Nature Photographers Network
This is the preserved archive of the original Nature Photographers Network. The community continues at naturephotographers.network.

Selective Sharpening Techniques

Discussion

14 replies
Posted by Scott Borger on January 25, 2007 at 2:22 PM
Selective Sharpening Techniques
I’ve been experimenting with selective sharpening techniques. At times, I use multiple layers for different parts of the images that I want to sharpen. For example, I create another layer for the sky and leave it unsharpened while I sharpen the rest. However, when you have a tree in the middle of the sky (see posted image), this does not work very well. I’m wondering what the best method for performing selective sharpening is (without creating a bunch of noise in the sky/highlights, etc). Also, what is the best overall sharpening software (and why you believe this) and is it a good value? What do you think about the CS3 bicubic sharpener when compared to other software? Thanks in advance for your replies.
Posted by Kalin Wilson on January 25, 2007 at 3:54 PM
Scott,
I recently started using Pixel Genius' Photokit Sharpener, http://www.pixelgenius.com/sharpener/index.html. There is an in depth review of it on Luminous Landscape. Also, there is a tutorial on the PG site.

I like it's approach to applying sharpening in different stages of image processing. Also, it offers a large variety of sharpening techniques that you can tailor to what you need. it uses sophisticated techniques that avoid color changes and you can control how the effects are applied because everything is done on a seperate layer. This is cool because the sharpening calculations are based on visible layers. So you can apply a sharpener, turn off it's layer, apply another, and compare by turning layers on and off. Then using layer blending and masks you can control just where you want the sharpening. I've had really good results with it.

Of course, a plugin like photokit sharpener isn't doing anything you can't do yourself. One technique you might try is converting to LAB and sharpening only the L channel. Most actual noise will be in the A and B (usually B) channels and this can actually be smoothed to remove the noise.
Another technique, that can be combined with the above, is to sharpen a seperate layer (copy or stamp) to the degree you want globally and fade the filter effect with luminosity blend, then use a layer mask to reveal the sharpening just where you want it (the tree).

I'm sure you'll get plenty of other suggestions. There are definately lots of ways to skin the sharpening chicken!

Kalin
Posted by bill meikle on January 25, 2007 at 3:59 PM
You can also select an area that you want to sharpen and just sharpen that part of the image, as you say lots of different ways to do it. I use PS CS3 and it does the basics well enough.

Bill

Posted by Brent Murphy on January 25, 2007 at 5:29 PM
I've used a number of sharpening techniques and plugins. Currently, I use two:
LMC54's Edge Sharpening - Simple, but useful. It uses the threshold tool to find the edges in the drawing, using a slider to determine how many edges it finds. Then it runs USM on that selection. Quick and easy to use.

Fred Miranda's Intellisharpen II - I use this occassionally, too. It has a lot more options than the one listed above.

Not sure Fred Miranda's is worth the cost, though. I've been very happy with the results from the free LMC54 sharpener, so I use that much more often. It's great for the photographer on a budget.
Posted by Aleksei Saunders on January 25, 2007 at 5:48 PM
I'll second Kalin in really liking the Pixel Genius' Photokit Sharpener and the way it fits into the workflow.

I'll often run the capture sharpening routine, create a mask over the new, sharpened layer, then "paint in" the areas I'd like sharpened. It's fairly easy to avoid sharpening large amounts of sky. If you're patient, you can do a lot of the little stuff with a fine brush.

The kit also offers a creative sharpening set, for areas that need a little more work, and then a group of output sharpeners for various printing styles. It's not the cheapest software (@ around $100), but it is useful and easy to use.

Cheers,
Alex
Posted by Scott Borger on January 25, 2007 at 7:36 PM
Does Pixel Genius work with CS3? Their website only speaks to CS2.
Posted by Kalin Wilson on January 25, 2007 at 7:58 PM
I would expect it to. Many of the Pixel Genius principles are Adobe consultants, like Jeff Schewe. Most of the plugin makers probably won't update their advertising or post CS3 specific upgrades until CS3 is released.

