Ad
Ad
Ad
Pages: [1] 2 »   Bottom of Page
Print
Author Topic: Sharpening - LR vs. Photokit  (Read 10627 times)
Phil Corley
Jr. Member
**
Offline Offline

Posts: 64


WWW
« on: August 24, 2013, 04:37:08 AM »
ReplyReply

Wonder if someone could help with my confusion (or maybe lack of knowledge)...

Taking the same landscape image and exporting in LR with Standard sharpening and then with the same image, edit in Photoshop CS5.1, resize and then output sharpening, medium edge using Photokit.

If all is equal and the image is correctly capture sharpened in LR, should the results be the same - given a lot of Photokit code was used in LR 5 sharpening?

If the PS outputted image is sharper, does this tell me something about the capture sharpening (or lack of) in LR?

Basically, I find I always get better results for my images doing the round trip into Photoshop (resize, sharpen) than just using LR, so I am wondering if this is due to something I am not getting right in LR or it is just better output sharpening using PS & Photokit than LR?

Appreciate all thoughts and opinions (and Jeff's view if possible)

Many thanks

Phil
Logged
David Sutton
Sr. Member
****
Offline Offline

Posts: 874


WWW
« Reply #1 on: August 24, 2013, 05:10:05 AM »
ReplyReply

If you are comparing the results of the output sharpening used in the export dialogue in Lightroom with the settings in Photokit on a resized file, I think you are comparing apples with oranges.
Are you sharpening for print or screen?

Edit: I wouldn't expect the two methods to look the same. But if you want to do a comparison then don't resize in Photoshop. That changes the characteristics of the file.
« Last Edit: August 24, 2013, 05:25:37 AM by David Sutton » Logged

Phil Corley
Jr. Member
**
Offline Offline

Posts: 64


WWW
« Reply #2 on: August 24, 2013, 07:49:10 AM »
ReplyReply

Both are being output for screen as JPEGs and the question really is; am I comparing apples with oranges or should they both be an apple?

True the resize in Photoshop will be different, but this is part of the question as well - with LR 5 I use the Export dialog to resize and sharpen, in Photoshop I use Image Size with Resample Image using Bicubic then output sharpen using Photokit; for web; medium edge.

I am just trying to understand why I am seeing a difference and am I causing this by less than optimum processing in LR

Thanks

Phil
Logged
luxborealis
Sr. Member
****
Offline Offline

Posts: 946



WWW
« Reply #3 on: August 24, 2013, 12:07:23 PM »
ReplyReply

If the PS+PhotoKit version is "better", what makes it so? Including examples of both full and 100% view crops of PS+PhotoKit and LR versions would be helpful for us to better understand your dilemma especially since your definition of "better" may be different from that of others.
Logged

Terry McDonald
Revealing the art inherent in nature
- visit luxBorealis.com.
Have a read of my PhotoBlog and subscribe!
David Sutton
Sr. Member
****
Offline Offline

Posts: 874


WWW
« Reply #4 on: August 24, 2013, 05:14:35 PM »
ReplyReply

Both are being output for screen as JPEGs and the question really is; am I comparing apples with oranges or should they both be an apple?

True the resize in Photoshop will be different, but this is part of the question as well - with LR 5 I use the Export dialog to resize and sharpen, in Photoshop I use Image Size with Resample Image using Bicubic then output sharpen using Photokit; for web; medium edge.

I am just trying to understand why I am seeing a difference and am I causing this by less than optimum processing in LR

Thanks

Phil
No, you have apples and oranges.
The jpeg conversion engines are different in Photoshop and Lightroom. The "quality" settings are different too.
Also, if you are not doing creative sharpening in Photoshop you are losing image quality. Possibly a lot of quality.
Furthermore, apart from sharpening for print, the whole sharpening process is somewhat subjective. That is part of what makes it interesting looking at other's work. Your "less than optimal" may be my "too crunchy".

Logged

RikkFlohr
Newbie
*
Online Online

Posts: 36


« Reply #5 on: August 25, 2013, 03:58:15 PM »
ReplyReply

Realize, of course that there is a well-publicized bug that occurs in LR 5.0 that has been fixed in LR 5.2 RC that affects the application of Noise Reduction and Sharpening upon export.

