Posted: Thu Oct 23, 2008 1:17 am
I'm a bit late in adding this information, but perhaps it will help. Two of the best authorities on sharpening are the late Bruce Fraser, and Dan Margulis.
Bruce Fraser wrote the book "Image Sharpening with Adobe Photoshop CS2" (published by Adobe and Peachpit Press). Although we are now up to Photoshop CS4, the information is still up to date. Basically, Fraser says you need to sharpen first for the source (camera, scanner, etc.), then for the image type (low, medium or high frequency), and finally for the output (monitor, printer, etc.). This is most easily done with actions that pause at various stages to allow you to tune the sharpening to the image, but can be run as a single action without user intervention. The one-pass sharpening, Fraser feels, doesn't work for most images. The book is a must for anyone serious about learning the theory of what and why we sharpen, and has a lot of detailed images showing the steps and results, along with why he chose the steps that he likes.
Dan Margulis, a fan of Mike's Curvemeister by the way, has a neat sharpening routine that I am currently using with great success. I have just built an action for 64-bit CS4 but it should work in 32-bit versions of CS3 as well. Dan's steps are as follows:
1) Load your image into Photoshop and make the color corrections and alterations. Duplicate the image (I flatten it before duplicating to avoid problems with the action).
2) I use the split screen in CS4, but it isn't necessary, as you can toggle back and forth. Convert the original image to Custom CMYK Profile and set the box to Heavy GCR, 70% black and 340% total ink. This bumps up the black a bit in the "K" channel. Click OK twice to generate the false separation.
3) Click on the black "K" channel, convert the image to Grayscale and discard the color CMY channels.
4) Invert the remaining "K" channel
5) Run Auto Levels on the "K" channel
6) Run Gaussian blur at 2.0 pixels on the "K" channel
7) Return to the original RGB image and duplicate the background layer
8) Add a mask to the new layer
9) Load the inverted black channel from the copy to the RGB mask.
10) Sharpen again as needed after cropping, then flatten.
Dan uses a heavy black mask made of the inverted and blurred black channel and feels it is superior to using masks created from the RGB channels or even the "L" channel in LAB. The mask effectively blocks any sharpening in non-edge areas such as sky where you don't want it. You can also tweak or adjust the mask density by using the sliders in the Levels window or by using a curve. The "K" channel in CMYK blocks sharpening in most mid-tone areas and is more effective at "finding edges" than Find Edges. Alternately, you can use Curvemeister to create the mask in the "K" channel and bump the contrast using a curve.
If anyone would like a copy of either of these action sets for Photoshop, let me know and I'll forward a copy to you.
Bruce Fraser wrote the book "Image Sharpening with Adobe Photoshop CS2" (published by Adobe and Peachpit Press). Although we are now up to Photoshop CS4, the information is still up to date. Basically, Fraser says you need to sharpen first for the source (camera, scanner, etc.), then for the image type (low, medium or high frequency), and finally for the output (monitor, printer, etc.). This is most easily done with actions that pause at various stages to allow you to tune the sharpening to the image, but can be run as a single action without user intervention. The one-pass sharpening, Fraser feels, doesn't work for most images. The book is a must for anyone serious about learning the theory of what and why we sharpen, and has a lot of detailed images showing the steps and results, along with why he chose the steps that he likes.
Dan Margulis, a fan of Mike's Curvemeister by the way, has a neat sharpening routine that I am currently using with great success. I have just built an action for 64-bit CS4 but it should work in 32-bit versions of CS3 as well. Dan's steps are as follows:
1) Load your image into Photoshop and make the color corrections and alterations. Duplicate the image (I flatten it before duplicating to avoid problems with the action).
2) I use the split screen in CS4, but it isn't necessary, as you can toggle back and forth. Convert the original image to Custom CMYK Profile and set the box to Heavy GCR, 70% black and 340% total ink. This bumps up the black a bit in the "K" channel. Click OK twice to generate the false separation.
3) Click on the black "K" channel, convert the image to Grayscale and discard the color CMY channels.
4) Invert the remaining "K" channel
5) Run Auto Levels on the "K" channel
6) Run Gaussian blur at 2.0 pixels on the "K" channel
7) Return to the original RGB image and duplicate the background layer
8) Add a mask to the new layer
9) Load the inverted black channel from the copy to the RGB mask.
10) Sharpen again as needed after cropping, then flatten.
Dan uses a heavy black mask made of the inverted and blurred black channel and feels it is superior to using masks created from the RGB channels or even the "L" channel in LAB. The mask effectively blocks any sharpening in non-edge areas such as sky where you don't want it. You can also tweak or adjust the mask density by using the sliders in the Levels window or by using a curve. The "K" channel in CMYK blocks sharpening in most mid-tone areas and is more effective at "finding edges" than Find Edges. Alternately, you can use Curvemeister to create the mask in the "K" channel and bump the contrast using a curve.
If anyone would like a copy of either of these action sets for Photoshop, let me know and I'll forward a copy to you.