The plugin is basically a set of fancy actions integrated into PS via the automate menu. The best thing to do would be to download the demo and install it to both CS2 (still have that installed don't you?) and CS3. See if it works.

Have fun!
Posted by James H Egbert on January 25, 2007 at 10:29 PM
In CS3 and CS2 I used Smart Sharpen and tweaked it based on the Image Size and resolution.
Posted by Aleksei Saunders on January 26, 2007 at 12:53 PM
Scott,

Yes, it works in the beta version of CS3 I'm using, so I can't imagine the final version would dump compatability.

Cheers,
Alex
Posted by Rene Pirolt on January 26, 2007 at 2:28 PM
Scott - not sure if this is what you are asking, or if you have already tried this.

Using a separate layer for the tree would work very well. I don’t have the original image, but it worked very well on your jpeg.

Just create a duplicate layer of the original image, and place it above the original.
Sharpen the duplicate layer using your favorite sharpener – USM, Smart Sharpen, etc.
Pay attention to only the tree and down worry if the rest of the image is over sharpened.
Once you are done, just use the eraser tool with a hardness of zero to erase everything on the duplicate layer, but the tree and the area immediately around it.

If you are using USM you can keep the noise low by using a very high number for the Amount%, keep the Radius very low, and then move up the Threshold starting at zero.

Hope this helps.
Posted by Tony Bouchard on January 26, 2007 at 2:31 PM
Scott,

My normal method for selective sharpening is magic wand, tolerance 16-> quick mask -> gaussian blur -> sharpen. Many skies are easy enough to select with the magic wand, and the gaussian blur helps to blend in the edge of the selection. For more complicated selections my favorite method is the channels palette -> duplicate channel -> levels, then move the levels sliders to make the part you want to select all black or all white. You can then paste the duplicate channel into a mask.

For your image above, I was able to select the sky with about 5 clicks of the magic wand. I find using the eraser or paintbrush is way too time consuming and imprecise.
Posted by Nat Coalson on January 26, 2007 at 3:00 PM
I'll cast my vote for Pixel Genius PhotoKit Sharpener also. The most versatile, complete sharpening solution I've seen.
Posted by Jesse Speer on January 26, 2007 at 4:02 PM
There's a lot of good comments on PK Sharpener ... but the beauty of the "software" is that it's mostly automated. For most images, you don't need any of the manual processes describe above. When running the "capture" sharpening, I rarely do more than adjust the opacity of the sharpened layer. For "creative" sharpening, it sets up a mask (in basic terms) and lets you choose a paintbrush - about a 5 second process - and then you're off and running with your selective sharpening. And then I find that "output" sharpening never needs to be altered. Anything that shaves the guesswork and time from the processing process is a great value, indeed.
Posted by Les Barstow on January 26, 2007 at 4:27 PM
I imagine LMC54's edge sharpening is a lot like the traditional edge-sharpening techniques. The non-automated way to do this is to copy your image to a new layer, do an edge-detect, desaturate the edge-detect, invert it, adjust the brightness/contrast of the edge-detect to select an appropriate amount of detail to sharpen, blur somewhat, copy the layer, select the main image layer, copy over to a new channel, load the channel as a selection, select RGB(A) channels, then apply sharpening to just the edges.

This works really well in conjunction with a slight gaussian blur on scans where you don't want to sharpen grain, but do want to tighten the edges. On your particular image, you'd want to paint out any edges in the sky before saving your efforts into the new channel.

I haven't gone through the effort of this manual procedure since going digital, but there are some images where it works quite well. Sadly, Photoshop's edge detection routines are a bit limited; you may have to do edge-detection on individual (RGB) channels, and/or merge a few edge-detection attempts... The GIMP has a wider selection of edge detection routines which come in handy when you're having trouble teasing out edge details in your images; a quick load of the image there, saving the final edge-detection and loading it back into PS for use as a selection mask, has helped me more than once. :frown:
Posted by Scott Borger on January 26, 2007 at 11:07 PM
Thanks to all who provided suggestions. My current workflow is similar to that described by Rene and Tony. However, I really like the time saved by the Photokit Sharpener. I'll download it and give it a try.