Read here: http://blogs.adobe.com/lightroomjournal/2013/06/lightroom-5-hot-issues.html

Are the images with which you are testing affected by this?
Logged

Rikk Flohr
Sagas in Light
Phil Corley
Jr. Member
**
Offline Offline

Posts: 64


WWW
« Reply #6 on: September 15, 2013, 11:04:22 AM »
ReplyReply

Sorry for the delay in getting back - been too busy :-(

I have put 5.2 RC on and it is now sharpening "better", however it still doesn't match the output from PS using Photokit.

So, as mentioned in this thread, I ruled out the resizing - as I resize in PS and then sharpen.  If I resize the image in PS and then export a version from LR the sharpening (at standard) is almost, if not, the same as Photokit is giving me on a Medium Edge.

It seems the visible difference (to me) is that resizing with PS provides a much better end result than LR - but don't get me wrong LR is good, it is just that for my images PS seems to be doing it better.

I suppose the question of this thread has changed to: How does LR do it's resizing and should there be a difference when compared against PS - using Resample: Bicubic Sharper (reduction) taking a D800E image down to 1000 pixels.

Appreciate all replies as it all helps my learning.

Thanks

Phil
Logged
luxborealis
Sr. Member
****
Offline Offline

Posts: 946



WWW
« Reply #7 on: September 15, 2013, 06:22:07 PM »
ReplyReply

I reiterate...
If the PS+PhotoKit version is "better", what makes it so? Including examples of both full and 100% view crops of PS+PhotoKit and LR versions would be helpful for us to better understand your dilemma especially since your definition of "better" may be different from that of others.
Logged

Terry McDonald
Revealing the art inherent in nature
- visit luxBorealis.com.
Have a read of my PhotoBlog and subscribe!
PhotoEcosse
Sr. Member
****
Offline Offline

Posts: 562



« Reply #8 on: September 16, 2013, 04:09:04 AM »
ReplyReply

Not directly related to the reason you are detecting differences in result, but.....

Do remember that one of the most common "faults" identified in digital images by competition and salon judges is "over-sharpening". You only have to do a quick scan of the images in the Landscape & Nature Photography section of this forum to find some horrific examples (as well as some superb images, of course).

I find that the standard Lightroom defaults give me pretty much the degree of sharpening I want for "normal" images. If I want to exaggerate the effect of sharpening for some dubious "creative" purpose, I tend to try for the effect locally by applying the "detail enhancer" filter in ColorEfexPro4 to the desired section(s) of the image with Control Points.

I think what I am saying, in a convoluted, round-about manner is "optimum sharpening and maximum sharpening are not the same thing".
Logged

************************************
"Reality is an illusion caused by lack of alcohol."
Alternatively, "Life begins at the far end of your comfort zone."
bjanes
Sr. Member
****
Offline Offline

Posts: 2756



« Reply #9 on: September 16, 2013, 08:38:21 AM »
ReplyReply

Wonder if someone could help with my confusion (or maybe lack of knowledge)...

Taking the same landscape image and exporting in LR with Standard sharpening and then with the same image, edit in Photoshop CS5.1, resize and then output sharpening, medium edge using Photokit.

If all is equal and the image is correctly capture sharpened in LR, should the results be the same - given a lot of Photokit code was used in LR 5 sharpening?

If the PS outputted image is sharper, does this tell me something about the capture sharpening (or lack of) in LR?

Basically, I find I always get better results for my images doing the round trip into Photoshop (resize, sharpen) than just using LR, so I am wondering if this is due to something I am not getting right in LR or it is just better output sharpening using PS & Photokit than LR?


The LR/ACR sharpening incorporates many of the features of Photokit, but the capture sharpening in LR/ACR has been improved by the addition of deconvolution. Deconvolution actually restores image detail rather than merely simulating sharpness by enhancing edge contrast (see Roger Clark). Photokit capture sharpening relies mainly on the unsharp mask, which is older technology. IMHO, one can get better results with capture sharpening by activating deconvolution in LR/ACR by moving the detail slider to the right. For landscape work, Eric Chan (who implemented the deconvolution in LR/ACR) often uses a low radius and moves the detail slider all the way to the right, maximizing the use of deconvolution.

