. If the codec or video software doesn’t automatically convert the color space, then a slight shift in color will occur.

These color shifts are only visible on a properly calibrated HD monitor that supports the 709 color space. However, most users are not designing with these hooked up so they won’t see the issue.

Fortunately, it is rarely a problem as the color conversion process is fairly transparent and is generally handled internally by the video software application. For example, when Adobe After Effects converts from RGB to Y’CbCr in a video codec, the codec handles the conversion internally. When you export the same After Effects project to both an SD and an HD output, the process should work without issue.

Adjustment Layers

I’ve already mentioned adjustment layers as a helpful tool for pulling down Levels and for use in the Gradient Map feature. Adjustment Layers are the only true way to perform nondestructive editing on an image. The adjustment layer can be blended, masked, or deleted at any time. Additionally, if you double-click on the Adjustment Layer’s thumbnail, the Image Adjustment dialog box comes back. Nearly every image adjustment available in the Image>Adjustments menu and is offered as an Adjustment Layer (the notable exceptions are the Shadow/Highlights, Equalize, Match Color, and Replace Color commands).

Masking Adjustment Layers

If you want to apply an adjustment only to part of a layer, make an active selection as you would for any other change (see Chapter 4, “What About Transparency?”). Then, when you create the Adjustment Layer, it will automatically mask the adjustment to apply only to your selection. It is a good idea to feather the edges of your selection (Select>Modify>Feather) for a more believable adjustment. Otherwise, you will see a hard edge that makes the adjustment very easy to spot.

The sign can use a little “punching up.” (Photo by James Ball.)

The new image has a different “feel.” If I change my mind, editing is a double-click away.

Adjustment Layers can also have gradient masks applied. The Adjustment Layer automatically comes with a mask applied. Just click on its thumbnail (the big, empty white one), and you will see a border around the mask’s icon. You can now add a gradient or hand-painted mask in which to blend the adjustment. To see these two techniques applied, open up the file Ch09_AL_.psd.

Photo Filter Adjustment Layers

Adobe added to the “real-time” color-correction tools with the introduction of Photo Filter Adjustment Layers. These 20 different adjustments simulate using colored glass filters when the photo was taken. These “filters” can be used to adjust the color balance and color temperature of the light in the scene. Besides the built-in presets, you can also choose custom colors from the Photo Filter interface using the standard color picker.

The filter offers several choices, but they can be simplified into three groupings:

Let’s try adding a Photo Filter to an image. Open the file Ch09_Photo_Filter1.tif from the chapter’s folder on the DVD-ROM. To apply a Photo Filter, do the following:

Fixing Common Problems

Our next chapter, , “Repairing Damaged Photos,” will tackle more difficult photo damage. Here we will look at six common color problems that will affect you frequently.

Red Eye

I’m not talking about what you have after a late-night editing session. The fill flash bouncing off the back of the eyeball causes red eye. This is a common problem for photos taken with consumer-quality cameras, because the flash and lens are very close together. Dark rooms only aggravate the problem, because the pupils are open wider. If it’s your camera, look for a red eye reduction mode; it will strobe the flash, thus cutting down on the red eye. You can also use a secondary flash with a sync cable if you are using a professional camera.

However, chances are you didn’t take the pictures. Unless the photos are from a visit to the Inferno, you will need to clean up this problem. The discoloration won’t necessarily be red, just look for glowing eyeballs that shouldn’t be there.

Color Cast

Sometimes the light cast conditions are universally off. This could be seen as a blue cast throughout the entire shot. Maybe the wrong filter was used, or the image was not white-balanced. Sometimes the problem is simply at the developing end. (When you go to a one-hour photo center at your local superstore, chances are the machine is on autopilot and is being run by someone with little or no experience.)

Fixing Shadows and Highlights

One of the most flexible image adjustment commands is called Shadow/Highlight. The purpose of the adjustment is to help salvage images where the subject is silhouetted due to strong backlight. It can also be used to improve subjects who have been washed out by the camera’s flashbulb. The command does not just lighten or darken an image; rather it makes adjustments based on neighboring pixels.

When first opened the tool is too basic, offering just two sliders. However, clicking on the Show More Options box opens this tool up significantly. Let’s give it a try:

While this command simplifies the process of fixing shadow/highlight problems, it is not available as an Adjustment Layer. Therefore, it is a destructive editing command, which means you should work with a duplicate file or layer so you can return to the original photo if needed.

Skin Tones

It seems like old photos tend to have problems with skin tones. I look back at childhood photos from the 1970s and 1980s, and I know that my skin was never that red. Improvements in film quality and processing have cut down on these color shifts, but the proliferation of digital cameras has made this problem all too prevalent. Picture fading can cause a photo to lighten or darken over time. Generally, this is an even distortion, however, so a slight adjustment is all that is needed.

Tinting a Photo

Imagine that you are working on a project showing different people, and all of your sources are full color, except one. What do you do? Dump it? Look for a tornado to carry you over the rainbow? No, you regress a bit and color it.

This technique is simple, but often overlooked.

To see this technique open the image Ch09_Recolor.psd from the chapter’s folder.

Can you spot the fake? The top picture is actually a hand-colored grayscale version. The original photo is on the lower right. (Photo by James Ball.)

Hands On | 09

Enhancing Still Images for Video—with Jayse Hansen

As an editor or motion graphic artist, you will often be sent still images to include in your video. These images will often be rather uninspiring and may not fit the overall mood of your video. By harnessing some basic techniques, images can quickly be made to look more professional and meaningful using Photoshop.

For example, you might be tasked with creating a somber biographical video about an individual. The client hands you a box of mismatched photos. Some are color, some are black and white, and some are an awful orange color from the seventies. Simply adding these photos “as-is” to your video would make the end result look messy, inconsistent and amateur.

Therefore, in order to make the video piece have a uniform quality—we want to adjust the photos so they all have a similar look to them. Because this hypothetical piece is somber, we will make the photos a classic sepia-toned. Keep in mind, however, that this same process can be used to create wildly different looks and variations.

You’ll find Exercise 09 on the DVD-ROM.