Processing Images with Photoshop ______________________________________________________________________________ This "How-To" was aquired for another source on the internet (I don't remeber where). Align and Combine Images I recommend using your astro imaging software (e.g., CCDSoft v5, MaxIm DL, etc.) to do your alignment and combining. These programs usually support sub-pixel alignment, and Photoshop does not. [Note: You can simulate sub-pixel alignment in Photoshop by enlarging the images 2x, 3x, or 4x, and then performing the alignment. After alignment, downsize to the original dimensions.] 1. Perform data reduction (dark subtraction, flat fielding). 2. Look for flaws (remaining hot/cold pixels; dark spots from cosmic ray hits on the dark frame; bright spots (false stars, arcs, lines) from cosmic ray hits on the images); bright/dark columns and rows. Fix these before you move on. 3. Align all images with each other. I recommend aligning all individual reduced images before you do any combining. This gets all of the alignment over at one time. Otherwise you have to align the individual images you use to build you four master images, and then you have to align your master images. Every alignment step introduces a small amount of softness into your images. 4. Determine what combination method to use for making the four master images. If there are a lot of hazards (airplanes, meteors, cosmic ray hits, etc.), use a median combine. If the images are clean, or can be cleaned up with pixel editing, sum or average will give you better S/N than median combine. 5. Create the four master images. 6. Combine the R, G, and B images into an RGB images (usually in MaxIm DL or Photoshop). 7. Optional: Remove gradients from the combined images now using automatic tools in Mira, MaxIm DL, etc. If you are using Photoshop for most of your processing, and the automatic tools won't work well enough to suit your requirements, hold off on gradient removal. 8. RGB combine using MaxIm DL or Photoshop. Check point: you now have two files for export into Photoshop: an RGB file in 16-bit TIFF format, and a Luminance image in 16-bit TIFF or FITS format. [Note: if you are only doing an RGB combine, no L layer, you only have the one file. Skip steps relating to the luminance layer below]. Histogram Adjustments in Photoshop Perform the following histogram adjustment steps on both RGB and L images in Photoshop or your favorite image editor. 9. Adjust Levels by lowering the white point so that you are just able to see the brightest areas of nebulosity or galactic detail. (Note: depending on the image, you may want to lower the white point further. The lower you set the white point in this step, the greater the overall contrast in your image. The higher you leave the white point in this step, the lower the contrast but the greater the range of brightness levels represented in your final image.) While you are lowering the white point, raise the black point as needed if the background gets bright. You may raise the black point at any time during processing if the background seems bright. Just don't go so far as to eliminate any of your dim details! Repeat this step several times if you find yourself lowering the white point by a large amount, to avoid coarse adjustments. 10. Use Curves to generally brighten the midrange details. You have two basic ways you can use curves at this point (with a few variations that I will ignore until I can write a tutorial on them <g>): drag a point upward from the left half of the curve, which will brighten dim areas at the risk of burning out the bright areas; drag a point upward from the right half of the curve, which will brighten the bright areas while keeping dim areas fairly dim. The first method is good on images that have very good signal to noise ratio, or for the first 1-2 times you apply curves to an image that needs a lot of curves adjustment. If necessary, flatten the right-hand portion of the curve to prevent it from clipping (see text of book re: clipping). This brightens dim details without making stars over-sized or bloated. Following use of this technique, your last Curves adjustment should be used to emphasize the bright areas appropriately, since you have been suppressing brightening them. This technique favors bringing out dim details when there is a very bright core to cope with. If necessary, raise the black point as you adjust curves to keep the background fairly dark gray. [NOTE: Stop short of full application of curves! You want to have room left for additional curves adjustments a little later on. But don't leave too much adjustment for later, or the following conversion to 8-bits per channel might add objectionable noise.] Check point: convert to 8-bits per channel. Image Corrections and Conversions 11. Gradient removal 12. If the image does not have a large object or much nebulosity: Remove gradients by making a duplicate of the image layer, then apply Dust & Scratches filter with a radius of 5-15 to remove most stars, then clone to remove remaining stars/nebulosity/galaxies. Set opacity to 75-95%, and blend mode to Difference. Poof! Gradients are gone. 13. If the image has a large object or extensive nebulosity, you'll need to use one of the manual techniques outlined in the book to remove gradients. 14. [Optional] If necessary, using the Channels palette, fine tune the black point for the RGB channels. If you used the 4a gradient removal method, your background should be in excellent shape as a result. If you did manual (or no) gradient removal, then raise the black point of each channel to make it a very dark gray. Use your eye to get all channels as close to the same shade of gray as possible.5. Save these two images. Check point: You now have two 8-bits per channel images that look pretty good, one RGB and one L. 15. Convert the L image to RGB. 16. Flatten the L image, and save it under a new name. I suggest <objectName>_LRGB.PSD, for example, to show that this will be an LRGB image. (Use RGB if no L!) Create the LRGB Image 17. Flatten the RGB image, then copy and paste the RGB image into the L image. Make sure the color image is above the L image. Close the RGB image without saving the flattening!!! You only flattened it to combine the two layers into a single layer. Check point: Verify alignment of the two images by temporarily setting the opacity of the RGB layer to ~50% 18. Set the blend mode of the RGB layer to Color. The result is an image that takes all of its color information from the RGB layer, and all of it's luminance detail from the L layer. Tune the LRGB Image 19. Observe the quality of the color. If it looks grainy, apply a moderate Gaussian blur to the entire RGB layer. Moderate would be a radius from 0.8 to 2.0. If the image is very noisy, use an even larger radius. Note that the larger your radius, the greater the reduction in color saturation. You are trading color saturation for freedom from color noise. 20. Temporarily hide the RGB layer. 21. Examine the L layer to determine which brightness levels contain noise, and which do not. Select the noisy brightness levels (usually the background, but may include a portion of the dim details in nebula and galaxy images) using the Color Range tool, and apply a very light Gaussian blur. I like 0.3 or 0.4 radius for this. For extremely noisy backgrounds, apply a 0.6 radius. 22. Examine the L layer to identify the brightness levels that contain very good S/N. This is not simply the inverse of the selection in step 11. You may want to omit stars from this selection, as you could wind up creating halos around them when you sharpen. Use Color Range, and select the bright areas but not the white areas; this will help you avoid selecting stars if you choose to do so. Apply an unsharp mask to the selection, usually at 50%, a radius of 1.0 to 2.2, and 0 for threshold. See text for more on use of this tool. Check point: Show the RGB layer. Look at your image and decide how you might want to adjust the histogram. 23. Make minor histogram adjustments, using levels and curves, to finalize the appearance of the image. Use levels to raise the black point so you have a very dark gray background (or as dark as the S/N permits). Use Curves to brighten the brightest nebula detail to give you satisfactory overall contrast. Use Curves to emphasize any portion of the brightness range that needs it. Use Curves to hide noisy areas that remain after any of the above. Perform a final Levels adjustment to set the black point precisely where you want it.