Sharpening in Photoshop

Sue and I ran a beginner's digital night recently down at the ADAPS clubrooms and one of the things we did was test three types of sharpening.

1) Unsharp mask directly on the RGB file
2) First converting to Lab mode and applying unsharp mask to the lightness channel, then converting back to RGB
3) The High Pass method

These were all done on duplicate copies of the same image, we then viewed them all at the same magnification and took a consensus on the preferred shot.

Option 1 was clearly the best result, which was not what we expected. It just goes to show that we need to try all techniques for ourselves without necessarily relying on what publications tell us!

Can anyone else try this and confirm our result?

John,

That's an interesting thought... I tried the three methods with a shot I took yesterday of my new born nephew. I sharpened the image quite strongly (100%/2.1/0), and cut the same selection out of each method to view. The selection was an area of baby face with the soft fine facial fair, gently lit from the side. Consequently there's quite a bit of contrast already in the hair.

My results showed that Lab was the most natural. The other two methods sharpened the noise too (ISO400, *istD). The Lab method effectively sharpened the contrast in the hairs, making them more visible.

Then I tried a shot of some Lego - nice hard angles and even colours. In that case I would say that straight RGB sharpening looked best, followed by High Pass, then Lab.

I would say, then, that the subject dictates which sharpening method to use . Personally, I tend to make a duplicate layer and RGB USM. That way I can change the opacity of the sharpening layer, or even create a layer mask to selectively paint in sharpening (great for portraits where you just sharpen the eye, nose, mouth details). This wasn't my idea, but one from Russell Brown - www.russellbrown.com (Look for Painting with Sharpen - works in Photoshop Elements too )

I'd be interested to hear others' views too

Matt

Hi Matt

Nice one!

The picture that we chose for our test was in fact an architectural shot, which coincides pretty well with your Lego image. So I suspect you are quite right that it depends upon the subject matter and we can't use one method only for all pictures.

It really emphasises that we can't slavishly follow any one recommendation and we should experiment for ourselves using others' advice only as a starting point. Bad news for those of us short of time!

Thanks for following up my original message with some practcal experimentation.

Best wishes,

John,

I always enjoy a challenge Thanks for your suggestion - just trying out the methods made me sit up and think.

Also, I took a good long look at Russell Brown's site. He has some superb tricks for PS. Even the sharpening one "Advanced Sharpening" is brilliant. Watch the video, but it works like this:
1. Copy the layer
2. USM quite hard (a little more than you would normally)
3. Copy the USM'd layer
4. Make one USM layer mode "Lighten"
5. Make the other USM layer mode "Darken"
6. Now adjust the opacity of each of the two USM layers to get the effect you want.

It's fantastic for skin and hair - you can regulate how much the sharpening makes hightlights and how much it makes shadows. Then you can combine the approach with a layer mask to paint in the sharpening over specific areas. Cool

(There's also a really neat trick for making B&W using two Hue and Saturation layers - go and watch it )

I have also been looking at FixerLab's Focus Fixer - it did a pretty neat job of a genuinely out of focus image I had. However, it can be used as a 'general' sharpener too.

I have also read, and often found in practise, that two USM stages can work well - a larger radius, lower percentage, then a small radius higher percentage.

Wow! Just sooooo many tools available and so little time to play

Keep smiling
Matt

Thought I'd add a link to this thread about "Smart sharpening":

http://www.luminous-landscape.com/tutorials/smart_sharp.shtml
(a variation for Photoshop Elements: http://www.retouchpro.com/tutorials/index.php?m=show&id=69)

Fantastic!!!

I also found a variation of my previous technique that produces "punchy" sharp images (works well on tigers ):
* Copy background layer twice
* Top layer - USM 300%/0.7, layer mode = Soft Light, Opacity somewhere around 20-30%
* Middle layer - USM 150%/3, layer mode = Darken, Opacity somewhere around 40%

The soft light has the effect of increasing saturation and contrast. Now I must go and learn the layer modes properly

Similar effect: http://www.luminous-landscape.com/tutorials/high-pass-sharpening.shtml

Have fun!
Matt

Nice tips there gentlemen!!

Also, I find that different methods somehow works differently on the final output - whether for prints or screen/web display. And Matt is right. We must find the method that best suits the subject/texture and their "edginess" (if there's such a word ).

In most cases, I'm already happy with the standard Photoshop USM
.. although I play around with the USM values according to the image size and/or output. For screen display for instance (images not exceeding 1024 x 76, I tend to use 50-120%, 0.3-0.4, 2-4 .

If my image is intended for large prints (especially A4 or A3), I find radius of 0.5 to 0.8 works best. Again, this depends on the subject.