Skip navigation
ACRFREAK
ACRFREAK Calculating status...
Currently Being Moderated

anti deconvolution - ACR Capture Sharpening

Jul 27, 2011 6:49 PM

anti deconvolution - ACR Capture Sharpening

 

I cannot remember where I came across this word "anti deconvolution" (likely years back debating here about how best to use Capture Sharpening). I am considering making training videos on "The Ultimate Sharpening Workflow for Fine Art Prints" and want to make sure what I am saying is sound and maybe even accurate! So I am double checking my facts.

 

I believe I was told by someone here that when the Capture Sharpening Radius is set to 0.5 and the detail set to 100 (with the appropriate amount dialed in) that it does something called "anti deconvolution" and can actually extract a very small amount of real detail (as apposed to simply sharpening halos) out of a raw file.

 

Is that a sound statement?

 

BTW, I am a professional landscape photographer (big fine art enlargements) and my preferred way to use capture sharpening is by double/triple processing raw files for various areas then masking them accordingly in PS. But since I am almost always trying to emphasize the ultra fine high frequency details in an image (except sky's and soft moving water) I have found that:

 

Amount (varying)

Radius 0.5

Detail 100

Masking  (varying)

 

Has worked best for me.

 

Cheers.

3354 Views   17 Replies   Latest reply: 01af, Sep 12, 2011 12:53 AM
 
Replies
  • Jeff Schewe
    Currently Being Moderated
    1. Jeff Schewe,
    Jul 27, 2011 8:35 PM   in reply to ACRFREAK
    Report

    It's not "anti deconvolution"...it's called simply deconvolution kernel sharpening...similar to the Smart Sharpen's lens correction. ACR, with Detail set above 50, is a simple sharpening using a generic point spread function (PSF). Ya might want to google it...oh, and it works regardless of the radius setting. The deconvolution is dependent on the Detail slider, not the radius.

     
    |
    Mark as:
  • ACRFREAK
    Currently Being Moderated
    2. ACRFREAK,
    Jul 27, 2011 11:35 PM   in reply to Jeff Schewe
    Report

    Oops, that's right, just deconvolution.  I'm doing a lot of research on it right now.

     
    |
    Mark as:
  • deejjjaaaa
    Currently Being Moderated
    3. deejjjaaaa,
    Jul 28, 2011 8:06 AM   in reply to Jeff Schewe
    Report

    Jeff

     

    > ACR, with Detail set above 50

     

    Eric Chan was saying that it is not a "switch" (for example @ 50) from one type to another type of sharpening but a gradual blend of USM and deconvolution sharpening when you start moving the slider from 0 to 100... do you have another information ? please clarify.

     
    |
    Mark as:
  • Yammer P
    Currently Being Moderated
    4. Yammer P,
    Jul 28, 2011 8:34 AM   in reply to deejjjaaaa
    Report

    deejjjaaaa wrote:

     

    Eric Chan was saying that it is not a "switch" (for example @ 50) from one type to another type of sharpening but a gradual blend of USM and deconvolution sharpening when you start moving the slider from 0 to 100

     

    I remember that too. Maybe the other type doesn't kick in until 50.

     
    |
    Mark as:
  • Jeff Schewe
    Currently Being Moderated
    5. Jeff Schewe,
    Jul 28, 2011 9:00 AM   in reply to deejjjaaaa
    Report

    deejjjaaaa wrote:

     

    Jeff

     

    Eric Chan was saying that it is not a "switch" (for example @ 50) from one type to another type of sharpening but a gradual blend of USM and deconvolution sharpening when you start moving the slider from 0 to 100...

     

    It's more complicated than that...but when you get to the 40-60 range (which is why I said 50) it becomes predominately but not exclusively deconvolution. The default setting of 25 I don't think has much. Also know that halo suppression also impacts the results. Again by about 50, there's little or no halo suppression.

     
    |
    Mark as:
  • Bill_Janes
    Currently Being Moderated
    6. Bill_Janes,
    Jul 28, 2011 4:49 PM   in reply to deejjjaaaa
    Report

    deejjjaaaa wrote:

     

    Jeff

     

    > ACR, with Detail set above 50

     

    Eric Chan was saying that it is not a "switch" (for example @ 50) from one type to another type of sharpening but a gradual blend of USM and deconvolution sharpening when you start moving the slider from 0 to 100... do you have another information ? please clarify.

     

    Here is Eric's link for his suggestion for landscapes:

     

    http://forums.adobe.com/message/3066522#3066522

     
    |
    Mark as:
  • Noel Carboni
    20,917 posts
    Dec 23, 2006
    Currently Being Moderated
    7. Noel Carboni,
    Jul 30, 2011 1:03 PM   in reply to ACRFREAK
    Report

    ACRFREAK wrote:

     

    I believe I was told by someone here that when the Capture Sharpening Radius is set to 0.5 and the detail set to 100 (with the appropriate amount dialed in) that it does something called "anti deconvolution" and can actually extract a very small amount of real detail (as apposed to simply sharpening halos) out of a raw file.

     

     

    I don't know if you've got the terms right, but those are EXACTLY the settings I have found, through much experimentation, bring out the most possible usable detail from my raw images.

     

    -Noel

     
    |
    Mark as:
  • ACRFREAK
    Currently Being Moderated
    8. ACRFREAK,
    Jul 30, 2011 2:40 PM   in reply to Noel Carboni
    Report

    Noel,

     

    Me too!

     

    I get the feeling that the way it works (R 0.5 / D 100) combined with the fact than deconvolution sharpening is being implemented that actual/real detail (even if minimal) is being accessed.

