How Much Sharpening?

[Steve P]

I more or less shoot exclusively with Raw these days and have been converting through EX. I especially like the way it allows you to adjust exposure! This can change however as I sometimes prefure CS. I guess it depends on how the mood takes me on the day.
My question is sharpening or should I say by how much. I guess it's going to be down to personnel taste but I'd be interested to know how others handle it.

Steve

[Swampy]

You might do a search for sharpening Steve. There was a discussion about it and how people do it and how much.

I've built a filter that lets me selectively mark the edges only to sharpen and then, I sharpen according to the needs of the particular picture. Somewhere between 50-75%, .7 and 0 in PS. Which seemed to be the conclusion as to where to sharpen. All in PS.

[Steve P]

So you've built a filter, clever stuff, Interestingly I tend to use higher figures than yourself, depending on what the subject is of course. I've read many photoshop manuals that suggest many different examples. For instance when shooting buildings/landscapes one can go to 50-75%, 2.0, 1. Group shots something like 50-75%, 1.0, 0. Yes I know this will depend on many factors.

Scott Kelby has a whole chapter on the subject and he's not shy when he wants to be.

Cheers Swampy

[HulaMike]

Steve, you'll get as many solutions as posters here I think. Sharpening is such a subjective thing. I have used a Fred Miranda plug-in called Intellisharpen II and its predecessor for over a year now and like it quite well. What it does is analyze the image and sharpen areas differently. Close focus objects are sharpened more than distant objects or noise for example. It can be found here:

http://www.fredmiranda.com/shopping/IS

Another way to manually focus selectively can be found in this Russell Brown tutorial. It must be similar to the way Swampy does things. Here's that link, scroll down:

http://www.russellbrown.com/tips_tech.html

[Steve P]

Hey Mike, cheers mate! May well give it a look. Your right though, sharpening is pretty much down to the individual is it not. When I look at how I use to do it I cringe as I didnt apply enough. This whole thing has been an amazing learning experience for me!

Tis always funny looking at old pics. I was looking through some old digital photos I took with my old Olympus E20 and thought to myself it wern't a bad camera really, I just needed to know how to sharpen the image.

[NZDoug]

I asked the same question at Photo District News and the main reply was
http://www.pixelgenius.com/sharpener/ :)

[HulaMike]

Hey Mike, cheers mate! May well give it a look. Your right though, sharpening is pretty much down to the individual is it not. When I look at how I use to do it I cringe as I didnt apply enough. This whole thing has been an amazing learning experience for me!

Tis always funny looking at old pics. I was looking through some old digital photos I took with my old Olympus E20 and thought to myself it wern't a bad camera really, I just needed to know how to sharpen the image.

How true. I opened some old files edited pre CS and found that I could greatly improve them with a little shadow/highlight recovery. It never ends really. I guess we should adhere to the advice to save all our pics in an unedited form before we work on them because in a year or so as the technology mateurs (and we get better) we'll all be able to improve them. ;)

[easternherp]

As you have said Steve, it is down to personal taste and what the final image is going to be used for. If for the web and other low res files then small amounts of sharpening like swampy does, for higher res images then higher like you have said. I tend to do 150%–200%,2.0–5.0,4 levels.
As you said it depends on the image. I use the High pass filter quite a lot as it gives a better sharpening where you need it rather than over the whole image.

[HulaMike]

Andrew, could you post your high pass filter technique? How does that work?

[Eddie the Gnat]

Duplicate layer, apply HP filter so everything is pretty much grey with only edges showing up, desaturate (optional), change blending mode to opacity and season to taste (sorry for jumping in here Andrew :o )

Any unsharp mask numbers are very much dependent on image size - the rule of thumb I always use is to sharpen as little as you can possibly get away with! :)

I do like the HP technique though.

[HulaMike]

Duplicate layer, apply HP filter so everything is pretty much grey with only edges showing up, desaturate (optional), change blending mode to opacity and season to taste (sorry for jumping in here Andrew :o )

I do like the HP technique though.

Got the basic technique down but there is no "opacity" selection available in blending mode in CS. Do you mean "multiply"? I can adjust the opacity of any filter, but I don't see an opacity filter.

[deMille]

Just a thought;

Convert the image to LAB color, select the "Lightness" channel, use "unsharp mask", re-convert to RGB. You can also sharpen individual color layers, should you desire with great results.

A billion choices ... play with them all until you find what you like for your own images.

Dale

[easternherp]

Duplicate layer, apply HP filter so everything is pretty much grey with only edges showing up, desaturate (optional), change blending mode to opacity and season to taste (sorry for jumping in here Andrew :o )

Any unsharp mask numbers are very much dependent on image size - the rule of thumb I always use is to sharpen as little as you can possibly get away with! :)

I do like the HP technique though.

Hi Eddie,

no problem about jumping in.
I vary the blending mode from overlay to soft light and hard light for different subjects.
As others have said, there are no rules just personal taste.

[Eddie the Gnat]

Sorry, Mike - I meant overlay. Then you can tweak the opacity.
Apologies for the confusion, but at least I made myself look stupid! :stupid:

E.

[HulaMike]

Thanks guys. Looks promising on the few pics I tried. Hard to judge sharpening on the high pass image tho, probably comes with experience. Thanks for the tips.

[Tom V]

I almost never use in-camera sharpening. I do all sharpening in Photoshop.

I sharpen for two reasons, so I sharpen in two steps.

1) I sharpen to compensate for the filter that is over the sensor (some say it is clear glass, other say it's a high-pass filter - whatever it is, or even if it's not the reason), that make images from the sensor look soft. This is a high power (150% - 250%) sharpen over a little range, with no (0) threshold for noiseless images from TIF or RAW files at low ISO, or a minor (3-10) threshold for JPG or noisy images. The actual number depends on: subject matter (hard or soft edged), and amount of noise / contrast / jpg jaggies in the image.

- then I do all my PS fooling around: Levels, Curves, Color Balance, retouching, etc. I save my RGB image (usually layered by now) as my master .psd file.

Then I adapt a copy of the master RGB file for the output size and method. I resize the image, convert to CMYK (usually), and then move to Sharpen #2

2) I sharpen to compensate for the blurring effect of the output method. Inkjet printers usually are pretty sharp, so I don't do much unless the image is large. Printing to a screen process (magazine ad) breaks my image into patterns of dots, which work against the image's natural sharpness. I have my image sized so that every 4 pixels (2 high x 2 wide) equals one halftone dot. To me, it means that I am loosing half the detail, and that any sharpening must span more than 2 pixels to even be noticeable. For images going to magazines, I USM 100% / 2-4 px radius (depending on subject matter) / 0-10 threshold (depending on noise).

After USM for magazine output, I can see a slight outline effect in areas of higher contrast. This is fine, and will be mostly hidden by the halftone screen process. I used to only use 2 px radius, but now I am favoring 4 px radius.

If I output an image on inkjet that was sharpened for a magazine, the outline is visible. If I intend to have inkjet prints for a final output method, I would use less USM in the second step (if any).

I view my image onscreen at 200% or larger when doing USM. Never judge USM or try to do precise selections at 33% or 66.6% screen magnification - you are seeing the pixels "interpolated" to match your monitor resolution.

[HulaMike]

Why 200% and not 100%?

[Tom V]

Why 200% and not 100%?

I assume you are talking about the screen magnification. I feel that viewing the sharpness with the on-screen image at 200% gives me a more accurate representation of the printed sharpness. When I view at 100%, it may LOOK sharp, but at 200% - I KNOW if it is sharp.