Many experienced users prefer the use of specialized deconvoluiton restoration algorithms such as ImageMagic, Topaz InFocus, or Richardson-Lucy. The latter is not readily available as a plugin for Photoshop. Deconvolution can also be applied to output sharpening, but in my experience the Photokit and LR/ACR output sharpening works well. What have others experienced?

Regards,

Bill
Logged
BartvanderWolf
Sr. Member
****
Offline Offline

Posts: 3415


« Reply #10 on: September 20, 2013, 04:04:36 AM »
ReplyReply

Hi,

Output usually requires resampling of the original image size. Resampling will introduce artifacts, because we start with an abstraction of reality, pixels with a limited number of discrete values, and we create new pixels at a (sometimes fractionally) different position. This will generally also introduce a certain amount of blur. When we assume, although that is not a given, that the resampling operation produced relatively few (blocking/ringing/halo) artifacts, we still need to address the resulting blur. In addition to the resampling blur, we may need to pre-compensate for further blur caused by the output modality and medium used.

Many experienced users prefer the use of specialized deconvolution restoration algorithms such as ImageMagic, Topaz InFocus, or Richardson-Lucy. The latter is not readily available as a plugin for Photoshop.

One can also use RawTherapee to process TIFFs, and RT offers a decent implementation of the Richardson-Lucy deconvolution sharpening. Deconvolution is the most natural way of restoring actual resolution, instead of simply boosting local edge contrast, but it benefits from relatively low noise data as input.

Quote
Deconvolution can also be applied to output sharpening, but in my experience the Photokit and LR/ACR output sharpening works well. What have others experienced?

Photokit and other USM or High-pass based methods may be a convenient way of compensating for the loss of output contrast, especially when we upsample for large format output. However, nothing will look as natural as deconvolution, which attempts to exactly reverse the cause of the blur that reduces the contrast. Not only will it optimize the detail at the pixel level, but it will also boost the MTF curve of the entire image. It's as if a gray veil is lifted from the image, yet it will not boost intentionally hazy areas. It just restores the original signal, if applied in the correct amount.

The effect of deconvolution therefore depends very much on image content, and for some images the difference between traditional contrast boosting methods and real sharpening may be subtle. One of the benefits of deconvolution is that it can target the blur PSF that characterizes the resampling blur, pretty well.

Cheers,
Bart
Logged
Tim Lookingbill
Sr. Member
****
Offline Offline

Posts: 1144



WWW
« Reply #11 on: September 20, 2013, 03:11:37 PM »
ReplyReply


...Many experienced users prefer the use of specialized deconvoluiton restoration algorithms such as ImageMagic, Topaz InFocus, or Richardson-Lucy. The latter is not readily available as a plugin for Photoshop. Deconvolution can also be applied to output sharpening, but in my experience the Photokit and LR/ACR output sharpening works well. What have others experienced?

Regards,

Bill

I've been experimenting lately with LR4's sharpening algorithms on 100% previews of 6MP Pentax Raw PEFs for upscaling to 36in. poster size after recently purchasing LR from becoming accustomed to CS3's sharpened previews.

The results after finding the best sharpening combo viewing at 400% pixel size vew for upscaling to this size and exporting out of LR deliver impressive 100% tiff previews on landscape detail considering the extreme level of upsampling.

I find it helps in coming up with sharpening settings by viewing at a pixel level due to overcoming what I've gleamed from LR4's upscaling texture characterization on tiff results viewed at 100% using Glossy Paper/High Sharpening on export. I've noticed different output sharpening settings are required according to the level of detail on an image by image basis viewed at 400% in LR4.  

Below are comparison screengrabs where I applied an additional USM of low Amout/high Radius on the LR4 exported tiff in Photoshop. Remember this is a 6MP Raw. I could never get these results working in CS3 and upscaling in CS3 PS. Don't know if could get the same results out of CS5 ACR and upscaling in PS. I recall trying but remember getting zipper like textural artifacts. Exporting out of LR4 has made it much easier with better 100% previews.
« Last Edit: September 20, 2013, 03:14:21 PM by Tim Lookingbill » Logged
Tim Lookingbill
Sr. Member
****
Offline Offline

Posts: 1144



WWW
« Reply #12 on: September 20, 2013, 03:25:44 PM »
ReplyReply

Correction on the final sharpening in Photoshop on the tiff above. I used CS3 PS Smart Sharpen setting of 40Amount/20Radius-(Gaussian Blur)-Shadow/Highlight blending each set at 50/50.
« Last Edit: September 20, 2013, 03:31:02 PM by Tim Lookingbill » Logged
BartvanderWolf
Sr. Member
****
Offline Offline

Posts: 3415


« Reply #13 on: September 21, 2013, 05:12:34 AM »
ReplyReply

I've been experimenting lately with LR4's sharpening algorithms on 100% previews of 6MP Pentax Raw PEFs for upscaling to 36in. poster size after recently purchasing LR from becoming accustomed to CS3's sharpened previews.

