Quick sharpening question

rickp · July 21, 2010

Hey guys,
quick question on sharpening and noise reduction. When you guys start your PPing do you sharpen your image first then reduce noise or the other way around? And also, do you run both unsharp mask and High pass of just one method?

R.

eur0ed · July 21, 2010

I personally always sharpen last. Also, another thing people often don't think of...I see many images posted that have been sharpened aggressively across the whole image...when a selective sharpen on just the subject can give a rather unique pop with a lot less aggressive modification.

Richard · July 21, 2010

I do capture sharpening and noise reduction in ACR. You cannot control the order as both are internal. That's the first step, and starting with ACR 6, usually the only NR I have to do. If additional NR is required, I do that as the first step after cropping using the Noiseware plugin in PS. If you are using some other NR program, you should run it before sharpening. Output sharpening is always my last step. I hardly ever use more than one method on a single frame, and I never apply two methods to the same part of the image. When I use more than one method (or different settings for different areas with the same method) I work on separate, masked layers.

rickp · July 21, 2010

I agree about sometimes sharpening just the subject.
Also when I do sharpen I do it on separate layers. My noise reduction program is Topaz Labs DeNoise 4.0

The other part of this that kind of confuses me is how much Radius to apply. I would think there's not a set number but do you guys have any recommendations?

arodney · July 21, 2010

I do all capture and output sharpening (which are quite different! See http://www.creativepro.com/story/feature/20357.html) in Lightroom. You can do capture sharpening in ACR as well.

pathfinder · July 21, 2010

Like Richard, I do my capture sharpening in LR3 ( ACR 6.1 ) along with any needed global noise reduction. I make frequent passes through Photoshop for local noise reduction or local sharpening via an adjustment layer with a mask if the image needs local editing in addition to global editing. I also favor NoiseWare in Photoshop, but the new noise control in ACR 6.1 is pretty darn good.

Output sharpening I do via the printing panel in LR 3.

kwcrow · July 21, 2010

rickp wrote: »

I agree about sometimes sharpening just the subject.
Also when I do sharpen I do it on separate layers. My noise reduction program is Topaz Labs DeNoise 4.0

The other part of this that kind of confuses me is how much Radius to apply. I would think there's not a set number but do you guys have any recommendations?

I usually do 2 passes of sharpening.
First pass is for local contrast enhancement. Dan Margulis calls it HIRALOAM (High Radius Low Amount). He recommends bumping the sharpness amount to 500% and then playing with the radius until you get a good shape to your picture. This is hard to describe, but for example if you are trying on a portrait, you don't want to completely wash out the cheeks, nose and chin, but give shape to the face. Once the radius is set to your liking, you reduce the amount to your liking. For me it is usually between 20 and 70 for amount.

Then I do a pass of normal sharpening with my radius usually between .5 and 2 and amount to your liking.

pathfinder · July 21, 2010

For Local contrast enhancement, I use the settings posted by Michael Reichman in Unsharp Mask Amount 15-25%, Radius 50 Threshold 0-2 and then Edit>Fade> Luminosity Blend to avoid color changes..

Some more about sharpening

I do not think of Local Contrast Enhancement as sharpening, so much as a bit of mid tone contrast. Capture Sharpening is best done in Raw Conversion.

Richard · July 22, 2010

I find that the ACR/LR Clarity adjustment is largely equivalent to the HIRALOAM method, though you cannot control the radius. OTOH, it lets you set a negative clarity if you want, which is sometimes useful for creating surreal effects. I don't know how to do that with USM.

rickp · July 22, 2010

Excuse my ignorance but what is HIRALOAM?

R.

Richard · July 22, 2010

High radius, low amount. Instead of emphasizing edges, it enhances contours by increasing local contrast. It uses the USM filter but with different settings. If you are sharpening edges you might use radius=1 and amount =100; with HIRALOAM you would typically set radius=30 and amount anywhere between 20 and 40.

ziggy53 · July 22, 2010

rickp wrote: »

Excuse my ignorance but what is HIRALOAM?

