P's Post-Processing 2: Sharpening (1/2)

[Petteri Sulonen]

Harry Shepherd wrote:
[snip]

Here's another way, this is the one I prefer.
Open image.
1. create a duplicate layer.
2. select High Pass filter with a setting of approx 45.
3. select Soft Light.
4 adjust opacity to taste. I find around 35-45 is sufficient.

I find this works very well for printing, but I haven't gotten satisfactory results for screen-rez pictures. Maybe I need to experiment some more...

Petteri
--
http://www.seittipaja.fi/index/

 

[Kevin T128195]

God knows why this thread hasn't achieved the HOT status, but in my view this is one of the best, this is the kind of threads that keep this forum going. Come on guys, where are the regular "students" who submit their works, we shy people who never posted anything have learn from threads like this immensly.

Come on fellow newbies and pros alike, please keep Petteri interested in keeping these wonderful series comming!

 

[frank bayley]

What's the difference between the Unsharpen and Sharpen functions? Do they not do the same thing: the end result is the same? When would you use one rather than the other?

Cheers
Frank B.
Ontario CANADA

 

[Petteri Sulonen]

What's the difference between the Unsharpen and Sharpen functions?
Do they not do the same thing: the end result is the same? When
would you use one rather than the other?

Actually, I'm not quite sure what Sharpen does. To take a wild guess, I think it uses some sort of heuristics to guess the "right" parameters for USM and then applies it (a bit like Auto Levels).

I do know that it doesn't give you the chance to set any parameters. Since different pictures require different amounts and types of sharpening (amount, radius, threshold), Sharpen won't give you optimal results except in the situation it was designed for. (I just tried it on an ISO3200 picture from the 10D, and all it did was accentuate the noise.)

So, to answer your query: I never use Sharpen, and always use USM instead.

BTW, it's called Unsharp Mask, not Unsharpen. The very strange name comes from an abstruse darkroom technique that does more or less the same as the filter.

Petteri
--
http://www.seittipaja.fi/index/

 

[Jeremy Rosenberger]

Or, if you're concerned about bandwidth usage due to the popularity this series will undoubtedly have, I will offer to host the pages, images, etc. on my site.

Cheers,
Jeremy

Excellent, as always. You should format your lessons as web pages
and post them on your site.

--
Jeremy L. Rosenberger
http://www.frii.com/~jeremy/

 

[Ken Coviak]

Three days and no one has turned in their homework yet? Well, we can start with this one. The work was done with Paint Shop Pro 7.

I started with a photo that was actually taken with a film camera (Minolta XGM) about 4 years ago, and scanned from the print. I ran across it a few days ago, and it seemed to be a good candidate for this assignment. This is a view of the Chicago skyline, taken from the campus of Northwestern University looking south toward the city.

I started with a defog operation. The suggested parameters of radius 60, strength 20, and clip 0 seemed to work well. I played with them a bit, raising and lowering each one, and there was either minimal effect or it made things worse. I used the suggested parameters for the defog. This made a noticable improvement, brightening the color in the sky and increasing the overall contrast.

The image was resized to 400 pixels wide, and a sharpening operation was done with USM. I tried it a few times, using different starting points, but each time seemed to end up with about the same settings. The radius had to be reduced to about 0.6 or 0.5 to eliminate the halo effects. The strength had to be reduced to about 110 to 150 to avoid artifacts along some of the edges, especially the skyline edge. The clipping needed to be about 5 to 10 to avoid accentuating noise. I ended up using radius 0.5, strength 150, and clip at 5. The before and after photos are shown below.

Ken

 

[Petteri Sulonen]

Or, if you're concerned about bandwidth usage due to the popularity
this series will undoubtedly have, I will offer to host the pages,
images, etc. on my site.

Thanks for the offer: I think I'll take you up on that, once we get that far. Mirrors might be useful, if indeed the site does generate a lot of traffic.

I've started work on the website, as a matter of fact. Here's my current to-do list:

+ Design directory layout -- 80% complete
+ Collect materials -- 80% complete
+ Tag materials as XML(1) -- 40% complete
+ Create XML TOC files
+ Install and configure Apache Tomcat and Cocoon
+ Create XSLT's for HTML page generation
+ Create pipelines for page generation
+ Create CSS styles and graphic widgets
+ Configure Cocoon for CLI use
+ Batch generate site off-line
+ Publish first version
+ Create XSLT's for PDF generation

+ Collect selection of completed assignments from students (and permissions to use them)
+ Batch generate and publish second version

It may seem a bit ambitious, but it'll pay off in the long run -- I intend to use this as a framework for other articles I'll put on the web as well, as well as any future lessons. It's much easier to maintain this than static site. Besides, I've done it before, with previous-generation tools. ;-)

Petteri
--
http://www.seittipaja.fi/index/

 

[Petteri Sulonen]

Three days and no one has turned in their homework yet? Well, we
can start with this one. The work was done with Paint Shop Pro 7.

[snip]

Good work on both counts.

Did you try using radius 0.2, amount 500% on the downsampled image? IMO it usually works well on web-size stuff. The result you got is very close to optimal, though, so I don't know if you would've gotten anything better.

You might also want to try the Fade USM trick, if you overshoot the mark slightly. (Does PSP7 permit this?)

Petteri
--
http://www.seittipaja.fi/index/

 

[Ken Coviak]

Good work on both counts.

Did you try using radius 0.2, amount 500% on the downsampled image?
IMO it usually works well on web-size stuff. The result you got is
very close to optimal, though, so I don't know if you would've
gotten anything better.

You might also want to try the Fade USM trick, if you overshoot the
mark slightly. (Does PSP7 permit this?)

Petteri
--
http://www.seittipaja.fi/index/

I had not tried radius 0.2. In giving it a go this morning, it produces a very good result.

USM with radius 0.5, strength 150:



USM with radius 0.2, strength 500, applied twice:



To my eye, radius 0.5 seems to give a slightly better result for this photo. It also took quite a bit of tweaking and scrutiny to arrive at that result. The application of radius 0.2 was done quicky and easily, with very little effort.

I am not aware of a feature like Fade USM in PSP7.

Ken

 

[Jeremy Rosenberger]

My e-mail address is easy enough to find, so let me know when you're ready to put something up!

Generating HTML from XML markup is a very nice way of doing this sort of thing. The server that I use for hosting doesn't offer much in the way of dynamic content; is it a reasonable assumption that all the XML-to-HTML transformation stuff would be done at content creation/update time, and the result HTML/JPEG/etc. is what would be published to the site?

Cheers,
Jeremy

--
Jeremy L. Rosenberger
http://www.frii.com/~jeremy/