Ad
Ad
Ad
Pages: [1]   Bottom of Page
Print
Author Topic: ACR 5.7 and Deconvolution Sharpening for Diffraction  (Read 7154 times)
markgoble
Newbie
*
Offline Offline

Posts: 36


« on: April 25, 2012, 09:30:28 AM »
ReplyReply

Hi,

Does ACR 5.7 have the ability to do Deconvolution Sharpening for Diffraction. I normally shoot at f16 or f22 for maximum DOF, with a 5dm2. I'm just starting to learn sharpening techniques, and would like to RESTORE some of the degrading effects of diffraction.

Thank you in advance for your help.

Mark
Logged
AlanPezzulich
Newbie
*
Offline Offline

Posts: 19


« Reply #1 on: April 25, 2012, 05:28:13 PM »
ReplyReply

No. Deconvolution tries to reconstruct an image that is out of focus. There is not much out there that does a reasonable job using deconvolution. NASA used it to fix pictures from the Hubble telescope. THey had the full specs for the lens and also knew exactly what the focusing error was. If I remember correctly it took 6 hours on a supercomputer to deconvolve one picture.
Logged
Schewe
Sr. Member
****
Offline Offline

Posts: 5469


WWW
« Reply #2 on: April 25, 2012, 06:01:01 PM »
ReplyReply

Does ACR 5.7 have the ability to do Deconvolution Sharpening for Diffraction.

Actually, I think it does...what I'm NOT sure is what version it went in...I "think" it was ACR 5.7 but it may have been ACR 6. The deconvolution kicks in when you run the Detail slider up. If you move the slider to 100, it's all deconvolution similar to what Smart Sharpen does in Photoshop when set to remove Lens Blur (which is the PSF ACR uses).
Logged
Keith Reeder
Full Member
***
Offline Offline

Posts: 242


WWW
« Reply #3 on: April 26, 2012, 04:45:03 AM »
ReplyReply

There is not much out there that does a reasonable job using deconvolution.

Sure there is, Alan. Richardson-Lucy Deconvolution is a commonly-enough used sharpening algorithm.

As Jeff says, deconvolution can be called from ACR/PS, and I use Raw Therapee's RL Deconvolution sharpening algorithm on a regular basis.

Interesting and informative thread on the subject here: http://www.luminous-landscape.com/forum/index.php?topic=45038.0
« Last Edit: April 26, 2012, 05:04:09 AM by Keith Reeder » Logged

Keith Reeder
Blyth, NE England
markgoble
Newbie
*
Offline Offline

Posts: 36


« Reply #4 on: April 26, 2012, 10:23:11 AM »
ReplyReply

Actually, I think it does...what I'm NOT sure is what version it went in...I "think" it was ACR 5.7 but it may have been ACR 6. The deconvolution kicks in when you run the Detail slider up. If you move the slider to 100, it's all deconvolution similar to what Smart Sharpen does in Photoshop when set to remove Lens Blur (which is the PSF ACR uses).

Jeff,

Does, (2nd Edition) Real World Image Sharpening with Adobe Photoshop, Camera Raw, and Lightroom, cover ACR 5.7 CS4? I need to learn this stuff, it seems like my old darkroom days were so simple.


Logged
Schewe
Sr. Member
****
Offline Offline

Posts: 5469


WWW
« Reply #5 on: April 26, 2012, 12:19:02 PM »
ReplyReply

Does, (2nd Edition) Real World Image Sharpening with Adobe Photoshop, Camera Raw, and Lightroom, cover ACR 5.7 CS4?

Yes...but the exact description of the detail slider is wrong. Eric had changed what algorithms the detail slider tweened...so remember that towards 100 is deconvolution :~)
Logged
bjanes
Sr. Member
****
Online Online

Posts: 2783



« Reply #6 on: April 26, 2012, 02:00:04 PM »
ReplyReply

Sure there is, Alan. Richardson-Lucy Deconvolution is a commonly-enough used sharpening algorithm.

As Jeff says, deconvolution can be called from ACR/PS, and I use Raw Therapee's RL Deconvolution sharpening algorithm on a regular basis.