R.

"HIRALOAM" is an acronym for the "Un-Sharp Masking" (USM) settings of "Hi Radius, Low Amount", equating to a type of global sharpening.

rickp · July 22, 2010

got it. Thanks

R.

arodney · July 22, 2010

rickp wrote: »

Excuse my ignorance but what is HIRALOAM?

R.

Another unnecessary made up term (like “false Profile” or “Impossible Colors”)!

See:

http://luminous-landscape.com/forum/index.php?showtopic=24097&mode=threaded&pid=183603

Schewe has the skinny on this.

As the Chinese proverb says: The first step towards genius is calling things by their proper name. Not sure what that makes of those that feel the need to invent terms that don’t need to be invented.

BinaryFx · July 23, 2010

A common misconception, "HiRaLoAm" and "Clarity" are two totally different things. What they do have in common is higher radius and lower amounts than found in "traditional sharpening for some sort of print output", which is probably why they are thought of as being similar some commentators. HiRaLaAm moves are great for portraits and anywhere that the illusion of extra shape/dimension is agreeable (HiRaLoAm is more akin to retouching and colour correction than forms of sharpening or local contrast enhancement). HiRaLoAm uses the threshold feature to target wide edges and ignore smaller details that are affected by Clarity/LCE moves. Clarity/LCE moves use a much larger radius than HiRaLoAm. Two different techniques with two totally different goals.

Knowing the background history here, I find the comments over naming to be petty.

Stephen Marsh

http://binaryfx.customer.netspace.net.au/ (coming soon!)
http://members.ozemail.com.au/~binaryfx/
http://prepression.blogspot.com/

Richard · July 23, 2010

BinaryFx wrote: »

A common misconception, "HiRaLoAm" and "Clarity" are two totally different things. What they do have in common is higher radius and lower amounts than found in "traditional sharpening for some sort of print output". HiRaLaAm moves are great for portraits and anywhere that the illusion of extra shape/dimension is agreeable. Clarity type moves use a much larger radius than HiRaLoAm. Two different techniques with two totally different goals.

OK, so what precisely does clarity do? The only explanations I have seen are rather vague.

arodney · July 23, 2010

Richard wrote: »

OK, so what precisely does clarity do? The only explanations I have seen are rather vague.

Its a Midtone contrast tweak. It *appears* to produce sharper images because in reality, the only way we sharpen is by adjusting the tones of edges and such. Think of producing a sharpening (say USM), then ONLY targeting the midtones. In Holberts technqiue, its using High pass and Overlay blend modes. Mac’s Photoshop technique uses the Blend If to adjust just this effect to the midtones. It starts to fade in at a value of 50, is full strength at 70. Runs full strength from 70 to 185, then fades back to 205.

Richard · July 23, 2010

Thanks, Andrew. Very helpful.

BinaryFx · July 24, 2010

Richard wrote: »

OK, so what precisely does clarity do? The only explanations I have seen are rather vague.

Richard, there would be at least two ways to answer this question - if the user manual is not providing the required answer:

1) from intimate behind the scenes knowlege
2) from "reverse engineering" the command

Option 1 may not be that easy, unless one knows what trade secrets go into the process.

Option 2 is easier for the common end user.

Render an image into Photoshop with zero clarity settings. I would then dupe this image with a different name and close down the original render. Next I would open the same image in ACR and put in a med-high clarity setting without changing any other slider and then render this into Photoshop.

Layer one image over the other, set the upper image to difference blend mode and flatten. Then use the image/adjustments/equalize command to "magnify" the differences so that they are easy to visually identify.

Sincerely,

Stephen Marsh

http://binaryfx.customer.netspace.net.au/ (coming soon!)
http://members.ozemail.com.au/~binaryfx/
http://prepression.blogspot.com/

Quick sharpening question

Comments