How best to sharpen landscapes

I am curious as to what is the best way to sharpen landscapes. I an sure there is no once correct answer but typically what do you do in Photoshop?

Which tool and typical settings?

thanks in advance!

I've been using the TLR Lanscape Sharpener action lately and like it. It doesn't oversharpen and leaves you room for the final sharpening before printing. There also is a set of sharpening actions in the tools section that have numerous sharpening actions including ones tailored for various media at various dpi--Larry



http://www.thelightsright.com/TLRLandscapeSharpener

I've been having luck with a Photoshop action I've put together - essentially it's a high pass sharpener. It duplicates your background layer, sets the duplicate layer (on top) to "Overlay" mode and then converts that layer to a smart object. Then a filter ("Other": "High Pass") is run. When I built the action, I inserted a stop at the "High Pass" point, so I can choose the amount I want to run on each image. I use a small number, usually 3 to 5. It seems to work very well.
Hope this helps.
If anyone wants the action, just send me an email and I'll email it back as an .atn action.

Lately I generally use a two pass sharpening on a smart-object layer that is a duplicate of the topmost adjusted image layer. The first pass is a USM with low amount, 10% to 20%, though it can run to as high as 40% depending on the image. Radius set to 60 and threshold to 0. The second pass is a Smart Sharpen that can vary wildly depending on the size of the output image and the frequency of the image. Low frequency = large areas with little detail (sky with no clouds) and high frequency = lots of small detail (leaves on a tree). By sharpening on a separate layer I can mask areas I don't want sharpened. Sometimes I do just one or the other. It depends on the image.

Thanks much,

Dave and Jim, What is the advantage of making the new layer a smart object layer instead of just a duplicate background layer? I am still pretty basic.

Peter

Thanks much,

Dave and Jim, What is the advantage of making the new layer a smart object layer instead of just a duplicate background layer? I am still pretty basic.

Peter

Sure thing Peter. The basic advantage of using smart objects is similar to using an adjustment layer, rather than applying adjustments directly to the actual background (or duplicate). With Smart Objects, you can very easily go back and readjust the smart object by double on the new filter layer. So it is infinitely adjustable, as long as you save your master file UN-FLATTENED.

Thanks Dave!

I like a little high pass, a lot of luminance and edge sharpening (which involves layers and some heavy USM)... love the PhotoKit Sharpener routines. I prefer that any sharpening does not affect color or noise, at the pixel level, along the edges.

Also keep in mind that if you get the tones and contrast nailed, you will need much less sharpening. The perception of "sharpness" is subjective and is all about contrast, especially along the high contrast edges.

Making captures in-focus and with proper DOF always helps... try to think about overall sharpness while making the capture (including lens selection), because some of it can't be made up for later...

What needs to be sharp? What doesn't?

"try to think about overall sharpness while making the capture (including lens selection), because some of it can't be made up for later"

This really is critical. Nothing beats a stable tripod--add weight to it if necessary and don't extend the center column. Also, don't forget that you can take several shots with different focal points and blend them later to get critically sharp focus--I think this beats using the hyperfocal distance approach--Larry

Does any one use PhotoKit Sharpener on their landscapes?

Just picked up Nathaniel's reference abovr to PhotoKit, thanks.

Just picked up Nathaniel's reference abovr to PhotoKit, thanks.

I like FocalBlade sharpening software. When I was looking at the different sharpening software it seems to match what I was looking for the best. It allows surface sharpness and edge sharpness to be adjusted separately and doesn't hugely increase the file size (unless of course you sharpen in a duplicate background layer). Fairly easy to use but some flexibility in adjustments. I also use what I call contrast sharpening where I use smart sharpen with a high radius of maybe 30 and a low percentage of maybe 15 or 20. This actually gives some interesting results and you might want to give it a try. Sometimes I use it after applying FocalBlade sharpening.

I've been playing around with a variant on the highpass sharpening approach, and so far it's been my favorite of any method I've tried..

1. Duplicate your base image (perhaps use the smart layer approach)
2. Highpass filter, somewhere around 0.5-1 radius works well - note, you will hardly see any lines.. that's ok!
3. Desaturate the highpass layer
4. Create a curves layer adjustment tied to the highpass layer (either when you make it select, 'create clipping mask', or option click between the curves layer and highpass layer after you've made it
5. Click in the center of the curve, to keep the mid grays the same
6. Make a very steep curve, this will dramatically increase the contrast of the lines, giving you strong sharpening at a very small radius, and thus very little haloing and lots of fine detail sharpening. Also, you can control the degree to which light/dark edges get enhanced by shaping the curve
7. Set the blend mode of the highpass layer to overlay
8. Use masks to appropriately select areas to be sharpened

It might sound like a lot of work, but it goes real quick. I like highpass layers as they add together in overlay mode, so you can stack different levels of sharpening (ie. different radii), instead of USM layers of which you can really only have one.