Interesting and informative thread on the subject here: http://www.luminous-landscape.com/forum/index.php?topic=45038.0

Yes, the deconvolution sharpening with ACR (detail all the way to the right) does a good job of improving softness induced by a low pass filter, but I don't know if the deconvolution algorithm is optimal. It must be relatively abbreviated, since it is much faster than 20 iterations of RL in Raw Therapee. Jeff says the PSF used by ACR is to remove lens blur, but I don't really know what lens blur is--defocus, spherical aberration, other? Since blurring in the image results from multiple sources, Bart van der Wolf (our resident expert in deconvolution) recommends a Gaussian PSF.

Regards,

Bill
Logged
Schewe
Sr. Member
****
Offline Offline

Posts: 5469


WWW
« Reply #7 on: April 26, 2012, 02:35:52 PM »
ReplyReply

Jeff says the PSF used by ACR is to remove lens blur, but I don't really know what lens blur is--defocus, spherical aberration, other?

All I know is that Eric has said the PSF is similar to the Lens Blur option in Smart Sharpen...
Logged
Keith Reeder
Full Member
***
Offline Offline

Posts: 242


WWW
« Reply #8 on: April 26, 2012, 04:38:18 PM »
ReplyReply

Hi Bill,

as you might imagine I can't comment on the implementation decisions involved here - I'm just bringing to Alan's attention that there are more deconvolution opportunities out there than I think he realised.
Logged

Keith Reeder
Blyth, NE England
markgoble
Newbie
*
Offline Offline

Posts: 36


« Reply #9 on: April 26, 2012, 05:50:34 PM »
ReplyReply

Very good information,
Thank you
Logged
ErikKaffehr
Sr. Member
****
Offline Offline

Posts: 7332


WWW
« Reply #10 on: April 26, 2012, 10:15:06 PM »
ReplyReply

Hi,

In my view you would use lens blur to correct for defocus and gaussian to correct for diffraction. Why? Defocus would essentially correspond to a small uniformly lit disk. Diffraction is quite similar to a bell curve.

Best regards
Erik

Yes, the deconvolution sharpening with ACR (detail all the way to the right) does a good job of improving softness induced by a low pass filter, but I don't know if the deconvolution algorithm is optimal. It must be relatively abbreviated, since it is much faster than 20 iterations of RL in Raw Therapee. Jeff says the PSF used by ACR is to remove lens blur, but I don't really know what lens blur is--defocus, spherical aberration, other? Since blurring in the image results from multiple sources, Bart van der Wolf (our resident expert in deconvolution) recommends a Gaussian PSF.

Regards,

Bill
« Last Edit: April 26, 2012, 10:17:36 PM by ErikKaffehr » Logged

BartvanderWolf
Sr. Member
****
Offline Offline

Posts: 3579


« Reply #11 on: April 27, 2012, 02:44:55 AM »
ReplyReply

In my view you would use lens blur to correct for defocus and gaussian to correct for diffraction. Why? Defocus would essentially correspond to a small uniformly lit disk. Diffraction is quite similar to a bell curve.

Hi Erik,

That would work as a first approach, but I would also try the other option. The problem is that we do not know exactly which algorithms are used. There is a likelihood that the Gaussian type produces more halo, especially when the required radius and amount need to be found by trial and error. I can imagine that the lens blur type of sharpening (in more accurate mode) is potentially somewhat less susceptible to creating such halo artifacts if the correct radius is used. Therefore, i'd say try them both (or a mix) because we are rarely confronted with only one of the blur types, it's usually a mix.

Cheers,
Bart

Logged
Pete Berry
Sr. Member
****
Offline Offline

Posts: 269


« Reply #12 on: April 27, 2012, 12:47:30 PM »
ReplyReply

All I know is that Eric has said the PSF is similar to the Lens Blur option in Smart Sharpen...

With the main difference I can see that Smart Sharpen lacks a masking or threshold filter to protect smoother areas from noise accentuation. I've got to wonder why not as in ACR, and CS5's USM?

Pete
Logged
Keith Reeder
Full Member
***
Offline Offline

Posts: 242


WWW
« Reply #13 on: April 29, 2012, 02:59:41 PM »
ReplyReply

Smart Sharpen has fade sliders for highlights and shadows, Pete - they do a good job of protecting smooth areas from unwanted sharpening.
Logged

Keith Reeder
Blyth, NE England
Pages: [1]   Top of Page
Print
Jump to:  

Ad
Ad
Ad