     

    Side note: I am even starting to experiment with an ultra, ultra small amount of very fine grain simulation (amount on 5 - almost completely undetectable) at the raw stage as a way to create the illusion of even more detail (I have not completed the testing on that).

     

    Of course for fine art printing (after the uprez and sharpening routine) a more aggressive grain simulation at various amounts and types are used as a last step to create the illusion of more detail in a print as well as a more organic look.

     

    Cheers

     
    |
    Mark as:
  • Noel Carboni
    20,917 posts
    Dec 23, 2006
    Currently Being Moderated
    9. Noel Carboni,
    Jul 30, 2011 3:14 PM   in reply to ACRFREAK
    Report

    If you're not already doing so, try converting directly to upsampled resolutions as well.

     

    -Noel

     
    |
    Mark as:
  • ACRFREAK
    Currently Being Moderated
    10. ACRFREAK,
    Aug 6, 2011 1:40 PM   in reply to Noel Carboni
    Report

    I bring the file into PS at it native res.  Finish it and save as a master.  Then once I am ready to make a print, I size it all the way up using smoother.  For typical DSLR shots and printing on the Chromira or the Lightjet, if the image is larger than about 30 inches I will res up to its final size at 200ppi.  If it is below 30 inches I choose 300ppi.  Then of course advanced sharpening is applied... Lastly grain.

     
    |
    Mark as:
  • ACRFREAK
    Currently Being Moderated
    11. ACRFREAK,
    Aug 6, 2011 1:41 PM   in reply to ACRFREAK
    Report

    In my testing I have seen no benefit to rezzing up in ACR then going up the rest of the way in PS.  I think Smoother does it best from native to max size.

     
    |
    Mark as:
  • Noel Carboni
    20,917 posts
    Dec 23, 2006
    Currently Being Moderated
    12. Noel Carboni,
    Aug 6, 2011 2:36 PM   in reply to ACRFREAK
    Report

    I think it boils down, then, to what kind of input is best for your "advanced sharpening" step.  I'm using my own fractal sharpening actions for that.

     

    -Noel

     
    |
    Mark as:
  • MadManChan2000
    Currently Being Moderated
    13. MadManChan2000,
    Aug 14, 2011 8:15 PM   in reply to Noel Carboni
    Report

    If Detail is set to 100 in PV 2010, you will be using only the deconv-based method in ACR.  Please use carefully and judiciously, as this will try to extract as much detail as possible with minimal regard for noise and other artifacts that may already be present in the image.  (Of course, if you find it is too strong in some areas of the image, you can use local adjustments, e.g., minus Sharpness to back off on the problematic areas.)  I do personally use Detail 100 often, but only in low ISO landscape images with very fine details (twigs, blades of grass, etc.).

     

    If set at a lower value (e.g., 40 to 60) you will be using a blend of the deconv-based method, and another (smoother) method, and I recommend that for more general purpose photography.

     

    Eric

     
    |
    Mark as:
  • Hudechrome
    Currently Being Moderated
    14. Hudechrome,
    Aug 16, 2011 8:50 PM   in reply to MadManChan2000
    Report

    Moving this kind of sharpening into PS and employing the Surface Blur in the range of 2 to 5 for both sliders takes some of the edge off images that are pushed to far. I frequently do this to avoid that digital sharp look.

     

    But then, I am drawn to the look of the early photographers like Julia Margart Cameron, or Lartigue, where his very sharp look is tempered by that softer overlay. Even certain images of Edward Weston possess this.

     

    Perhaps digital can return to that sensibility. For now, I'll tweak with the Surface Blur.

     
    |
    Mark as:
  • 01af
    Currently Being Moderated
    15. 01af,
    Sep 8, 2011 9:27 AM   in reply to MadManChan2000
    Report

    MadManChan2000 wrote:

     

    If Detail is set to 100 in PV 2010, you will be using only the deconv-based method in ACR.

    Huh? I am confused ... isn't the Detail slider gradually changing (blending) between Deconvolution-based Sharpening and Unsharp Masking, with the former at the slider's "0" end and USM at the "100" end!? And if Deconvolution-based Sharpening really is at "100" then what's at the "0" end? Certainly not USM ... but "another (smoother) method"? So USM (or anything similar to USM) is not involved at all? As I said—I am confused. Can Eric (or anyone) please shed a little light on this ...?

     
    |
    Mark as:
  • MadManChan2000
    Currently Being Moderated
    16. MadManChan2000,
    Sep 11, 2011 4:42 PM   in reply to 01af
    Report

    No.  The deconv-based method kicks in as Detail approaches 100.  Smaller values of Detail mean smoother sharpening (and no, it's not USM).  The important point is that ramping Detail up towards 100 means that ACR will be extracting as much fine detail as it can, and this will result in an image with more bite and texture. This can be good for low-ISO landscape or architecture work with very fine details, and very bad for portraits or other situations where you want smoother results.

     
    |
    Mark as:
  • 01af
    Currently Being Moderated
    17. 01af,
    Sep 12, 2011 12:53 AM   in reply to MadManChan2000
    Report

    I see ... so my understanding that the Detail slider changed between deconv-based sharpening (at 0) and unsharp-masking (at 100) was just plain wrong. That relieves my confusion. By the way, the practical usage of the Detail slider was clear always; I was just wondering about the underlying technology.

     
    |
    Mark as:
Actions

More Like This

  • Retrieving data ...

Bookmarked By (0)

Answers + Points = Status

  • 10 points awarded for Correct Answers
  • 5 points awarded for Helpful Answers
  • 10,000+ points
  • 1,001-10,000 points
  • 501-1,000 points
  • 5-500 points