The Color245 On-Line User's Manual

• What Color245 Does
• Color245 Basics in a Jiffy
• Color Balance Your Monitor
• How to Shoot Precise color Images
• Hints and Tips for Color Imagers
• Color245 Tutorial

---

What Color245 Does

Color245 allows you to resize, equalize, and precisely co-register three source images in microscopically fine steps. In addition, Color245 has tools to fine-tune the color balance of the celestial object you have imaged for a precise, true-color representation. Try the sample images to see how it works.

Color245 does not change the source files that it works from. This means that you can experiment freely without fear of corrupting your original image data. When you load an RGB image set, Color245 automatically creates three "generic" color files with the names RED.FTS, GRN.FTS, and BLU.FTS, and only these are altered during the creation of a color image.

---

Color245 Basics in a Jiffy

Follow these six steps to create a color image:

• Load your RGB images. Use the Load RGB (LR) command to specify the red, green, and blue-filtered source images. For optimum results, you should prepare these images for compositing using CB245 or Multi245 software. If for any reason you want to reexamine the images, use the View RGB (VR) command.
• Resample the RGB images. Images should be resampled to "square-pixel" dimensions. Use the Resample RGB (RS) command to resize the images. Images taken in the Cookbook 245's 252-wide mode should be resampled to 320x240, and images in 378-wide mode should be resampled to 378x284 pixels. Finally, focus-mode images with the camera in 378-wide mode should be resampled to 215x240 pixels for maximum resolution.
• Resample the RGB Images. The sources images may have been taken with a CCD with rectangular pixels. In 252-wide mode, the Cookbook 245 has 25 x 19-micron pixels; in 378-wide mode, the pixels are 17 x 19 microns. Since most software assume that images have square pixels, the images should be resampled to a square-pixel format. To resample, use the RS command.
• Equalize the RGB Images. The source images may have different exposures, but the Equalize Color (EC) command adjusts them to the exact same specifications. For deep-sky images, a low point of .01 and a high point of .995 usually work well. For planetary images, use .001 and .9999. If the gamma value is greater than 1, the images will be brighter; below 1 the images become darker. For the most accurate color, set the gamma value to 1.0. Note that equalizing automatically creates good color balance for most images.
• Register the RGB images. You can bring the three color images into precise alignment by two methods: blinking and centroiding. Blink Registration (BR) blinks the red and blue images against the green images; you simply press the arrow keys until there is no motion left. For Centroid Registration (CR), you select a star and the images are precisely aligned to subpixel accuracy.
• Adjust the Color. The Adjust Color (AC) and View Color (VC) functions provide two different "windows" on your color image. The View Color image is bright and contrasty with a limited color range, and the Adjust Color view is soft but a had a long and delicate scale of colors well suited for creating a precisely balanced color image. Between them, you can tell how your color image looks, and, if necessary, perform corrective steps to change the color or brightness of the image.
• Save the Color Image. Using the Save Color (SC) command, enter the file name for the output image and approve the saturation and gamma correction parameters. Color245 creates your choice of a 24-bit color TIFF or a 24-bit Windows Bitmap (BMP) file.

---

Color Balance Your Monitor

Test the color balance of your monitor by giving the name of one image for all three colors. The screen display should appear in shades of neutral gray. If the image has a green, blue, magenta, or other color cast, adjust the color balance of the monitor so that you see neutral gray image. If these controls are internal, have a technician adjust the monitor for you.

Most monitors distort the brightness that they display, with the middle values appearing too dark. Using the Set Display (SD) function, you can correct this distortion. For most monitors, a monitor gamma value of 1.9 does an excellent job.

---

How to Make Precise Color Images for Color245

Although tricolor imaging is simple in theory, it is difficult to put into practice. For realistic results, you must use a filter set that closely approximates the spectral response of the display phosphors on your computer monitor and the tricolor sensors in the human eye. Wratten filters #25 red, #58 green, and #47 blue have long been considered the standard filter set for photographic tricolor work.

