NX/NX2:Gen - Commands - Sharpening multiple times

Joined
Aug 27, 2006
Messages
1,727
Location
The Garden State, USA
Can globaly sharpening an image multiple times be beneficial? if so what is generally the best approach and in what cicurmumstances would you employ this method?
 
Joined
Sep 13, 2007
Messages
15,405
Location
Northern VA suburb of Washington, DC
Can globaly sharpening an image multiple times be beneficial? if so what is generally the best approach and in what cicurmumstances would you employ this method?
One classic use of sharpening is a technique called Local Contrast Enhancement (LCE). Some would argue that it's not sharpening per se, even though the technique requires using a USM starting point of about 10/50/0 in Capture NX. You would apply it before applying the normal USM. Though I might usually use 50/5/4 as my normal sharpening, if I have used the LCE I sometimes have to change the first parameter to about 40.

Once you reduce a file's size for output to the web, it's typical to apply another round of USM. Otherwise, after dramatically reducing the size some of the sharpness that was in the larger file is gone. I use about 10/5/0 for Web output.

Keep in mind that the starting USM parameters that you use will change somewhat depending on the camera because of the different file sizes that different cameras produce.
 
Joined
Oct 14, 2006
Messages
254
Location
Colorado, USA
You generally want to sharpen an image in three steps:

1) Sharpening to remove the effect of the antialiasing filter on your sensor. This is often referred to as capture sharpening.

2) Sharpening locally to enhance parts of the image. This is usually referred to as creative sharpening.

3) Sharpening globally to counteract the effects of resizing/resampling or to mitigate dot gain when printing. This is known as output sharpening.

As Mike points out, you can also use the USM and other sharpening tools to affect local contrast.

The starting points for each of these techniques will vary depending on the model of your camera and the printer/paper combination you use to print.

-Jason
 
Joined
Oct 9, 2005
Messages
25,321
Location
Moscow, Idaho
If that doesn't explain in a nutshell why I only try to help out occasionally by responding to posts as opposed to why Jason writes successful books, nothing will. :biggrin:

Mike,
I constantly tell my students that explaining things to others is the best way to develop their own comprehension and curiosity. And sometimes the way someone else (other than us stuffy high-brow profs) might explain things can help others learn better. So please, help away!:smile:


Now if I could only follow my own advice . . . . .:redface:
 
Joined
Jul 1, 2008
Messages
67
Location
U.S.A.
Capture Sharpening

A comment on Capture Sharpening (C-S for short): Bruce Fraser was the original author of the three-phase sharpening method. His guidelines on Capture Sharpening included the following:
  • As with all forms of sharpening, it is best to reduce noise before sharpening.
  • C-S is often restricted to the midtones to prevent blowout/blockup at the ends of the tonal spectrum early in the image editing process (we need to leave headroom for other operations). Since NX2 does not have tonal masking, applying sharpening to important subject areas via a Selection Control Point or by painting the result can work OK.
  • C-S is often restricted to edges via an edge mask so we don’t punch up noise or other artifacts which show up mainly in the smooth areas. The only way I know how to do this in NX2 is with High Pass sharpening.
  • C-S is applied gently since more sharpening is applied later.
More details on C-S and the other phases can be found in Bruce’s CS2/CS3 books or his book on sharpening. Note that this info is somewhat tangential to the OP’s question since it concerned global sharpening.

Also, for perspective, authors such as Thom Hogan generally sharpens twice and not three times. I don’t think he mentions that it in his article on sharpening, which is found here: http://bythom.com/sharpening.htm and is worth reading.

Jim


Jim
 
Joined
Sep 13, 2007
Messages
15,405
Location
Northern VA suburb of Washington, DC
A comment on Capture Sharpening (C-S for short): Bruce Fraser was the original author of the three-phase sharpening method. His guidelines on Capture Sharpening included the following:
  • As with all forms of sharpening, it is best to reduce noise before sharpening.
  • C-S is often restricted to the midtones to prevent blowout/blockup at the ends of the tonal spectrum early in the image editing process (we need to leave headroom for other operations). Since NX2 does not have tonal masking, applying sharpening to important subject areas via a Selection Control Point or by painting the result can work OK.
  • C-S is often restricted to edges via an edge mask so we don’t punch up noise or other artifacts which show up mainly in the smooth areas. The only way I know how to do this in NX2 is with High Pass sharpening.
  • C-S is applied gently since more sharpening is applied later.
More details on C-S and the other phases can be found in Bruce’s CS2/CS3 books or his book on sharpening. Note that this info is somewhat tangential to the OP’s question since it concerned global sharpening.