Also, for high iso or otherwise noisy images, I sometimes run neat image on my highpass layer, which will leave the sharpening lines, but eliminate the fine noise.

Lastly, I've started sharpening less and less, likely for the reasons that Nat stated. If you have good contrast and a good quality raw with all the detail possible, then you really don't need much other than fine detail enhancement.

For a while now, I've been using a hybrid mix of a sharpening routine that I gleaned from the old "Radiant Vista" website and bits from a now outdated book, "photoshop 7 artistry". In effect, an edge mask is created that is used to direct the power of the USM filter to areas that really need it, and stopping it from sharpening noise, and smooth toned areas. Its a little long winded, but I have created an action and it runs very quickly once you tune the settings to your camera. I'm using a Canon 40D.

1. Create duplicate layer of your processed master file.
2. Make it a smart object.
3. In the channels palette, find the channel with the best edge contrast, and duplicate it.
4. You should be looking at a monotone image now, with your copied channel active.
5. Run the "find edges" filter, then invert the result. (ctrl + I) You should see a mostly black image, with the edges fairly well defined in white.
6. Filter-> Maximum. Radius 2px.
7. Filter-> Median. Radius 4px.
8. Gaussian Blur. Radius 3 px.
9. Image->Adjustment->Levels. Adjust the sliders until areas to be left unsharpened are black, and areas to be sharpened are white.
10. Now, set this copied channel as a selection (by alt clicking it)
11. With the selection active click back on the RGB channel. You should now be seeing a color image, with "marching ants" roughly around the areas to be sharpened.
12. Click View -> Show -> Selection Edges to hide the "marching ants". You will want them out of sight for the next step.
13. Add USM accordingly to the type of image you are working on. Since this will be a smart filter, you can go back and adjust later. I usually end up at settings of about 300, .06, 0 for a native 40D file size.

Whew! Thats it. You can go back and adjust your USM settings at any time, or even paint on the mask to add or subtract areas that need more or less sharpening. Doing this all on a smart layer allows me to resize the file to any size and only have to tweak the USM settings once I'm at my desired output size. As a final step, you can set the blending mode of this smart layer to "Luminosity" to prevent any color shifts due to sharpening, but I've never really needed to do so. Hopefully this all makes sense. Give it a try. If it works for you, make an action to save time. I've found it works very well on defining grasses, foliage and tree lines without any noticeable halos or other uglies coming into play.

TLR tools are very good - flexible actions where you can put in parameters and then at the end, you can change opacity to your liking. There are 3 scripts for different stages - I use them all the time. Though for Capture sharpening, I am using Lightroom now, but for Selective/Creative sharpening and Output/Print sharpening - these TLR scripts are very good

Aaron--would you be willing to email copies of the action? Thanks ljpuckett@comcast.net

Arnab--which of the TLR tools are you speaking of? I know of the Sharpening Toolkit and the Landscape Sharpening action--are there others?

Larry

Most of the methods above (some very complicated) apply sharpening globally. These get complicated because they are always trying to strike the right balance between areas of the image that are undersharpened, oversharpened, and sharpened just right (a.k.a. the "Goldilocks Zone"). The end result is a lot of twisting and turning of opacity sliders, selection edges, channels, modes, and various filters. Some of the explanations above look like a Twister game hit by a terrorist attack -- total friggin' chaos. Even if using an action or a third party program, you still are getting a global application, crossing your fingers in the hopes that the action will get it right -- which it doesn't always do.

Whatever sharpening method you use, do it as a layer. That way you can sharpen as much as you want for the undersharpened areas, and mask out for the areas that need less sharpening -- basically, you are no longer applying sharpening globally, but rather it is applied locally as needed. And because you can adjust the opacity of the brush that you use to mask out certain areas, each local area can be given just the right amount of sharpening -- making Goldilocks one happy camper. For example, hard edges need less sharpening to avoid halos and other artifacts, so you can use a soft brush to locally reduce the opacity of the sharpening layer for these area until you get the perfect amount of sharpening. Other areas of the image that need more sharpening can be left at full opacity, or slightly masked as necessary.