The results after finding the best sharpening combo viewing at 400% pixel size vew for upscaling to this size and exporting out of LR deliver impressive 100% tiff previews on landscape detail considering the extreme level of upsampling.

Hi Tim,

Those look like reasonable sharpening settings for this type of subject. The optimal radius and amount do depend on the hardware, camera + lens and aperture used, so may require different settings for other combinations of those variables.

Quote
I could never get these results working in CS3 and upscaling in CS3 PS. Don't know if could get the same results out of CS5 ACR and upscaling in PS. I recall trying but remember getting zipper like textural artifacts. Exporting out of LR4 has made it much easier with better 100% previews.

It may have to do with the Bicubic Smoother resampling algorithm, which produces less pronounced artifacts and therefore allows to use more, or more accurate, sharpening. The type of subject matter makes a lot of difference as well, because with landscapes in general busy subject detail will allow to hide some of the artifacts. Modern architecture or similar subjects will show edge artifacts without mercy.

I've attached two test images that can be used for testing the upsampling algorithms, and for honing in on the output sharpening. It has all sorts of spatial frequencies, and a sharp slanted edge. The first image is the 'original scene', the second is a simulation of that scene after lens and capture/demosaicing blur. One should assign a profile, and resize with the same percentage that one would use to reach the native resolution required by the output modality, and convert to the output profile.

Sharpening after upsampling offers much more solid results, provided that the resampling doesn't introduce to many artifacts. By avoiding oversharpening earlier in the conversion process, one will create fewer issues when upsampling and sharpening for final output.

Cheers,
Bart
Logged
Bryan Conner
Sr. Member
****
Offline Offline

Posts: 514


WWW
« Reply #14 on: September 21, 2013, 06:00:56 AM »
ReplyReply

Thanks for the chart.  What is the recommended procedure for utilizing the chart?  Print and make a capture?  I agree with some of the other comments on FocusMagic.  It really is a great sharpening tool.  Thanks for sharing your knowledge here on the forum.
Logged

BartvanderWolf
Sr. Member
****
Offline Offline

Posts: 3415


« Reply #15 on: September 21, 2013, 06:46:13 AM »
ReplyReply

Thanks for the chart.  What is the recommended procedure for utilizing the chart?  Print and make a capture?  I agree with some of the other comments on FocusMagic.  It really is a great sharpening tool.  Thanks for sharing your knowledge here on the forum.

Hi Bryan,

One could view the Star_GB07+.png as a captured image, after Raw conversion, before capture sharpening.

That leaves 2 possible workflows.

One is to capture sharpen as one is accustomed to, then upsample for output and either apply output sharpening oneself, or let the application (e.g. Lightroom) add some. To take media influences into account it's best judged as printed output, but with a bit of experience one can also judge on screen.

Another workflow is to skip capture sharpening, upsample for output and apply output sharpening like above.

Depending on the quality of the upsampling, the second method will give cleaner results, especially when deconvolution output sharpening is used.

In my experience, the file with a 0.7 radius Gaussian Blur, is typical for a good lens capture with a modest aperture (modest diffraction). One can try other amounts of blur on the original which is not blurred, a Gaussian blur with a Radius o 1.0-1.2 approximates a narrow aperture effect. An actual diffraction pattern blur is more complex, but the combined effect of lens aberrations, aperture, AA-filter, sensel aperture, and demosaicing, does approximate a Gaussian blur.

Perfect sharpening of the blurred version should come close to looking like the un-blurred version, without jagged edges or halo and ringing artifacts. The amplitude and smoothness of the sinusoidal grating should be approximated as close as possible (one can try sharpening in difference layer blend mode in Photoshop to achieve that). To pre-compensate for the media losses (e.g. ink diffusion), one can then exaggerate the sharpening a bit at the output pixel level.

