DPChallenge: A Digital Photography Contest You are not logged in. (log in or register) 
 

DPChallenge Forums >> Individual Photograph Discussion >> Sharpening question?
Pages:  
Showing posts 1 - 25 of 35, (reverse)
AuthorThread
 06/06/2005 07:43:05 AM · #1


Too much, too little, not enough? I'm learning a new technique for sharpening images,a nd I'm trying to get a feel for what looks good.
Any input is apprecaited.

And yes, I know the model is fuzzy. :)

06/06/2005 08:16:37 AM · #2
just purrrrfect, IMO.
 06/06/2005 08:24:48 AM · #3
left comment. I think it's right on, and if you were to tinker, perhaps push it a "tad" more.
 06/06/2005 09:35:28 AM · #4
Looks pretty good to me. A cat's fur may not be the best subject to test your sharpening methods, but the whiskers look just right. Nice picture btw.
 06/06/2005 06:23:37 PM · #5
what better model to use than one who's fuzzy to begin with! :)

Any more input?

Clara
 06/06/2005 06:32:19 PM · #6
Looks good to me. No little jaggies to be found. ;-)
 06/06/2005 06:33:11 PM · #7
What technique didja use, eh?
 06/06/2005 06:33:34 PM · #8
Looks really great.
 06/06/2005 06:34:18 PM · #9
Oh yeah, please do tell us what the new technique is! :-D
 06/06/2005 06:45:26 PM · #10
This is probably a trick and NOTHING was done.
Prolly got a new lens or something.
:)

It does look great, regardless of how you acheived it.
Now just need to level the horizon and add a bunch or frames to it and it'll be perfect.
06/06/2005 06:45:57 PM · #11
I went to my USM filter, set it to Amount 225, Radius .5 and Threshold 0. Then go to the edit menu, select Fade Unsharp Mask. Set the mode to Luminosity. Adjust the slider from there to get the degree of sharp you want.

Courtesy of the Photoshop CS book by Scott Kelby. :)

Clara
 06/06/2005 06:47:09 PM · #12
Originally posted by BradP:
Now just need to level the horizon and add a bunch or frames to it and it'll be perfect.


You forgot dodging and burning it. ;)

Clara
 06/06/2005 06:49:53 PM · #13
Was this at full size or after resizing for DPC?

This technique is basically the same as appying USM to the lightness channel in Labcolor from what I've heard, but then again, why take the work of a lunatic (me).

Oh and I didn't forget dodging & burning - thought that was a given - kinda' like breathing!

Message edited by author 2005-06-06 18:59:48.
 06/06/2005 06:57:15 PM · #14
Originally posted by BradP:
Was this at full size or after resizing for DPC?


After. As always, the very last thing I did. I've had the book for about 5 months. I decided to do something radical this weekend and actually read it. ;)

Originally posted by BradP:

This technique is basically the same as appying USN to the lightness channel in Labcolor from what I've heard, but then again, why take the work of a lunatic (me).


My understanding is it's basically the same. But I'm scared of Labcolormode. :) This was fewer steps.

Originally posted by BradP:

Oh and I didn't forget dodging & burning - thought that was a given - kinda' like breathing!


Did I mention it's too hot to burn? I ordered pizza cause I refuse to cook in my breadbox.

Clara
 06/06/2005 06:58:54 PM · #15
It is the same as convertig to LAB, running USM on the lightness channel, then converting back. The advantage of this is that you don't convert to the LAB color space and back. This means you won't shift colors that are within the gamut of sRGB but outside of LAB.
I'd add to this that I highly recommend setting threshold to a non-zero value, to avoid bringing up noise. A value of 2-3 seems to work well for DSLR images shot at moderate ISO (<=400).
 06/06/2005 06:59:42 PM · #16
Mmmm... breadbox pizza...
06/06/2005 07:04:23 PM · #17
Originally posted by kirbic:

I'd add to this that I highly recommend setting threshold to a non-zero value, to avoid bringing up noise. A value of 2-3 seems to work well for DSLR images shot at moderate ISO (<=400).