• Recently Edmund Scientific has introduced red, green, and blue dichroic color separation filters with ~90% transmission that do an excellent job at a very reasonable cost ($60). Because the Edmund filters transmit more light than the Wratten filters, exposure times are considerably shorter with Edmund dichroics than with Wratten filters.
• However, the high infrared sensitivity of the TC245 CCD chip means that neither Edmund nor Wratten filters cannot be used by themselves make accurate color images. This is because standard blue, green, and red filters transmit a great deal of near infrared light. CCD tricolor images shot with standard filters come out with a strong purple cast.
• To block the infrared, you must add an extra filter. The Corion NR-400 infrared rejection filter does an excellent job. Murnaghan also offers an IR exclusion filter. Both are expensive, but because their transmission is 90% or more across the visible spectrum, these filters cause little loss of light. Used with an infrared rejection filter, both the Wratten or Edmund filters give accurate CCD color rendition of daytime scenes, planetary images, and stellar images.
• With the Cookbook CCD camera, you can shoot one or more images of the object through each color filter combination, that is, through the red plus IR rejection, the green plus IR rejection, and the blue plus IR rejection. For true color, an IR rejection filter must always be used. You will soon discover that the TC245 has low sensitivity to blue light. To make a blue image with the same pixel values, you need to integrate 5 to 8 times as long in blue as in red. For this reason, multiple imaging stack-and-track software like Multi245 is recommended. Blue integrations up to one hour are quite feasible.
• On the same night you shoot your images, obtain a set of dark frames, and if possible, make flat field frames also. In an ideal world, you would shoot separate flat fields for each of the colors because the color sensitivity of the chip may vary over its surface. If because of time constraints you can shoot only one flat field, shoot it through the green filter.
• Calibrate each of the images. If you are using CB245 or Multi245, be sure to use Dark Matching to obtain the lowest possible noise level. Good calibration is important to avoid sky backgrounds that change in color.
• If you have shot multiple frames in each color, use Multi245 to register all of the images to a common star, then average the images taken in each color. This will give you red, green, and blue tricolor frames with very high signal to noise ratios and already in perfect register.
• For the best color, avoid the non-linear scaling laws. Use stretch or linear scaling even though this usually entails some loss of highlight detail. Non-linear scaling laws have high gain at low PVs and low gain at high PVs, so you get intensely colored noise in the sky background and weak or washed-out color in the celestial object, which is the opposite of what you want.
• The color-related functions ask for a saturation parameter. It is tempting to set this to some value other than 1 (normal color), but resist if you can. Strong saturation usually looks fake. If you must enhance the saturation, stick to values under ~1.4.

---

Hint, Tips, and Things to Try

The centroiding algorithm is pretty robust. With a little care, you can get it to centroid on planetary features, such as white ovals in Jupiter's atmosphere. Use the + and - keys to enlarge and reduce the size of the centroiding box. If the box is a tight fit to a bright highlight, it will center reliably on the highlight. Blinking is nice, but centroiding is best.

Color245 treats the green image as the master image. If you want to center an object in the color image, load the green image before registering the images and use the auxiliary translate (TR) command to center it. Use integer values for the shift. When you register the images, you can register the red and blue images to a properly placed green master.

If your images are rotated with respect to one another, determine the pixel coordinates where they fit best. Load RED.FTS and use the auxiliary Rotate (RO) command to rotate it around this point. Clockwise angles are positive. After rotating, save this image and then do the same for the BLU.FTS image. If you are not certain how much rotation to apply, save RED.FTS with a different file name, and save the rotated version as RED.FTS. Determine the rotation angle by trail and error and then apply the total rotation to the original image.

Color245 is designed to make accurate, honest, astronomical color CCD images quick and easy to produce. The images resulting from Color245 are rich and full. By departing from the default color, saturation, and gamma settings, you can create exaggerated and false color images, but doing so is not the primary purpose or intent of Color245. If you stick with the default settings and are surprised by the softness and delicacy of color output from Color245, consider the possibility that you are seeing accurate color.

The images that Color245 produces are 24-bit color TIFF and BMP images that are 100% compatible with all major graphics, word-processing, and desktop publishing programs. Use any graphics program to convert them to GIF, JPEG, TGA, or other format of your choice. LViewPro is recommended as an outstanding shareware program for making file format conversions.

---

Color245 Tutorial

On the distribution diskette you will find sample images of Jupiter, Saturn, Mars, and M51, plus the three-image set used in this tutorial. The files are called TREERED.FTS, TREEGRN.FTS, and TREEBLU.FTS. The image shows a red pipe wrench and the pair of pliers with a blue handle duct-taped to the trunk of a big tree, so don't be too surprised when you see it. After you have experimented with this image, try the others. Don Parker took the planetary images and Jack Newton took the image of M51.