Cheers,
Bart
Logged
bjanes
Sr. Member
****
Offline Offline

Posts: 2756



« Reply #16 on: September 21, 2013, 09:55:00 AM »
ReplyReply


One can also use RawTherapee to process TIFFs, and RT offers a decent implementation of the Richardson-Lucy deconvolution sharpening. Deconvolution is the most natural way of restoring actual resolution, instead of simply boosting local edge contrast, but it benefits from relatively low noise data as input.

Photokit and other USM or High-pass based methods may be a convenient way of compensating for the loss of output contrast, especially when we upsample for large format output. However, nothing will look as natural as deconvolution, which attempts to exactly reverse the cause of the blur that reduces the contrast. Not only will it optimize the detail at the pixel level, but it will also boost the MTF curve of the entire image. It's as if a gray veil is lifted from the image, yet it will not boost intentionally hazy areas. It just restores the original signal, if applied in the correct amount.

The effect of deconvolution therefore depends very much on image content, and for some images the difference between traditional contrast boosting methods and real sharpening may be subtle. One of the benefits of deconvolution is that it can target the blur PSF that characterizes the resampling blur, pretty well.

Cheers,
Bart

Bart,

How does one implement output sharpening with deconvolution? A Gaussian PSP seems to work reasonably well with capture sharpening, but does it also apply to output sharpening? What radius and strength should one use with what algorithm and how does one judge the results other than by the laborious process of making test prints at the final output resolution. It is generally agreed that one can not judge the results on screen, since screen resolution is far below that of the printer.

In developing PhotoKit sharpener Bruce Frasier and his colleagues reportedly made literally hundreds of prints to judge the effects of various parameters. With regard to PhotoKit,
one can simulate the workflow (e.g. Glenn Mitchell, TLR), but the PhotoKit people say that such efforts lack the magic numbers (sharpening parameters) required for optimal results.

Although the two pass (with optimal creative sharpening) workflow of Frasier et al is widely accepted in the photographic community, you mentioned in a previous post that one could skip capture sharpening and sharpen only the final image. An interesting concept, but how would one implement such a workflow?

Regards,

Bill
Logged
BartvanderWolf
Sr. Member
****
Offline Offline

Posts: 3415


« Reply #17 on: September 21, 2013, 02:19:49 PM »
ReplyReply

How does one implement output sharpening with deconvolution? A Gaussian PSP seems to work reasonably well with capture sharpening, but does it also apply to output sharpening?

Hi Bill,

Yes, while never perfect, a Gaussian approximation comes a long way. This does assume that an algorithm is used that does a pretty straight forward interpolation. One can test for this by analyzing the slanted edge after interpolation, e.g. with my Slanted edge evaluation tool.

Quote
What radius and strength should one use with what algorithm and how does one judge the results other than by the laborious process of making test prints at the final output resolution. It is generally agreed that one can not judge the results on screen, since screen resolution is far below that of the printer.

I do not necessarily agree that it is not possible to judge the results on screen. One aspect can even be better judged on screen, and zoomed in, and that is to see if we are introducing artifacts which are a clear sign that we're doing something wrong. When we sharpen a blurred signal to the point that we add new information that was not there in the original unblurred signal in front of the camera, we are introducing artifacts. The other aspect is indeed difficult to judge, and that is how much we must exaggerate, or rather pre-compensate, for losses that have yet to occur (e.g. ink diffusion), and at the viewing distance scale.

A possible workflow that I suggest, until one has enough experience to judge some subject matter by eye, is to include a slanted edge or the above test images as a layer in Photoshop, or process them exactly the same (same resampling percentage) as our target image in Lightroom. That will allow to objectively see the effects of our capture blur + resampling blur and output sharpening, not obscured by present or not present real image detail.

Frankly, I'm amazed that e.g. Lightroom doesn't do something like that under-the-hood but rather seems to use Bicubic Smoother which produces artifacts, and then apparently applies some of the Photokit tricks to hide those artifacts while boosting edge contrast.

It should be almost trivial to use a smart pattern, which can be automatically analyzed after resampling to remove resampling blur. I've proposed one such pattern in an earlier post (see attached image). In fact, that pattern, or a similar one, can be used to determine the full 2D PSF if it rides along the entire processing workflow (behind the scenes, and one could theoretically even include demosaicing and lens correction effects).