Okay- I'm still suffering from brain lag this week. Why does a zero value on threshold bring up noise?

Clara- Yay! Sharpening 101 today!
06/06/2005 07:30:21 PM · #18
The threshold setting determines "how different" two pixels have to be before the sharpening effect is applied. If the threshold is zero, the USM filter will try to apply sharpening between all pixels/areas which differ in color (like noise spots), accentuating them.

To avoid creating banding and artifacts in areas of smooth gradation (like flesh tones), I usually set the threshold to 5 or 7, meaning the areas have to be "pretty different" before the effect is applied.
 06/06/2005 07:32:31 PM · #19
Originally posted by blemt:
Originally posted by kirbic:

I'd add to this that I highly recommend setting threshold to a non-zero value, to avoid bringing up noise. A value of 2-3 seems to work well for DSLR images shot at moderate ISO (<=400).


Okay- I'm still suffering from brain lag this week. Why does a zero value on threshold bring up noise?

Clara- Yay! Sharpening 101 today!


The threshold value is the lowest difference between adjacent pixels that the filter will consider to be an edge. Therefore, if the standard deviation of the noise in the photo is less than about 1.5, as it is for most low-ISO shots, then most of the differences that aree due to noise will lie below the threshold, and thus not be sharpened.
 06/06/2005 07:55:51 PM · #20
Originally posted by kirbic:
... if the standard deviation of the noise in the photo is less than about 1.5, as it is for most low-ISO shots, then most of the differences that aree due to noise will lie below the threshold, and thus not be sharpened.

Is there an easy way to estimate this value from densitomiter readings? That might allow calculating an "optimum" TH level.
 06/06/2005 07:58:02 PM · #21
Heeeeellllllpppp! I lost my kirbic/GeneralE to English dictionary! I think justine had it last.

Translation for the statistically challenged?

06/06/2005 09:06:11 PM · #22
Eh, no stats necessary... When in doubt, just set it to 3 and let 'er rip. For noisy source images, if you find that the noise in areas of continuous tone looks worse after USM, undo and re-run USM with a higher threshold. If the noise is obtrusive just in the shadows, try running USM with a (softly graduated) mask to avoid running USM on the shadows, where you really don't need it anyhow.
Experiment with blue skies shot at various ISOs, you'll soon find out what works best.
 06/06/2005 09:12:03 PM · #23
Originally posted by kirbic:
Eh, no stats necessary... When in doubt, just set it to 3 and let 'er rip. For noisy source images, if you find that the noise in areas of continuous tone looks worse after USM, undo and re-run USM with a higher threshold. If the noise is obtrusive just in the shadows, try running USM with a (softly graduated) mask to avoid running USM on the shadows, where you really don't need it anyhow.
Experiment with blue skies shot at various ISOs, you'll soon find out what works best.


Ok, I followed below, but you got me there--how do you run USM with a mask?
 06/06/2005 09:46:18 PM · #24
Originally posted by blemt:
*snipped*
I've had the book for about 5 months. I decided to do something radical this weekend and actually read it. ;)

Clara


Too funny! Isn't it amazing what you find out you can do when a book is opened and read! Oh, and I think you did good with trying new technique. - Christine

Message edited by author 2005-06-06 21:48:58.
 06/06/2005 09:51:01 PM · #25
hard to tell without going full view
Pages:  
Current Server Time: 09/13/2025 07:15:34 PM

Please log in or register to post to the forums.


Home - Challenges - Community - League - Photos - Cameras - Lenses - Learn - Help - Terms of Use - Privacy - Top ^
DPChallenge, and website content and design, Copyright © 2001-2025 Challenging Technologies, LLC.
All digital photo copyrights belong to the photographers and may not be used without permission.
Current Server Time: 09/13/2025 07:15:34 PM EDT.