Scan Correct Color Smooth Texture Reduce Noise Remove Dust Repair Crop

There's a lot of technical stuff on this page, but don't worry... The rest of the site isn't as dry and dusty as this page is.

Common Terms

I use these terms repeatedly throughout my website. Most of the following links in this section go to wikipedia.

Cast: A cast is a tint of a particular color that affects the entire print, commonly seen in faded prints.

Clipping: Usually, overly bright or dark areas in the image that have lost some of their detail. Clipping is complex and deserves its own section. See below.

Color Depth: Color depth is also known as bit depth. See the Color Depth section below.

Color Space: An RGB color space is a color model used to accurately reproduce colors in images. The most popular color spaces for color images are sRGB and Adobe RGB. The most popular color space for grayscale images is Gray Gamma 2.2. See Color Standards below.

Color Profile: An ICC profile is a file or attachment for a specific device (such as a scanner or digital camera) or a standard color space. Epson scanners use an ICC profile called "EPSON SRGB". The sRGB color space uses an ICC profile called "sRGB IEC61966-2.1".

Gamma: Affects the brightness of an image. Most computer systems including Windows OS and Mac OS use a gamma of 2.2.

Histogram: A graph of the tones in an image, from the darkest on the left, to the lightest on the right. A sample rgb histogram is shown to the right.

Levels: Levels can be adjusted using a tool such as Adobe® Photoshop® Level Tool. A very good guide to using this tool can be found at Levels: Bright Lights, Dark Shadows, Contrasty-er Contrasts by Manggy on photomanggy.blogspot.com.

With the Levels Tool, you can make these adjustments: You can make these adjustments on the image as a whole, and you can also make these adjustments on individual colors. Adjusting levels for individual colors can help remove casts and correct color.

Color Depth

Color depth is also known as bit depth. Color depth is either the number of bits per pixel (bpp) of an image, or the number of bits per color (bits per channel, bits per component, or bpc) for each color channel.

An 8-bit grayscale image uses 8 bits per pixel, and a 16-bit grayscale image uses 16 bits per pixel. RGB color images use three color channels (Red, Green, and Blue), so a 24-bit color image (24bpp) uses 8 bits per color (8bpc). A 48-bit color image (48bpp) uses 16 bits per color (16bpc).

A 48-bit color image uses 65,536 shades of color for each channel. A 24-bit color image only uses 256 shades of color for each channel. There is an immense variation in color for 48-bit color images compared to 24-bit color images.

Because of the limited shades of color in 24-bit color images, making major edits to the image can be risky. Darkening the shadows, brightening the highlights, correcting the color, or increasing the contrast can cause posterization damage to your image. This damage is easily seen in the histograms, which will no longer look like smooth curves, but will have spikes and missing bars of colors. The damage can also be seen in the image. For example, blue skies might develop bands of color.

Posterization is explained in detail in Image Posterization on CambridgeInColor.com. Posterization can be prevented by using 48-bit color images.

Color Standards

Most of this section contains a few essential factoids about color spaces and a lot of links to Wikipedia and other expert sources. You can get much more in depth knowledge by clicking on any of the links below.

Gamma: Both color and grayscale images use gamma, which affects the overall brightness of an image. According to Gamma correction (standard gammas) on Wikipedia, most computers decode images (such as sRGB images) using a gamma of 2.2.

Grayscale Standards: According to the appendix Practical Color Management of "Preserving History" on archivehistory.jeksite.org (a website developed by Jim Kennedy that contains extensive information on archiving historical photographs and documents), the most common working color space for grayscale images is "Gray Gamma 2.2". He goes into detail about working color spaces in a section on his website titled Working Color Space.

RGB (Color) Standards: According to RGB color space (Applications) on Wikipedia, the most popular color space is sRGB although many graphic artists use Adobe RGB. In addition, according to sRGB (Usage) on Wikipedia, many cameras and scanners for consumers use the sRGB color space.

Windows Default Color Space: Windows has used default color space of sRGB for many years. According to an old article written in 1998 by Microsoft discussing color management and plans for making sRGB the default color space for Windows 98 and Windows 2000 can be found at Introduction to Color Management in Microsoft Windows Operating Systems.

