Color Targets for Digital Imaging

Timothy J. Vitale
October 2003

This is an update from a Cons DistList posting on Color Target Recommendations, a few months ago.

Discussion

Many color targets are available for use in digital imaging. I recommend the 24-patch ColorChecker (if using polarized light, see below). The Q-60, IT8.7, target series have their value, but they are not consistent enough for digital image calibration; they do have their uses, however, see below. The Kodak Q-13 and Q-14 don't have a consistent density values for my use, but they are "good enough" for film.

Many digital imaging professionals use the GretagMacbeth 24-patch ColorChecker, a 8" x 11.5" color target with its 1.625" sq matte-surface patches, consistent from target to target, costing about $60-70. The value of the ColorChecker is that the 6-patch gray scale is always neutral gray, and each patch is always the same density from target to target.

The Kodak Q-60, IT8.7, color target has 240 color patches with an additional 24-step gray scale (including Dmin and Dmax) and an image with flesh tones. The Q-60 grays are never neutral and do not have a consistent colorcast within the gray scale; they can't be used to neutralize grays. One can download measured XYZ and Lab values from a Kodak FTP site for each target <ftp://ftp.kodak.com/GASTDS/Q60DATA>. The "Y" values for the gray scale are used to calculate transmittance (T) and density (D) values.

The Kodak Q-13 and Q-14 have 19-step gray scales of different sizes. Robin Myers finds the targets acceptable for digital photography. See <http://www.betterlight.com/pdf/whitePaper/wp_gray_cards.pdf>, for an excellent review of gray cards by Robin Myers.

There is a smaller version of the ColorChecker that costs about $80-100; contact Robin Myers if your going to order one in a few weeks, he is getting together a group order at a lower cost.

The digital version, ColorCheckerDC, still has some problems, so it is best to say away for a while. The DC has many more patches, some are glossy to achieve a greater tonal range, but they can give serious reflection problems in some lighting situations. The DC costs roughly $300 ($420 with calibration). GretagMacbeth suggests using InCamera software ($150-300) to make ICC profile using the ColorCheckerDC or ColorChecker. I have done this with good results using the BetterLight scanning back. A new profile is required for each new lighting setup.

Many low-end Color Management tools, in the $300-700 range, often use the IT 8.7/2 target (there are several target manufacturers besides Kodak) because the values are measured with great accuracy and reported with Mean and Standard Deviation for each patch. The Kodak (IT 8.7/2) Q-60 R1 (reflective target on Ektacolor paper) versions cost roughly $65. Kodak (IT 8.7/1) Q-60 transparency targets E1, 4x5 Ektachrome, are currently $110 and Q60 E3, 35mm Ektachrome, are about $65; measured values are available at the same FTP site.

InCamera <http://www.picto.com/incamera/incameradetails.htm> uses the ColorChecker and/or ColorCheckerDC to make digital camera profiles. It is extremely simple to use.

See http://www.betterlight.com/pdf/whitePaper/wp_gray_cards.pdf for an excellent review of gray cards by Robin Myers.

Using the Q-60 Target

The gray scale in the Kodak Q60 R1 has a 2.5 Dmax, including the Dmin and Dmax patches at each end (before GS1 and after GS22). The E1 transparency target has a Dmax of 3.2. Ektachrome 100 film has exactly those properties, with an input range of densities from 0.1 to 2.5; it will output densities from 0.2 to 3.4 D. That is, if a Q-60 R1 print target is photographed with Ektachrome 100 film, the gray scale will have a Dmax of 3.2, if properly exposed and processed. Ektachrome 100 has a contrast enhancement factor of about 1.7; the shoulder and toe of the "density vs exposure curve" are not linear accounting for the difference in the value one can calculate from the slope above. Gray scale data from representative examples of the Kodak Q60 reflective and transparency targets are in <Macbeth Gamma & Q60 Data.xls>.

Using the Macbeth Colorchecker

Many use the ColorChecker gray scale to make sure that an image has been captured accurately. When each step of the captured gray scale is adjusted, using Photoshop >Image>Adjustments>Curves, to the RGB values of: 243/243/243, 201/201/201, 161/161/161, 122/122/122, 85/85/85 and 53/53/53 (+/- 1-2 RGB units), the gamma is uniformly 2.2.  These values yield a known density range of 1.45 D, with a Dmax of 1.5, at a gamma of 2.2: a very solid result.

