For a long time now I have used an action for output sharpening, which would duplicate my image and convert it to lab mode, then use find edges, then levels, small amount of blur, levels, and then onto Unsharp mask. From here I would sharpen a little more than necesaary and accept, then duplicate this new sharpened layer back into my finished RGB document, adjust opacity to suit and flatten the image.
I have now started to run the action, but stop right before Unsharp mask, then save the active selection back into my original RGB document as a new channel. Load this new channel as a selection and create a new layer from it in Luminosity mode, then sharpen this.
So my question if you are still reading this, is this a new way of sharpening? Or do a lot of other people use this or a similar technique?
If you would like to try the action you can download it here, I had trouble modifying the original, so this new action creates two copies of your original document, uses one to create the selection, and the other to sharpen, finally you end up with just your original RGB image, with a new "Sharpened" layer.
lightbox wrote:
So my question if you are still reading this, is this a new way of sharpening? Or do a lot of other people use this or a similar technique?
I use a similar technique. I have an action that merges visible layers onto a new layer (without flattening), builds an edge mask on that layer, and changes blend mode to luminosity. Then all I have to do it apply G/blur to suit.
Basically with the Levels>Gaussian Blur>Levels dialogs, you are trying to eliminate any flat areas in your image that lack detail or don't require sharpening ie shadows, clouds and flat surfaces. What comes out black will be sharpened, and what you convert to white will not be sharpened.
So when the first Levels pops up, make a fairly decent adjustment to whiten all flat areas that lack detail, and bring the blacks up to reveal the edges you want to sharpen. Next a small Gaussian blur is used to fade the transition of the sharpened edges into the flat areas. Doing this however can loose a bit of contrast created in the previous Levels, so after the blur another levels is required to bring back any blacks lost, this is usually just a small adjustment compared with the first levels.
Focus Magic (which I have) is a totally different process - the above method uses unsharp mask which "sharpens" by increasing the contrast between adjacent pixels - it is not therefore sharpening as such - merely making it look sharper. (and who cares - so long as it looks OK )
Focus Magic however doesn't use the Unsharp Mask principle to sharpen an image. It uses advanced deconvolution to reverse the way in which the image got out of focus to restore the original "in focus" image as much as possible.
What you may ask is deconvolution? Well Convolution is a neighborhood operation in which each output pixel is a weighted sum of neighboring input pixels. The weights are defined by the convolution kernel. Image processing operations implemented with convolution include smoothing, sharpening, and edge enhancement. So deconvolution must be the opposite!
Being serious though, Focus Magic is not actually intended for general sharpening. However - I have to say I think it does it pretty well - although you need a quick computer if you use it on big files, especially at setting of 3 and above. The big difference with this and Focus Magic is that Focus Magic tends to sharpen all detail to the same extent - whereas this method finds the edges and does not sharpen other areas - where it might not be required. I sometimes run Focus Magic on a selection or on a layer with a mask to avoid or control this.
The technical skills you say you lack are not that technical - if you run his action and fiddle with the dialogue boxes that come up you should get a decent result.
What you may ask is deconvolution? Well Convolution is a neighborhood operation in which each output pixel is a weighted sum of neighboring input pixels. The weights are defined by the convolution kernel. Image processing operations implemented with convolution include smoothing, sharpening, and edge enhancement. So deconvolution must be the opposite!
Sorry for any inconvenience, but in my haste to get the previous action to work I had to build in a few work arounds. I have worked out my differences with apply image, and created a new action which is a lot slimmer, only creating one duplicate of your original image. Also it brings up Unsharp Mask as a final step in the action, so if you decide you want more or less sharpening, it will be one of your last history states. You can find the new action here. Also if you would like to sharpen anything in your image that may not get sharpened using this action, you can always duplicate your background layer, and sharpen this using your favourite method, keeping the edge sharpen above.
My action creates an edge masks uses a stamped visible layers layer converted to a smart object so smart sharpen can be used as a smart filter, In addtion to the edge mask created there is and overall reveal all layer mask added which can be painted on to remove some of the sharpening effect. My so called sharpening adjustment layer blending mode is set to luminosity to not effect colors and the layer's blend if gray is adjusted as to not blow out highlights. You may want to try my action. Hint over-sharpen then play with the adjustment layer paint on its layer mask adjust it opacity. Sharpening is very personal. I use this action on all my images and find it easy to adjust, but then I created the action.
Here is a bit of a right-up on the action. It will only work with CS3 for it employs a smart filter. The action is interactive however I have turned off some of the dialogs I normally have on. The one level dialog I left on is to allow you influence the edge mask. That level step is followed by two steps I usually make interactive a blur followed by a second level which feathers and darkens the edge mask more. The action stops with the mask loaded as a selection. This is to allow you to modify the mask best done using quick mask. Change the mask or leave it as is. If you used quick mask get out of the mode so once again the marching ants show then press play to continue the action. The Smart Sharpen Dialog will then show the selection will be active but hidden. Make the settings you want try over sharpening a bit. The layer is blended with the layers below and you can adjust the sharpening amount later by changing the layers opacity and or by painting with black on its reveal all layer mask. The Smart sharping filter can also be readjusted for I use CS3 Smart Filter Object. http://www.mouseprints.net/old/dpr/CS3_Smart_Sharpen_Filter_Adjustment_Layer.atn
Mr Mouse wrote:
My action creates an edge masks uses a stamped visible layers layer converted to a smart object so smart sharpen can be used as a smart filter, In addtion to the edge mask created there is and overall reveal all layer mask added which can be painted on to remove some of the sharpening effect. My so called sharpening adjustment layer blending mode is set to luminosity to not effect colors and the layer's blend if gray is adjusted as to not blow out highlights. You may want to try my action. Hint over-sharpen then play with the adjustment layer paint on its layer mask adjust it opacity. Sharpening is very personal. I use this action on all my images and find it easy to adjust, but then I created the action.
Here is a bit of a right-up on the action. It will only work with CS3 for it employs a smart filter. The action is interactive however I have turned off some of the dialogs I normally have on. The one level dialog I left on is to allow you influence the edge mask. That level step is followed by two steps I usually make interactive a blur followed by a second level which feathers and darkens the edge mask more. The action stops with the mask loaded as a selection. This is to allow you to modify the mask best done using quick mask. Change the mask or leave it as is. If you used quick mask get out of the mode so once again the marching ants show then press play to continue the action. The Smart Sharpen Dialog will then show the selection will be active but hidden. Make the settings you want try over sharpening a bit. The layer is blended with the layers below and you can adjust the sharpening amount later by changing the layers opacity and or by painting with black on its reveal all layer mask. The Smart sharping filter can also be readjusted for I use CS3 Smart Filter Object. http://www.mouseprints.net/old/dpr/CS3_Smart_Sharpen_Filter_Adjustment_Layer.atn
eros wrote:
This leaves the image in it's original size.
How can we resize for web with your action?
You should resize before you do your final sharpening. I usally use Image Size check resample and constrain, set the pixel size in the top section for the width or hight I want then set the interpolation to Bicubic Sharper when down sizing for the web and click OK. Often I find I do not need to sharpen the downsized image. If I need to sharpen I run my action then save a jpeg image for the web.
)