Mac OS Default Color Space: According to a document titled Mac OS X v10.6: About gamma 2.2 on Apple's Support Website, Mac OS 10.6 Snow Leopard uses a default gamma of 2.2 while earlier versions of Mac OS use 1.8. Also, according to Lion 10.7 to sRGB Default by G. Bollard on gballard.net, Mac OS 10.7 Lion introduced a default color space of sRGB for color images.

During my testing and viewing of scanned images with Mac OS Preview, I found that an image without a profile will appear a bit brighter and have a slightly different histogram than the same image with a Gray Gamma 2.2 profile attached. This may indicate that the Mac OS default color space for grayscale images is not Gray Gamma 2.2. To avoid this issue, grayscale photos can be scanned in color using the sRGB color space.

Because sRGB has been the de facto standard for years, I suggest making color scans in the sRGB color space for both color and grayscale photos, and attaching the "sRGB IEC 61966-2.1" profile if possible. If you prefer making grayscale scans, I suggest attaching the "Gray Gamma 2.2" profile.

Types of Scans

These are the three main types of scans to choose from:
The choice of scan type depends on individual needs. For example, someone scanning historical photographs may want to scan photos in their present condition for archival purposes. Someone else who just wants to share some old 35mm family prints may want their scanner software to enhance and color correct during the scan for the sake of simplicity.

Whatever scan type is selected, I suggest archiving the scanned images and using working copies for editing.

You may wish to check out the links to the tests near the top of this page as well as Good and Best Practices for Making Digital Images by Jim Kennedy before deciding on the type of scan.

More details on each type of scan is below.


Raw Scan

A scanned image containing the raw RGB scan data. Neither a scanner profile nor an output color space are applied. No adjustments or enhancements or corrections or post-processing are performed. Clipping is never an issue.

The saved image file normally has a gamma of 1.0 (linear), but may differ depending on the scanner software and settings. The image will look very dark if the gamma is 1.0, otherwise it will just look dingy. In either case, the image will look unsaturated.

A longer description of raw files can be found in the Output | Raw File section of the VueScan manual on www.hamrick.com.

This type of scan is usually used by professionals, often when working with slides and negatives.


Straight Scan

Also called a natural scan. Both terms are common on the web and mean the same thing.

A scanner profile is applied for a straight scan. This converts the raw RGB scan data to true colors. An output color space is usually applied as well. This ensures that the colors stay the same when editing and saving the image with other applications.

The scanning software settings determine the exposure for the scan. The exposure controls the brightness and contrast for the scan. Proper exposure settings ideally result in an image that looks like the original photo in its current condition.

Other than exposure, no enhancements or corrections or post-processing are performed. Clipping is usually not an issue, although it is a possibility with some scanning software and/or out-of-gamut colors.

A straight scan represents a true representation of all the colors and tonal range of the original print. As such, it is a great starting point for color correction. It is also excellent for archiving historical photos.


Color Corrected Scan

"Color correction" and "exposure correction" are very common terms. When a scan is color corrected, it is usually exposure corrected as well. Anytime I talk about color correction on this website, I always mean both color and exposure correction.

To put it simply, an color and exposure corrected scan is a scanned image that had any color shifts fixed and color casts removed, and its brightness and contrast restored either from fading or from underexposure/overexposure of the photo at the time it was processed.

The image should ideally look as good as or better than the original photo when it was new, before it faded and changed color. Clipping can be an problem when using software to automatically correct color, though.

Color correction can be done during the scan or after. In some cases, the scanner software can perform better automatic color correction to images than image editors can. Epson Scan has an auto color correction feature that works very well, as I talk about in detail in the Color Correction Software section. If you prefer its results to other color correction software, you may wish to make color corrected scans.

Keep in mind that automatic color correction programs may work poorly with previously color corrected images. If you want the scanner program to do automatic color correction, I recommend making and archiving both straight scans and color corrected scans. This way, if you are unhappy with the scanning program's color correction, you always have a straight scan available for color correction with Photoshop or other image editor.

Most articles and guides I've read on the web recommend using an image editor rather than a scanning program to do color correction. However, color correction can be faster and in some cases better when it is done during the scan.

Clipping

Clipping happens when colors fall outside their normal range for various reasons. Clipping can be caused by selecting certain settings in scanning software, using automatic color correction software, increasing the contrast or brightness too much, and similar causes.

If an image is badly clipped, there will often be noticeable loss of detail in shadows and highlights. If the clipping is severe enough, there may be solid black areas in shadows (aka blocked shadows) or solid white areas in highlights (aka blown highlights).