These values were developed by Robin Myers (an Apple ColorSync color management software co-creator) <http://www.rmimaging.com/information/information_index.html>.  I used them extensively to make inkjet printed surrogates that match the originals remarkably well. It has been pointed out by Ben Blackwell and confirmed by Robin Myers that the newer 24-patch ColorChecker have variable results when used under polarized light. Robin Myers http://www.rmimaging.com/home.html can get the older ColorChecker when they are requested. The newer versions of the ColorChecker have gray patches that change density when viewed under polarized light; use ColorCheckers manufactured in 2000, and earlier.

Density of the Image

Conservators normally see artworks with a density range of 6 stops, or less. This is a dynamic range of about 1.8 D (1 stop equals 0.3 D).  If an artifact has deep darks with detail, suspect a greater than normal density range. 

Materials with known high dynamic range are: Q60 color targets, Kodachrome 25 slides, and Fuji Velvia 50/100 and Ektachrome 100/100GX, to name a few. Most slide films have a high tonal range, a Dmax of 3.0, up to 3.9 D. This produces colors that modern viewers find very acceptable when viewed as a transparency on a light box.

This built-in contrast enhancement does, however, yield back more-or-less the original tonal range of the image when the slide is bounced off a screen during projection, depending on the projector and screen material. A wall painted bright white has a low reflection compared to a projector screen; a gray wall reflects images that are truly awful. See the Da-Lite screen selection guide for a great deal of information on screens <http://www.da-lite.com/products/selecting.php>.

Using the Colorchecker Beyond its Inherent (1.45 D) Tonal Range

After you have captured the ColorChecker gray scale in the image and have corrected it to the above value check the darks for RGB values beyond 243. If you find that the image has darks beyond the 1.5 D "black" patch on the ColorChecker, you can determine the density of the darker areas using the RGB data below. Yes, there are only 12 RGB-values for the 1.5 - 4.0 density data to fit into; this is the problem when using 8-bit color with its 0-255 steps. The newest Photoshop, CS or Photoshop 8, uses 16-bit color with 65,538 steps. This is a significant improvement in the field of digital imaging.

Adjust the ColorChecker gray scale to the RGB values above. Then use the eyedropper tool with the >Image>Adjustments>Curves plot. Place the eyedropper tool on the suspected dark area of the scanned image, hold down the left mouse button and read off the RGB value, in the Info window. Compare it to the table below to determine the density of the area in question. You can check you ability to read values with the eyedropper using <gretagmacbeth_colorchecker-rgb-cmy-dkblks_v1.tif> color target. The corresponding data for the gray scale is in <macbeth_gamma_and_q60_data.xls> on the tab named "Density vs RGB" and under the column (K) titled "Calculated RGB values Gamma 2.2 Density 4.0." Note that a density beyond 2.8 D can only be captured if the device has the capability, see below.

Using the following equation, the full set of Myers' gamma 2.2 values to 4.0 D, were generated in Excel <macbeth_gamma_and_q60_data.xls>

EQ1: R/100 x (1 / 2.2) x 255

Where: r=100/[anti-log10] of D; d=density (log scale); and r=reflectance (linear).

The 12 steps in the 0.01 to 4.0 density series are:

0.01 d=255 RGB
0.05 d=243*RGB
0.23 d=201*RGB
0.44 d=160*RGB
0.7 d=123*RGB
1.05 d=85*RGB
1.5 d=53*RGB
2.0 d=31 RGB
2.6 d=17 RGB
3.4 d=7 RGB
3.8 d=5 RGB
4.0 d=4 RGB

Data marked with a star (*) are the RGB steps from the six patches (0.05 D to 1.5 D) in the Macbeth ColorChecker gray scale mentioned above. The value and the calculations to derive them can be found in the Excel worksheet <macbeth_gamma_and_q60_data.xls>.

