unsharp mask

Discussion of all digital SLR cameras under the Minolta and Konica Minolta brands
Forum rules
No more than three images or three external links allowed in any post or reply. Please trim quotations and do not include images in quotes unless essential.
mikehawthorne
Grand Caliph
Posts: 421
Joined: Thu Aug 07, 2008 5:41 am

unsharp mask

Unread postby mikehawthorne » Tue May 25, 2010 10:35 pm

Hi All

i have used unsharp mask often but i wondered how much is recomended by photoclubalpha members, when processing their photos.

mikehawthorne

Alan Shaw
Acolyte
Posts: 22
Joined: Fri Aug 01, 2008 9:17 pm

Re: unsharp mask

Unread postby Alan Shaw » Tue May 25, 2010 11:25 pm

Mike,
I can't answer your question specifically but would like to broaden your question a little and ask members: when you apply USM what process do you go through with each image to determine the right amount of sharpening? What do you look for to make your decisions?

There are many, many examples of tutorials and articles that say apply such and such an amount but I have not seen a discussion that explains "why" those settings in particular were chosen, or the process the author chose to arrive at those settings. It's the thinking behind the settings that I am especially interested in.

FWIW: Until recently I used Micrografx Picture Publisher (anyone else remember that program) and would zoom in on a prominent edge, apply USM until I'd introduced artifacts, and then knocked the amount and threshold back until they'd disappear. Crude, but it seemed to be effective enough for my purposes. I have recently swapped to Lightroom 2 and have not really had enough time to explore it beyond letting it suggest the right settings for me. So far so good.

Alan
Alan Shaw
Brisbane, Australia
-------------------------------------------------------------------
Visit my Railgrafx rail photography site:
http://www.railgrafx.id.au

Visit my Moreton Bay Model Railways Site:
http://home.iprimus.com.au/mbmr/index.html

David Kilpatrick
Site Admin
Posts: 6248
Joined: Sat May 19, 2007 1:14 pm
Location: Kelso, Scotland
Contact:

Re: unsharp mask

Unread postby David Kilpatrick » Wed May 26, 2010 1:13 am

Unsharp mark should always be used at final image size. For web use, settings between 0.5 and 0.8 pixel, threshold between 0 and 8, radius 50 to 100 will generally work well. For large web images (Flickr etc) use the higher values of pixel and radius; threshold depends entirely on noise levels in the image. If the image noise is low, set 0 every time and maximum detail will be retained.

For inkjet prints, values from 0.8 to 1.5 pixels, threshold 0 to 8, and radius 100 to 250 will normally work well but for poorly focused subjects increasing the pixel level to as much as 3, and radius to 500, may give a sharper (if crude looking) print.

For litho print, pixel 1.5 to 3, threshold 0 to 16, and radius 150 to 500 depending on subject can be a good range to work within. Newspaper repro can take higher values, stochastic screening (very fine, random) the lowest levels.

David

User avatar
Dr. Harout
Tower of Babel
Posts: 5833
Joined: Wed May 30, 2007 7:38 pm
Location: Yerevan, Armenia
Contact:

Re: unsharp mask

Unread postby Dr. Harout » Wed May 26, 2010 6:52 am

David Kilpatrick wrote:... If the image noise is low, set 0 every time and maximum detail will be retained.

David


Which parameter setting to 0, pixel and radius or all?
A99 + a7rII + Sony, Zeiss, Minolta, Rokinon and M42 lenses

Flickr

David Kilpatrick
Site Admin
Posts: 6248
Joined: Sat May 19, 2007 1:14 pm
Location: Kelso, Scotland
Contact:

Re: unsharp mask

Unread postby David Kilpatrick » Wed May 26, 2010 11:19 am

Threshold. Pixel and radius are the most useful variables. Pixel determines the extent of the reverse-mask effect on hard edge transitions; try setting it to something like 5 pixels and 500 percent, and you will see a very clear double line with 5 light pixels then 5 dark (or the opposite). If you have detail which is already down to being maybe 4-5 pixels wide (not much texture in an image is any smaller than 3 pixels) then obvious using 5 pixels mask will either ignore such detail or disrupt it.

The threshold setting determines what it does. 0 threshold means that USM will 'see' differences of just 1 step (in 256) between adjacent pixels. Even the grain or noise will be detected as detail, and sharpened. If you want to make grain look strong, simply use 0.5 radius (half a pixel), 0 threshold, and 500 per cent strength.

In most low-ISO images, there is not a significant value change from pixel to pixel within any area of plain tone, so setting 0 threshold is OK. You will get some increase in noise contrast, but not bad. To remove any tendency to get added noise, use higher threshold values - generally, up to 8 will be safe. Beyond 8, setting a higher threshold tells USM to treat noisy or detailed areas as if they are a single smooth tone. It begins to act like noise reduction software. As you increase threshold to higher values like 24, 50 or whatever you will see that noise is replaced by loss of information. The original detail in the picture is replaced by a smoothed tone.

One of the most useful filters in Photoshop is 'Sharpen Edges' which is a special type of USM. It applies a relatively high threshold value and a very small radius edge-effect, smoothing tones while defining high contrast edges. 'Sharpen Edges' is very useful for web pictures, as it can reduce the data size of the compressed JPEG while making the image appear crisp; you can use a higher quality JPEG setting without getting too large a file, especially for avatars and stuff like that.

David

User avatar
Birma
Tower of Babel
Posts: 6679
Joined: Sun Jun 29, 2008 3:10 pm

Re: unsharp mask

Unread postby Birma » Wed May 26, 2010 4:52 pm

Thanks to Mike for raising the question and to David for the very useful answer - thread duly bookmarked :)
Nex 5, Nex 6 (IR), A7M2, A99 and a bunch of lenses.

User avatar
Pirate
Grand Caliph
Posts: 286
Joined: Mon Nov 05, 2007 6:56 pm
Location: Liverpool, UK
Contact:

Re: unsharp mask

Unread postby Pirate » Sun Jan 09, 2011 11:56 pm

I use the following in CS5 (subtle), but can be increased if required after first adjusting with IDC.

Amount = 200% (increase to 300% + if more is required)
Radius: 0.2 pixels
Threshold: 0

However, I normally open my .ARW RAW files using Sony IDC v3.2 as the programme has a great sharpening feature called overshoot and undershoot. This allows for sharpening of light and dark edges where they join (like black and white), so you can sharpen either or both sides as opposed to the entire image. Once that's done, use the threshold slider to reduce noise to the adjusted side to smooth out. I then export as a .TIF to CS5 to finish off.


Return to “DSLRs pre-2006”

Who is online

Users browsing this forum: No registered users and 2 guests

cron