If only a single color is clipped, it may not be noticeable, though there may be loss of detail in that color. An article that has a good explanation on when this can occur can be found at Camera Histograms: Luminosity & Color on CambridgeInColor.com.

If you see signs of clipping in a histogram, you can use the Levels Tool in Photoshop to check for clipping. Option-click on the left or right Input sliders to see the pixels that have been clipped.


Clipping Severity Scale

I've come up with a clipping severity scale, which I use in my scanning software and color correction software tests. The severity of clipping in an image is determined by the number of primary colors (red, green, and blue) clipped to level 0 (darkest color level) and level 255 (lightest color level) as shown on a histogram (in GraphicConverter). The total number of clipped colors is divided by the total number of pixels in an image and multiplied by 100, resulting in a number between 0 and 300.

The scale contains the following four levels of severity, based on the resulting number: Note that images with pure white areas from tears or white tape may look clipped. I do not report the clipping in such images. I also do not report micro clipping since this level of clipping is not significant and is extremely common with scans and color corrected images.


Minor Clipping Example

Color Corrected Motorcycle Shadow Clipping Highlight Clipping
This example is of a black motorcycle in front of a white wall. The original print (not shown) is a relatively recent photo with a slight yellow cast. The color corrected image (upper left) has a strong contrast with extremely dark shadows.


The luminosity histogram (lower left) shows a short bar on the left and a very short bar on the right. This may indicate a minor amount of pixels have been clipped to pure black and pure white. It may also indicate too high of a contrast.

The color histogram (left) shows a short bar on the left which indicates a tiny amount of pixels clipped to 0. Since the bar is mostly gray, many of the clipped pixels will be pure black. Shadow clipping (upper middle) shows a small scattering of clipped pixels including pure black. These clipped pixels correspond to the darkest part of the image.

The color histogram also shows a very short bar on the right which indicates a very tiny amount of pixels clipped to 255. Since the bar is mostly gray, many of the clipped pixels will be pure white. Highlight clipping (upper right) shows a very tiny amount of clipped pixels (including pure white pixels) in shiny chrome parts of the motorcycle.

This small scattering of clipped pixels is unlikely to be noticeable or affect restoration. However, the concentration of black and white clipped pixels may point to a contrast that is too high.

This photo rates 1.9, or minor, on the clipping severity scale.


Moderate Clipping Example

Color Corrected Xmas Bush Shadow Clipping Highlight Clipping
This example is of a Christmas bush with brightly colored presents underneath. The original print (not shown) is only lightly faded. The color corrected image (upper left) contains presents that appear to have somewhat intense colors.


The luminosity histogram (lower left) shows no bars at all on either the left or right, which means pixels probably have not been clipped to pure black or pure white.

The color histogram (left) shows a tall bar on the left that contains gray, blue, and magenta colors. This indicates that a moderate amount of pixels of different colors were clipped to 0. Shadow clipping (upper middle) shows a large amount of colorfully clipped pixels in the Christmas presents.

The color histogram also shows a very short bar on the right which is colored magenta. Highlight clipping (upper right) shows a very tiny amount of clipped pixels.

The significant amount of colorfully clipped pixels in the Christmas presents may point to overly enhanced colors in the image. Overly enhanced colors can look artificial, vivid, and flat.

This photo rates 5.7, or moderate, on the clipping severity scale.


Major Clipping Example

Color Corrected Creek Shadow Clipping Highlight Clipping
This example is of a small creek in a forest on an overcast day. The original print (not shown) is a relatively recent photo with good contrast and very subtle fading. The color corrected image (upper left) has a strong contrast with an extremely white sky.


The luminosity histogram (lower left) shows no bar on the left and a ridiculously tall bar on the right. This may indicate highlights have been blown out, meaning a lot of pixels may have been clipped to pure white.

The color histogram (left) shows almost no bar on the left. Shadow clipping (upper middle) shows an extremely tiny scattering of pixels.

The color histogram also shows a ridiculously tall bar on the right that is almost completely gray. This almost certainly indicates a large amount of highlight clipping. Highlight clipping (upper right) shows that most of the sky has turned pure white.

This pure white part of the sky has now permanently lost its tonal range. Even if the highlights are darkened, this large part of the sky will always be a solid block of a single shade of gray or white.


This photo rates 19.5, or major, on the clipping severity scale.