Enhanced Colorchecker (On-Screen Only)

I have made my own on-screen enhanced version of the ColorChecker, with the 6 additional white, gray and black patches, with higher and lower density values <gretagmacbeth_colorchecker-rgb-cmy-dkblks_v1.tif> When the new chart is viewed in >Image>Adjustments>Levels>, the histogram shows data from far left (3 RGB) to the far right (255 RGB), almost the full range of 8-bit (255 + 0 steps) color. When the 12-patch data is viewed using >Image>Adjustments>Curves, and the eyedropper cursor is clicked on each white, gray or black patch, its presents a momentary circle on the "Curves" plot line, showing the RGB value in the boxes showing input and output RGB values for that point on the plot. This is proof that the full range of 0.0 to 4.0 D can be measure on the Photoshop, 0-255 Level/Curves scale.

Go to the enhanced color target created in Photoshop <gretagmacbeth_colorchecker-rgb-cmy-dkblks_v1.tif>It must remain a TIFF file or its colors will be altered by JPEG compression. Your working color space should be set to <Adobe RGB (1998)> in the >Edit>Color Settings>Working Space>RGB> or the color patches RGB values will be changed. Do not accept a change in the embedded color space.

Note that when the above target is printed on the Epson 9600 (or similar device), its gray values can't go much beyond 2.1 D. The image will have to be up-sized to about 180-360 ppi in Photoshop >Image>Image Size>Document Size>Resolution> for printing. Maximum density will depend on paper type, black ink used (PK or MK) and ICC profile <http://prographics.epson.com> used.

Color target

The densities of all the gray patches are marked. The color target's middle row, BGR and YMC, have my addition of the fully saturated BGRYMC patches just above the original (as printed and scanned) Macbeth version.

Dmax of Capture Device

Note that a density beyond 2.8 D can only be captured if the device has the capability. Some examples for today's digital devices are listed below with links.

Betterlight Scanning Backs <http://www.betterlight.com/products4X5.asp> can capture a density range or 3.3 (11 stops), across a full 0.0-4.0 D range <http://www.betterlight.com/downloads/Manuals-Tutorials/tonecurve_samples5.pdf>.

High-end Epson scanners have a Dmax of 3.6 <http://www.epson.com/cgi-bin/Store/consumer/consDetail.jsp?BV_usebvcookie=yes&oid=17065>. Modern low-end scanners generally have a Dmax of 3.0-3.2.

Nikon Coolscan 4000ED, a 35mm+ film scanner, is reported to have a range of 4.2 D, which I find hard to believe <http://www.nikonusa.com/usa_product/product.jsp?cat=1&grp=98&productnr=9282>.

B&W Film Step Wedge

Using another type of photographic target, the Kodak 1A Step Tablet (B&W film, 0.75" x 1.75"), CAT 152 3380, the actual response of a scanner can be evaluated. The 1A has 11 steps, ranging from 0.05 to 3.05 D. My 1A step wedge has the following values: 0.05, 0.35, 0.65, 1.25, 1.55, 1.85, 2.15, 2.50, 2.80, and 3.05. Exposing B&W film to a slightly de-focused gray card, a 1/3-stops steps, will yield frames with a range of uniform densities. An appropriate film frame can be added to the package to yield density values greater than 3.05; a 0.5 D film will shift all the values up 0.5D, e.g., 0.55, 0.85, 1.05... to ...3.55 D. This will facilitate the evaluation of scanners claiming a Dmax beyond the range of the step wedge. The densities of the film frames can be calculated using >Image>Adjustments>Curves and the eyedropper tool, after calibration with some known, such as the 1A step wedge.

Problem With the Reds

Remember that all scanners and digital cameras have problems with highly saturated reds. This is because they have CCD's with high infrared sensitivity. Infrared filters are used to cut off the 750-nm+ portion of the spectral data, but they also catch some of the red data as well. To get accurate results, the red values must be read with a colorimeter, such and GretagMacbeth Eye-One, and adjusted to the correct RGB/Lab values using the >Image>Adjustments>Curves.


[Search all CoOL documents]


URL: http://cool.conservation-us.org/byauth/vitale/color-targets/index.html
Timestamp: Thursday, 12-Nov-2009 11:27:24 PST
Retrieved: Sunday, 22-Oct-2017 23:07:06 GMT