Sharpening technique for photoshop CS2+

Because of some trials that some user did this and past week concerning sharpening I'm wondering if I can improve mine as well.
Please tell me the sharpening technique you mostly use.
Most sharpen options are the same with CS2 and up so that's why I said CS2+ but any sharpening technique is welcome.

I'm currently looking in smart sharpening and that looks quite interesting.

Not sure if I'm one of the users or not But I tried out MattMatic's PS sharpening action for a few weeks and found it excellent. It's only in high contrast areas (like tree branches) than Nik Sharpener is significantly better (far fewer artifacts). They are otherwise quite closely matched. Just FYI

Already knew all of them Tim, haven't used Nik though

It highly depends on the photo which sharpening works the best so I'm hoping that some members know a sharpening technique besides unsharpen or highpass or use those in a special way.
Just trying to expand the tools that can be used

Now that I use Silkypix for RAW conversion I find I do less sharpening in PS

However, when I do, it's normally like this:
* Duplicate layer
* Smart Sharpen
* Change layer blend mode to Luminosity
* Adjust opacity to taste
* (Optionally) create a layer mask and paint black/white to localise.

HTH
Matt

IMO photokit sharpener from PixelGenius is the best on the market.
It allows for Input, Output and creative sharpening in a very controlled way. It is expensive, but worth every penny.

Further more I use High pass sharpening a lot for fine detailed back grounds which have to show structure. Playing with the luminosity slider makes it very powerful and precise.

Those two are my main sharpening tools.

I've copied Matt's sharpening method as advised by Stefan but need a quiet two or three hours to ge my head round it!

I have a version of Photoshop which was lauched at the Great Exhibition in the 19th century, so up to this point have used either unsharp mask or Gareth's suggestion of two layers blended together, one of which has a High Pass filter.

Regards,

Andrew

Thank you Matt, aren't you using your old method anymore
I love what kind of control you've with smart sharpen but I haven't used it that much yet.

Martin, I'm using 64bit so I need to use the beta from that

Andrew.
Somehow I use unsharpen mask more then highpass, or even a combo of the two.

My method seems rather crude, but then my PP'ing skills are rather basic. I also have PSE 6, not CS2+. But FWIW I tend to use Adjust Sharpness>Remove: Lens Blur (I find this generally gives better results than Gaussian blur).

Is this hopelessly unsophisticated?

BTW what is Matt's (old) method?

Dr M - Matt's sharpening technique is explained in his article here. There are also download links at the end for a ready-made action for Photoshop. I don't know if PSE will be able to use it though, so you might be doing it all manually

Cheers Tim.

The simple way is to learn how to use Unsharp Mask, and this will be fine for most purposes.

For an A3 image from a 10-14MP camera, start at 100%, Radius 1.9 and Threshold 3.

For an EPZ or Web picture of 800 - 1000 pixels wide start at 50%, Radius 1 and Threshold 3.

There are many refinements after that within the same dialog box, depending on the image, but as a starting point those figures should serve well.

i just use unsharp mask also like john says, seems to work for me. i use CS4 extended, but unsharp mask should be the same on all of them.

Andrew, if your not far from bristol and want to upgrade your version of photoshop give me a shout.

Thanks Ian, I might well take you up on your kind offer: if my old lap-top could take the excitement!

Regards,

Andrew

USM is too obvious for me in many cases, I prefer something else....

- Take the original
- duplicate twice.
- on one, edge-detect and then convert to black and white.
- using the color levels, remove the grey (shift the black point up and the white point down) until you've got what could be a chalk drawing of the scene
- now gaussian blur it - somewhere between 4 and 10 pix works well for me
- Now select the entire pic.

On the second duplicate:
- decompose to LAB or HSV. If you're using LAB, sharpen L; if it's HSV, then V. I'll use L in the next section, as that's the way I do it.
- Select the pic.
- create a new channel. Name it "sharpmask" or similar
- make sure new channel is selected, not the pic and not the layers, and then paste. In GIMP, at this point you'd have to Anchor the layer.
- now go back to the channels and convert the channel to a selection.
- with the L layer selected, use USM. You can go pretty strong - 5, 1.0, 0 - and it will only sharpen the edges and pale sections of the pic.
It's these that give the "illusion" of sharpness, even if it's not perfect. It's especially obvious in small prints (up to around A4), because you're looking at them from a bit further away and so your brain fills in the other bits, even if it's not 100%.

I also tend to do any denoising heavily on the duplicate that will be the mask and more delicately on the other; that way, the mask is slightly larger (which you can influence with the gaussian blur) and the noise is not amplified.
That's the other reason you shouldn't try sharpening blue channels, as they get more noise - the luminosity is also really obvious.

Takes around 30min per pic on my T60, this machine is around 3 minutes.

Bret

Thank you all, seems every one has their own preference

Bret what a list but isn't that almost the same as USM?