Also, for perspective, authors such as Thom Hogan generally sharpens twice and not three times. I don’t think he mentions that it in his article on sharpening, which is found here: http://bythom.com/sharpening.htm and is worth reading.

Jim


Jim
That's really helpful, Jim. You provided some information that is new to me.

One quick question: You mention that NX2 doesn't have "tonal masking," a term that I'm not familiar with. You're probably aware that both NX and NX2 provide the ability to independently sharpen each of the color channels using USM. I realize that each channel can have very dark tones and very bright tones that can't be dealt with separately from the mid tones. Is that why you mention the inability to mask the tones?

Thank you mentioning that the OP asked about global sharpening. If all of the color channels are sharpened separately, the result is global sharpening. At least that's what I think is the case so I don't have to feel guilty about being off-topic. :biggrin:
 
Joined
Jul 1, 2008
Messages
67
Location
U.S.A.
Capture Sharpening

Mike:

Programs like Picture Window Pro can make masks based directly on tonal ranges. So to mask the midtones we construct a masking curve that selects all pixels in the brightness range of say, 30% to 70%. Using this definition of midtones, the shadows would then be 0% to 30% and the highlights would be 70%-100% brightness level. To make the brightness determination, PWP probably converts the image to HSB representation so B = Brightness is determined from a single channel.

The ranges I use in this example are arbitrary and we can select whatever ranges we want for a particular image. The resulting mask can be blurred/feathered or modified manually with other tools.

I am sure that Photoshop has the same capability, but I don't know the name of the tool.

Jim
 
Last edited by a moderator:
F

fgaston

Guest
With regard to tonal masking and sharpening only the midtones, I was reading a review about a PS plug-in called FocalBlade and it talks about sharpening only the luminosity channel, not any of the color channels. I assume since Jason has not suggested it, it wouldn't apply to NX, but I thought I would throw it out there for discussion. If it doesn't apply to NX, I'd be interested if anyone has found an equivalent way in NX to sharpen only midtones. Fred
 
Joined
Jul 1, 2008
Messages
67
Location
U.S.A.
Sharpening Midtones

Fred:

One way to approach the problem is to set negative Selection Control Points on the highlights and on the shadows and then apply USM to the rest of the image. This approach might work better than a positive selection control point set on something important in the image that is midtone in brightness.

My thought is that the highlights and shadows are more Hue and Saturation neutral than some point with a midtone brightness. So setting the point on the later might bring in more than we want since the implicit mask set by the control points algorithm presumably matches on Hue, Saturation, and Brightness.

I have not experimented with this since it just occured to me. Also, NX is supposed to sharpen only on the Luminance channel to avoid Hue shifts that sometimes occur when sharpening in RGB.

Jim

From NX2 Help: "The Unsharp Mask tool in Capture NX 2 is unique in that it always applies its sharpening to the luminosity of the image, which prevents any unwanted color shifts."
 
Last edited by a moderator:
Joined
Jun 10, 2008
Messages
61
Location
United States
Cool ;-)

Fred:

One way to approach the problem is to set negative Selection Control Points on the highlights and on the shadows and then apply USM to the rest of the image. This approach might work better than a positive selection control point set on something important in the image that is midtone in brightness.

My thought is that the highlights and shadows are more Hue and Saturation neutral than some point with a midtone brightness. So setting the point on the later might bring in more than we want since the implicit mask set by the control points algorithm presumably matches on Hue, Saturation, and Brightness.

I have not experimented with this since it just occured to me. Also, NX is supposed to sharpen only on the Luminance channel to avoid Hue shifts that sometimes occur when sharpening in RGB.
This works really well! Since I wanted pretty much the same "mask" for each, I linked several progressive sharpening steps (including capture, selective and output) and used negative control points to remove highlights and shadows from the mask. It's easy to place the points at the the bright and dark spots using double threshold in the histogram - once the points are placed, turn on the mask display and fine-tune to get just the mask you want!
 
F

fgaston

Guest
Linked Sharpening Steps

This works really well! Since I wanted pretty much the same "mask" for each, I linked several progressive sharpening steps (including capture, selective and output) and used negative control points to remove highlights and shadows from the mask. It's easy to place the points at the the bright and dark spots using double threshold in the histogram - once the points are placed, turn on the mask display and fine-tune to get just the mask you want!
John:
Have you just linked Jason's steps together--if not would you be willing to share details of the steps? Thanks, Fred
 
Joined
Oct 14, 2006
Messages
254
Location
Colorado, USA
This works really well! Since I wanted pretty much the same "mask" for each, I linked several progressive sharpening steps (including capture, selective and output) and used negative control points to remove highlights and shadows from the mask. It's easy to place the points at the the bright and dark spots using double threshold in the histogram - once the points are placed, turn on the mask display and fine-tune to get just the mask you want!
John-
While I can understand wanting to use the same mask for different sharpening routines, I'm a little confused as to why you would sharpen three times simultaneously.