By using layer masks, you might even find that sharpening becomes a relatively easy task. As a book publisher, I have to sharpen images all the time, so I've learned to keep it efficient, effective, and above all, simple. Here's my workflow: I create a duplicate layer, blast the image with Smart Sharpen until I get the overall sharpening effect I want for the undersharpened areas of the image, and then mask out areas that have gotten over-sharpened. Since I'm in a layer mask, I can adjust the opacity of the areas I am masking out -- sometimes using a brush with only 30-50% can be enough to take a hard edge off an over-sharpened area, leaving it perfectly in the Goldilocks Zone. It takes only a few seconds to get it done, and get it done just right.

I am gonna have to try all of these just to see what happens..then I will title them with each of your names and put them all together and compare....

I used to use Marc's Technique but have since found a slighty better, faster way to do it

orig size file.
sharpen 4 times
resize to 1100-1300 on the longest side.... This is what I am currently using over on Flickr

I have a really hard time getting file sizes down to 200 KB to post here so that is why I dont post here often...I will see if any of your ideas work!!!

Thanks, this has been an interesting read!

Darren - note that the techniques presented here are primarily geared to printing. In regards to web sharpening, I believe Marc's method is designed for bringing images down to 700ish pixels, for larger images in the 1200 range, what you describe is pretty much the same concept. Anyways, if you make a print comparison of all the techniques that would be really great to see!

I agree..I would love to see them all side by side of the exact same image.....but to notice a difference you would have to print them all at about 20x30 because I really dont think you would be able to tell much difference even at 16x20

Very true (you could print the images at the resolution of a 20x30, but cropped to an 8x10 to save money )

Prints or Web?

In the minority, I make large ultra quality fine art landscape prints for a living (partial moving quickly to full) and thus don't care for simple methods. Sheer results is always my aim at whatever complexity needed as long as the results are noticeably superior.

For starters I would read Bruce Fraser's "Image Sharpening" with Adobe Photoshop (the best book EVER on the subject). You will then know more than almost anyone in the universe (including the little green aliens with the huge craniums) about sharpening. I loved the book and read it until it literally fell apart. Then apply a few updates like slight "capture sharpening" in ACR.

I am of full belief that nothing can beat competent custom sharpening of individual layers in PS with the USM (but using a variety of advanced methods such as but not limited to: mid tone sharpening (where you control the sharpening effect to fall off near black 0 and white 255) sometimes sharpening through a mask to protect non edge areas (some call it "edge sharpening") localized sharpening (brushing the effect into specific areas) Luminance (for RGB) or LAB - L sharpening (to not cause hue shifts and saturation intensities at the pixel level) sharpening for specific frequencies in an image, as in high, medium and low...) for sheer sharpening quality. IMO the other tools people use (for prints) are to simplify, automate or speed up the process, but NEVER to improve on masterful USM work. To take a print to its absolute fullest, custom sharpening (the slower more complicated way) is the way to go and advanced use of USM rules. If I was sharpening for production, obviously I would not take this ultra scrupulous approach.

I make Fujiflex (poly) prints daily. I slightly "capture sharpen" (a slight pre sharpen in the raw pipeline) raw images in ACR, the mid to higher frequency areas only (by bringing versions/layers into PS and erasing the areas I want no capture sharpening - like the sky...).

I do a general global mid-tone sharpen to an image after up sizing it (upsampled by using Bicubic Smoother) in LAB mode (L Channel) in PS to generally "tighten" the image. Radius and amount varies from print to print. I then do custom localized sharpening (using the variety of methods previously mentioned) to the image in LAB mode (L Channel) to finish it off, again using the USM. Sometimes I will use masks to protect some of the non-edge areas from getting sharpening or too much of it. This overall approach takes a ton of work but, to my knowledge, it can not be matched or beaten if sheer quality fine art print results are your aim.

Also as someone else mentioned, if your contrast is right in an image, ESPECIALLY your mid tone contrast, then an image will appear sharper even without sharpening, and thus will need less sharpening to look ideal.

Side note: I do a ton of test printing! First at 12x18, then 24", then 30" then 40" then 50" then 60" and sometimes much larger depending on the aspect ratio and total original M-Pixels. I always sharpen with my screen res at its highest level so that you can see better what will happen to the print, also at one zoom level closer than "Print Size" in PS View menu. Also you need to sharpen to where the image looks slightly overdone on your monitor because there is a softening effect when a printer converts squares into dots for continuous tone printing.

Only by dumping A LOT of money and time into the machine can you really begin to master printing and sharpening. But what FUN!

Good luck.

I'm a bit late to this party.... I've not printed above 20 x 30. Given the same image printed at say 11x17 do you need a lot more amount, or radius, or threshold when printing to 30+ inches on the long side? Or are all those values held constant?