A plugin like Topaz Labs Infocus already allows to 'Estimate' the blur PSF, and deconvolve the image, and then offers to possibility add some more regular sharpening to amplify the effect or pre-compensate for output medium losses. It unfortunately doesn't allow to save the estimated settings for other images (I'm hoping version 2 will allow such things), but one can use the Generic setting as a simpler alternative instead (although for output sharpening a larger maximum radius would be helpful, and a more efficient memory management). FocusMagic also works fine with such an approach where one determines the optimal settings for a proxy image that has undergone the same resizing, and applies the same deconvolution on the actual image.

Quote
In developing PhotoKit sharpener Bruce Frasier and his colleagues reportedly made literally hundreds of prints to judge the effects of various parameters. With regard to PhotoKit, one can simulate the workflow (e.g. Glenn Mitchell, TLR), but the PhotoKit people say that such efforts lack the magic numbers (sharpening parameters) required for optimal results.

Well, there is an aspect of smoke and mirrors to that, although experience is also a component. The fact that they needed hundreds of prints may also signal a lack of fundamental insight, which is understandable when one is unfamiliar with Digital Signal Processing (DSP) theory. Of course, it is also a fact that we now have more computer power to do a lot of things nearly in real-time, and that was not available when these more traditional edge contrast and halo masking tools were developed.

Quote
Although the two pass (with optimal creative sharpening) workflow of Frasier et al is widely accepted in the photographic community, you mentioned in a previous post that one could skip capture sharpening and sharpen only the final image. An interesting concept, but how would one implement such a workflow?

One can use the method mentioned above, with a proxy image to serve as a guide. Mathematically it makes little difference if one cascades multiple deconvolutions with a smaller Gaussian PSF, or uses one with a larger Gaussian PSF (see Associative Property, which explains that "any number of cascaded systems can be replaced with a single system"). That's also one of the nice properties of Gaussians. So, besides having to cope with additional resampling artifacts, it's pretty much the same thing, but without cumulative rounding errors and halos building up to larger halos. One can even use resampling algorithms like those in Benvista's PhotoZoom Pro, which add (convincing fake) resolution to edges.

Cheers,
Bart
« Last Edit: September 21, 2013, 03:46:05 PM by BartvanderWolf » Logged
Hening Bettermann
Sr. Member
****
Offline Offline

Posts: 554


WWW
« Reply #18 on: September 21, 2013, 03:34:26 PM »
ReplyReply

Hi Bart,

I want to use a one-step sharpening, after up-sampling, but still just capture  sharpening, leaving the output sharpening to the print service. Upon your recommendation, I will use PhotoZoomPro for the re-sampling. Do I understand your above post so that PhotoZoom Pro will offer the best sharpening built-in, or should I use FocusMagic afterwards? (I think I collected elsewhere that FocusMagic is your favorite?)

Kind regards - Hening
Logged

BartvanderWolf
Sr. Member
****
Offline Offline

Posts: 3415


« Reply #19 on: September 21, 2013, 04:02:40 PM »
ReplyReply

Hi Bart,

I want to use a one-step sharpening, after up-sampling, but still just capture  sharpening, leaving the output sharpening to the print service. Upon your recommendation, I will use PhotoZoomPro for the re-sampling. Do I understand your above post so that PhotoZoom Pro will offer the best sharpening built-in, or should I use FocusMagic afterwards? (I think I collected elsewhere that FocusMagic is your favorite?)

Hi Hening,

Single step Capture+Output sharpening can be done with Photozoom Pro, which does a very good normal resampling + addition of edge sharpening (while adding edge resolution). However, that good normal resampling bit doesn't benefit from capture sharpening. The Photozoom Pro result has added edge sharpening, which makes it a lesser candidate for additional output sharpening.

So my suggestion is to use either a Capture Sharpening + Photozoom Pro upsampling, or a single step Capture + resampling deconvolution (assuming a good resampling algorithm that doesn't introduce artifacts). Very well controlled Capture Sharpening, can be followed by both (either an additional deconvolution after resampling, or Photozoom), but the Capture sharpening must be top notch.

Cheers,
Bart
Logged
Pages: [1] 2 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad