Ugly Hedgehog - Photography Forum
Home Active Topics Newest Pictures Search Login Register
For Your Consideration
Reality check: Deconvolution Sharpening.
Page 1 of 2 next>
Feb 13, 2016 12:06:04   #
Uuglypher Loc: South Dakota (East River)
 
Deconvolution Sharpening:

The statement was recently made that:
"Sharpening is not significantly effective at overcoming blur from out of focus, from poor lenses, or from camera shake."

This is simply a statement made from ignorance of the facts. If the statement had been qualified by "Edge sharpening..." it would have been true. However, the technique of deconvolutional sharpening is designed to sharpen images suffering from exactly those sorts of problems.

The algorithms of Deconvolution Sharpening are those that specifically determine the "point spread factor" (PSF) from the different sorts of image data convolution(dispersion) and then "reverse engineer" re-aggregation of the dispersed image data.
The classic basis of deconvolution sharpening is the "Richardson-Lucy" algorithm that has long been used by astronomers to re-aggregate image data of their typical "point-source" subjects (stars!). In Photoshop the sharpening filters based on deconvolution are "Smart a Sharpen" and "Shake Reduction". In ACR/Lightroom the "Detail" slider (to the right of its midpoint of excursion) utilizes deconvolution of image data which tends to brighten edge-sharpened halos to positive or negative effect depending, of course, on image size and halo width.

it should be noted that deconvolution sharpening is time- and memory-intensive !

Anyone with experience with Deconvolution sharpening care to comment...and perhaps post some before and after images?

Dave

Deconvolution Sharpening
Deconvolution Sharpening...
(Download)

Deconvolution Sharpening..workflow
Deconvolution Sharpening..workflow...
(Download)

Smart Sharpen
Smart Sharpen...
(Download)

Reply
Feb 13, 2016 19:59:15   #
minniev Loc: MIssissippi
 
Uuglypher wrote:
Deconvolution Sharpening:

The statement was recently made that:
"Sharpening is not significantly effective at overcoming blur from out of focus, from poor lenses, or from camera shake."

This is simply a statement made from ignorance of the facts. If the statement had been qualified by "Edge sharpening..." it would have been true. However, the technique of deconvolutional sharpening is designed to sharpen images suffering from exactly those sorts of problems.

The algorithms of Deconvolution Sharpening are those that specifically determine the "point spread factor" (PSF) from the different sorts of image data convolution(dispersion) and then "reverse engineer" re-aggregation of the dispersed image data.
The classic basis of deconvolution sharpening is the "Richardson-Lucy" algorithm that has long been used by astronomers to re-aggregate image data of their typical "point-source" subjects (stars!). In Photoshop the sharpening filters based on deconvolution are "Smart a Sharpen" and "Shake Reduction". In ACR/Lightroom the "Detail" slider (to the right of its midpoint of excursion) utilizes deconvolution of image data which tends to brighten edge-sharpened halos to positive or negative effect depending, of course, on image size and halo width.

it should be noted that deconvolution sharpening is time- and memory-intensive !

Anyone with experience with Deconvolution sharpening care to comment...and perhaps post some before and after images?

Dave
Deconvolution Sharpening: br br The statement was... (show quote)


I'll bite. But I don't claim experience, just a willingness to experiment. These aren't great photos by any means but I didn't have a picture of a guinea and I wanted one so I took the best of a bad lot and worked on it in PS with the shake reduction filter, which I am quite willing to try on such occasions. Sometimes the magic works, sometimes it doesn't, but it is worth a try if you don't have an option to reshoot.

soft guinea
soft guinea...
(Download)

slightly improved guinea
slightly improved guinea...
(Download)

Reply
Feb 13, 2016 22:17:03   #
Uuglypher Loc: South Dakota (East River)
 
minniev wrote:
I'll bite. But I don't claim experience, just a willingness to experiment. These aren't great photos by any means but I didn't have a picture of a guinea and I wanted one so I took the best of a bad lot and worked on it in PS with the shake reduction filter, which I am quite willing to try on such occasions. Sometimes the magic works, sometimes it doesn't, but it is worth a try if you don't have an option to reshoot.


hi, min,
That is definitely one of the better results I've seen with Shake Reduction!
Really nice job.
Dave

Reply
 
 
Feb 13, 2016 23:52:41   #
Billyspad Loc: The Philippines
 
A very worthwhile and well constructed tutorial Dave

Reply
Feb 14, 2016 01:09:00   #
Apaflo Loc: Anchorage, Alaska
 
Billyspad wrote:
A very worthwhile and well constructed tutorial Dave

Take a picture of a ruler as is done commonly to see DOF or test auto focus. Then try using deconvolution to remove blur from out of focus. Just try to get, for example, the equivalent of 2 fstops more DOF.

2 fstops just happens to be the amount of diffraction blur that can easily be removed using a sharpen tool.

All of a sudden the words "not significantly effective" in the statement Dave started off by objecting to become very pertinent. You can't get significant benefit without investing more than it is worth... unless you are NASA and have a Hubble telescope that is out of focus. All it took to refocus it with software was a few millions of dollars over several months.

The fact is that other than astronomers there are very few people finding a lot of use for Richardson-Lucy Deconvolution, or any of the variants. Typically what you get with general photography, after significant effort, is barely enough effect to be able to say it had an effect greater than other methods.

Reply
Feb 14, 2016 01:27:57   #
Billyspad Loc: The Philippines
 
Apaflo wrote:
Take a picture of a ruler as is done commonly to see DOF or test auto focus. Then try using deconvolution to remove blur from out of focus. Just try to get, for example, the equivalent of 2 fstops more DOF.

2 fstops just happens to be the amount of diffraction blur that can easily be removed using a sharpen tool.

All of a sudden the words "not significantly effective" in the statement Dave started off by objecting to become very pertinent. You can't get significant benefit without investing more than it is worth... unless you are NASA and have a Hubble telescope that is out of focus. All it took to refocus it with software was a few millions of dollars over several months.

The fact is that other than astronomers there are very few people finding a lot of use for Richardson-Lucy Deconvolution, or any of the variants. Typically what you get with general photography, after significant effort, is barely enough effect to be able to say it had an effect greater than other methods.
Take a picture of a ruler as is done commonly to s... (show quote)


Not an expert on such things Floyd but a download of Daves example makes quite a compelling argument.
What worries me more is that its common knowledge you and Dave have issues. You ban him from your forum

Reply
Feb 14, 2016 01:27:57   #
Billyspad Loc: The Philippines
 
Apaflo wrote:
Take a picture of a ruler as is done commonly to see DOF or test auto focus. Then try using deconvolution to remove blur from out of focus. Just try to get, for example, the equivalent of 2 fstops more DOF.

2 fstops just happens to be the amount of diffraction blur that can easily be removed using a sharpen tool.

All of a sudden the words "not significantly effective" in the statement Dave started off by objecting to become very pertinent. You can't get significant benefit without investing more than it is worth... unless you are NASA and have a Hubble telescope that is out of focus. All it took to refocus it with software was a few millions of dollars over several months.

The fact is that other than astronomers there are very few people finding a lot of use for Richardson-Lucy Deconvolution, or any of the variants. Typically what you get with general photography, after significant effort, is barely enough effect to be able to say it had an effect greater than other methods.
Take a picture of a ruler as is done commonly to s... (show quote)


Not an expert on such things Floyd but a download of Daves example makes quite a compelling argument.
What worries me more is that its common knowledge you and Dave have issues. You ban him from your forum

Reply
 
 
Feb 14, 2016 01:27:58   #
Billyspad Loc: The Philippines
 
Apaflo wrote:
Take a picture of a ruler as is done commonly to see DOF or test auto focus. Then try using deconvolution to remove blur from out of focus. Just try to get, for example, the equivalent of 2 fstops more DOF.

2 fstops just happens to be the amount of diffraction blur that can easily be removed using a sharpen tool.

All of a sudden the words "not significantly effective" in the statement Dave started off by objecting to become very pertinent. You can't get significant benefit without investing more than it is worth... unless you are NASA and have a Hubble telescope that is out of focus. All it took to refocus it with software was a few millions of dollars over several months.

The fact is that other than astronomers there are very few people finding a lot of use for Richardson-Lucy Deconvolution, or any of the variants. Typically what you get with general photography, after significant effort, is barely enough effect to be able to say it had an effect greater than other methods.
Take a picture of a ruler as is done commonly to s... (show quote)


Not an expert on such things Floyd but a download of Daves example makes quite a compelling argument.
What worries me more is that its common knowledge you and Dave have issues. You ban him from your forum and he pops up having started what some would view as a forum in direct competition with yours. Your argument here smells as if its personal and if that's the case apart from it being downright rude its not going to get you any respect.
As a friend Floyd this is not the place for vendettas, your above that.

Reply
Feb 14, 2016 03:29:19   #
Apaflo Loc: Anchorage, Alaska
 
Billyspad wrote:
Not an expert on such things Floyd but a download of Daves example makes quite a compelling argument.
What worries me more is that its common knowledge you and Dave have issues. You ban him from your forum and he pops up having started what some would view as a forum in direct competition with yours. Your argument here smells as if its personal and if that's the case apart from it being downright rude its not going to get you any respect.
As a friend Floyd this is not the place for vendettas, your above that.
Not an expert on such things Floyd but a download ... (show quote)

What's the need for you to argue personalities?

The fact is that with the Photoshop methods shown it can use deconvolution, but you cannot tell if that is what causes benefits seen, as opposed to either USM or HP Sharpening which is also being applied. Or, it usually is difficult for most people, but in this case it is very clear:

The detailed description provided very clearly indicates

the "Advanced" Highlight and Shadow controls really aren't
worth the fussing and fiddling trouble.

Since that is in fact the enable switch for use of deconvolution, Dave himself is repeating exactly what I said originally: no significant benefit.

Deconvolution sharpening has been around since the 1970's. It is not some new wonder dog. There is a very good reason that astronomers use it a great deal, as opposed to those doing general photography. For it to work well requires a Point Spread Function that closely matches the profile of equipment that produced the image. It also works well with a high signal to noise ratio, not so well with high ISO images.