The three filtered tutorial images were taken with a Cookbook 245 in 252x242 mode, and they have already been calibrated. If you had taken the images yourself, you would calibrated them using CB245, AIP245, or Multi245.

• Step 1: Load the r-g-b set using the LR command. Enter the path name of the directory where the files are located and the file names TREERED.FTS, TREEGRN.FTS, and TREEBLU.FTS. The Each of the files will be loaded and immediately cached as RED.FTS, GRN.FTS, and BLU.FTS. Color245 does not access the original files again, so the original images are never altered.
• Step 2: If you want to, view the images using the VR command. Each file is loaded and displayed. VR always loads the files RED.FTS, GRN.FTS, and BLU.FTS. You will not need this command very often, but it's handy to have it when you want it. You will notice that the green and blue images are much darker than the red image. This is normal.
• Step 3: To correct the aspect ratio, resample the images to 320 x 240. You can enter any size you wish, but to display correctly, the sample and line pixel counts must be in a 4:3 ratio. 320x240 is the default for images taken in 252-wide (.PA/B) mode, and 378x284 is the default for images taken in 378-wide (.P1/2/3) mode, however, you may find that 378x284 requires too much memory to run. If that is the case, resample to a smaller size such as 320x240. Each of the three images will be resampled to the new size. Resample Cookbook 211 (.PIX) images to 256 x 256, and 378-wide-mode focus images (.PC) to 215x240.
• Step 4: Equalize the images. This simply means that you apply the same scaling law to all three images. The default values are .01, .999, and 1, but for this image, values of .01, .99, and 1 work better. This is because the large shiny spot on the duct tape distorts the high end of the histogram. It is best to keep the gamma value at 1 for maximum color fidelity. After equalization, the red, green, and blue images are nearly the same brightness.
• Step 5: Register the images by blinking. Use the cursor arrow keys to move the small image until the features match, then press the [end] key. The red and blue images are registered to the green image. For these images, translations of (1,0 ) and (1,1) are quite close. It is possible to enter any values you wish, but for sub-pixel registration, it is easier to center by centroiding.
• Step 6: Register the images by centroiding. There is a handy specular highlight at (90,96) on the green image. You can change the centroid radius by pressing the + and - keys. Center the cursor pixel on the highlight and press the S key. You will get a readout of the centroid coordinates and be asked if you wish to continue. Press [end] and then mark the same highlight on the red and blue images. Color245 shifts the red and blue images into precise sub-pixel alignment with the green image.
• Step 7: Adjust the color by entering baseline and contrast values for each of the three images and a saturation value for the color. The default baseline and contrast values of 0 and 1 work well for the tutorial image, but experiment to see what happens as you change them. Negative baseline values brighten a color and value over 1 increases the color's contrast and brightness. Positive baselines darken a color, and values under 1 decrease the color's contrast and darken it. The saturation value should normally be left at the default value of 1. You may wish to experiment with baseline and contrast values of -100 and 1.2, for example, and a saturation of 1.2. The image forms on the screen with staggered pixels in pure red, green, and blue. Squint slightly to blend and judge the colors. The finalize the color adjustment, press Y. To exit without changes, press the [esc] key. To check another set of baseline and contrast values, press the [enter] key. Color245 changes the images when you press Y.
• Step 8: View the color image using the VC command. This produces a version of the image using intense and somewhat posterized colors. Step across the room to see the effect. It is best to leave the saturation at 1, and the dithering at a low value such as the default of .03, but feel free to experiment with higher values to get grainy but more blended color renditions.
• Step 9: Save the image in color using the SC command. The saturation is normally left at 1 and at the default gamma of 1.8. Color245 can save the image as a TIFF file or as a Bitmap file depending on the extension you give it, that is, whether you type the name as TREECOL.TIF or TREECOL.BMP. Use whichever is more convenient for you. If your images consistently come out too dark, raise the gamma value a few tenths and try again, and if they are always too light, lower the gamma value a bit. Gamma compensates for the brightness distortion introduced by your computer's monitor.
• Step 10: The finished image is a 24-bit color image. Convert it to a GIF or JPEG image with a conversion utility such as LViewPro, import it into word-processing or graphic-arts software, or print it with one of the wonderful new low-cost ink-jet printers now on the market.

The best way to learn is to experiment. Since Color245 does not alter the original images, if you mess up something, you can always return to the original images and try again. When it comes to judging color balance, experience in the color darkroom is valuable, and you will appreciate the computer's speed and precision compared to paper and chemistry.