While I am an advocate of multi-stage sharpening, each stage has a different purpose and comes in at a different place in the editing workflow.

Capture Sharpening: I do this on all my images, typically globally.
Creative Sharpening: I rarely do this, but sometimes it helps enhance an image (like sharpening eyes).
Output sharpening: I only do this after I've resized an image (especially downsampling) or when I'm ready to print. There is no value to applying output sharpening to an image if you aren't "presenting" it (on-screen or in print).

I think it is VERY easy to go overboard with sharpening. The essence of multi-stage sharpening is to work around the fact that sometimes, one size does not fit all.

-Jason
 
Joined
Jun 10, 2008
Messages
61
Location
United States
John-
While I can understand wanting to use the same mask for different sharpening routines, I'm a little confused as to why you would sharpen three times simultaneously.

While I am an advocate of multi-stage sharpening, each stage has a different purpose and comes in at a different place in the editing workflow.

Capture Sharpening: I do this on all my images, typically globally.
Creative Sharpening: I rarely do this, but sometimes it helps enhance an image (like sharpening eyes).
Output sharpening: I only do this after I've resized an image (especially downsampling) or when I'm ready to print. There is no value to applying output sharpening to an image if you aren't "presenting" it (on-screen or in print).

I think it is VERY easy to go overboard with sharpening. The essence of multi-stage sharpening is to work around the fact that sometimes, one size does not fit all.

-Jason
I agree, to a point. For *me*, I find that I'm not really interested in applying "capture" sharpening (RGB USM at 2%) globally - there is no need to sharpen the sky, for example, or deep shadows. Since my selective or "creative" sharpening (separate R/G/B USM at 4%, varying intensities) is also usually applied on the subject of my photo (and not the background or any other uninteresting or light / dark areas), I usually wind up sharpening the same area as the capture sharpening. And my output sharpening (High Pass at 1px) is, again, typically applied to the same areas (again, no need to sharpen out-of-focus areas, the sky or deep shadows).

So for me, 90% of the time, the same mask conveniently applies to all three sharpening passes and usually gets me the results *I* want. When it doesn't, I'm all for applying them separately.
 
Joined
Sep 13, 2007
Messages
15,405
Location
Northern VA suburb of Washington, DC
I'm not really interested in applying "capture" sharpening (RGB USM at 2%) globally - there is no need to sharpen the sky, for example, or deep shadows.
That's a really great point.

On the other hand, I've found that when the in-camera exposure is accurate there's no harm in sharpening those areas, especially the sky. As a result, I usually take the lazy way out and sharpen globally. But only because it's the fast (okay, lazy) method.
 
Joined
May 12, 2008
Messages
639
Location
Greenville, South Carolina
I was wondering how one tells if sharpening has been carried too far?

I have heard the term "halos" are the result of over sharpening?

Not knowing what "halos" are, can someone show or explain to me halos and over sharpening?

A over sharpnend picture would help greatly.

Thanks,

MikeT
 
Joined
Sep 13, 2007
Messages
15,405
Location
Northern VA suburb of Washington, DC
A over sharpnend picture would help greatly.
The first picture was sharpened years ago when I first got into digital photography. Notice the bright halo around the head, especially the Buddha's left side of the head. Notice also the grainy texture of the body.

Subscribe to see EXIF info for this image (if available)


The image below is a different crop, but it's from the same file and it's properly sharpened.

View attachment 224359
 
Joined
May 12, 2008
Messages
639
Location
Greenville, South Carolina
MikeB,
Thanks! This is a very good example of "halos" from oversharpening. I can see the "halos" over the left shoulder as well. I have had some of my pictures look slightly grainy after sharpening and went back to redo the sharpening because it didn't look right. I haven't seen the halos until now.

Now I know what to look for.

Thanks again MikeB!

MikeT
 

Latest posts

Links on this page may be to our affiliates. Sales through affiliate links may benefit this site.
Nikon Cafe is a fan site and not associated with Nikon Corporation.
Forum post reactions by Twemoji: https://github.com/twitter/twemoji
Copyright © 2005-2019 Amin Forums, LLC
Top Bottom