Deconvolution just doesn't provide that significant a cost to benefit ratio for general purpose photography. Otherwise, just as everyone here has heard of Unsharp Mask, everyone would know how to invoke Richardson-Lucy Sharpen in whatever editor they use.

Reply
Feb 14, 2016 05:09:13   #
Uuglypher Loc: South Dakota (East River)
 
The fact remains that Deconvolution sharpening has been undergoing refinements for years as a component of the sharpening techniques of a number of image processing applications and plug-ins. These are routinely used by many photographers who have expanded their sharpening repertoires far beyond the old standards of unsharp mask and high-pass blend sharpening, which, although still of utility, are far from being "...the only games in town" they once were.

You will find the opportunity to use deconvolution sharpening in:
Topaz "In Focus"
Adobe Photoshop "Smart Sharpen"
Adobe Photoshop "Shake Reduction"
"Raw Therapee"
"Irident (Raw) Developer"
"Focus Magic"
Adobe Camera Raw/Lightroom (upper half of excursion of "Detail" slider

You may already have been using Deconvolution sharpening without realizing it and have come to favorit because of far less risk of objectionable halos.

If you haven't yet used deconvolution, try it; you may like it!

Reply
Feb 14, 2016 09:57:23   #
346pak Loc: Texas
 
Great idea thanks for sharing that. I am going to have to try that.

Reply
 
 
Feb 14, 2016 10:31:18   #
pfrancke Loc: cold Maine
 
Uuglypher wrote:
The fact remains that Deconvolution sharpening has been undergoing refinements for years as a component of the sharpening techniques of a number of image processing applications and plug-ins. These are routinely used by many photographers who have expanded their sharpening repertoires far beyond the old standards of unsharp mask and high-pass blend sharpening, which, although still of utility, are far from being "...the only games in town" they once were.

You will find the opportunity to use deconvolution sharpening in:
Topaz "In Focus"
Adobe Photoshop "Smart Sharpen"
Adobe Photoshop "Shake Reduction"
"Raw Therapee"
"Irident (Raw) Developer"
"Focus Magic"
Adobe Camera Raw/Lightroom (upper half of excursion of "Detail" slider

You may already have been using Deconvolution sharpening without realizing it and have come to favorit because of far less risk of objectionable halos.

If you haven't yet used deconvolution, try it; you may like it!
The fact remains that Deconvolution sharpening has... (show quote)


Thank you Dave! And as always, if there is a particular area that benefits, then with masking you can selectively take what you need.

Reply
Feb 14, 2016 10:42:05   #
Apaflo Loc: Anchorage, Alaska
 
346pak wrote:
I am going to have to try that.

The difficulty seems to be in actually trying it.

Adobe editing tools make an effort to hide exactly what is being used. For example both Smart Sharpen and their "shake reduction filter" filter have been used as examples, but in neither case were the people who described successful editing aware of whether they were using deconvolution or not. In both cases it is very likely the examples shown are not the result of deconvolution, but rather a mixture of Unsharp Mask and an High Pass filter sharpen algorithm.

We are left with two people, Dave and myself, who say they have positively used deconvolution. We came to the same conclusion too. Dave said, the tools "really aren't worth the fussing and fiddling trouble", which is the same as where I had said it isn't "significantly effective".

Rather than use Adobe tools, try deconvolution in a tool package that doesn't mix it with other types of sharpening, and lets you evaluate just what deconvolution does. Compare it to a High Pass sharpen and an Unsharp Mask tool. For that matter, also try a Wavelet Sharpen too, which is another method at about the same level as deconvolution.

All that can be done without investing money. The G'Mic package for GIMP has those and others.

Reply
Feb 14, 2016 12:20:43   #
John_F Loc: Minneapolis, MN
 
Billyspad wrote:
....

You ban him from your forum and he pops up having started what some would view as a forum in direct competition with yours. ....


I didn't know before that individuals could erect their own personal forums and ban some people. I would hope the UHH would label these private preserves so the unwary like I not wander in expecting objective commentary. The private preserves should be identified upfront.

Reply
Feb 14, 2016 12:31:12   #
Apaflo Loc: Anchorage, Alaska
 
John_F wrote:
I didn't know before that individuals could erect their own personal forums and ban some people. I would hope the UHH would label these private preserves so the unwary like I not wander in expecting objective commentary. The private preserves should be identified upfront.

Perhaps what Billyspad said is not quite truthful.

It is actually relatively easy to be banned from any Section, or even from this entire forum. All you need to do is repeatedly and willfully abuse the posted rules.

Look up the most basic set of rules that Admin has posted, then try the patience of everyone by breaking them on purpose and saying nobody can stop you... see if nothing happens.

Reply
Page 1 of 2 next>
If you want to reply, then register here. Registration is free and your account is created instantly, so you can post right away.
For Your Consideration
UglyHedgehog.com - Forum
Copyright 2011-2024 Ugly Hedgehog, Inc.