Luminous Landscape Forum

Raw & Post Processing, Printing => Digital Image Processing => Topic started by: John V. on May 19, 2013, 04:52:33 pm

Title: What will work better? Sharpening workflow
Post by: John V. on May 19, 2013, 04:52:33 pm
I've done a lot of experimenting, but just on-screen so far.

Equipment: d800 with 105mm

Software: cs5, photokit sharpener 2

I'm shooting a collection of artwork I own for reproduction. The camera is stationary and the painting moves around on a wall. I shoot in chunks, then stich together (if anyone has any photomerge tips, I could use them. I haven't figured that out 100% either). Anyway, the camera is about 36" away from the wall, which gives me a ~10x7" chunk of the painting. Once I finish shooting/stitching, I end up with a huge file obviously, and the goal is to make an original sized (usually 50x50" or so) print at 300dpi. So the actual question (I follow the pk sharpener recommended workflow... capture sharpen>creative>output):

What would be the correct way to go about this? Should I apply the capture sharpening to the original (over sized) file and then down-sample to the desired dimensions@300dpi? Or, should I down-sample to the desired dimensions@300dpi and THEN apply capture sharpening? Or, a different way altogether?

The reason I don't shoot from further back is because the quality I'm looking for just isn't there.
It makes sense that the closer I shoot, the larger the image, the more pixels I end up dropping when re-sampling. But I've done some testing, and even after dropping all those pixels, the images still look great compared to shooting from say... 6' away or so (to get the 300dpi I need for print). Looking for that happy medium.

Thanks
Title: Re: What will work better? Sharpening workflow
Post by: Schewe on May 19, 2013, 05:11:59 pm
Are you shooting raw? If so, I would be inclined to do the capture sharpening in Camera Raw prior to stitching. Then you can do whatever PKS creative sharpening you my need and then use the output sharpening after the image is at the final size and resolution.
Title: Re: What will work better? Sharpening workflow
Post by: Joe S on May 19, 2013, 11:20:27 pm
I have previously used Photokit Sharpener for all my sharpening.   I have recently started doing my capture sharpening in Camera Raw and have compared the results.   The standard landscape capture sharpening recommended in the Evening  CS6 book (approximately 45, .8, 50, 10) in Camera Raw results in a considerably more sharpened image than the Photokit results.    I would appreciate any input.
Title: Re: What will work better? Sharpening workflow
Post by: Schewe on May 19, 2013, 11:27:10 pm
That's not unexpected...but the question is, is the result better or worse? You can always back off the presets and make your own new ones for your camera and the type of image you make.
Title: Re: What will work better? Sharpening workflow
Post by: bill t. on May 20, 2013, 12:08:10 am
You said you're testing "on screen."  You're wasting your time.  Make prints.   :)
Title: Re: What will work better? Sharpening workflow
Post by: Joe S on May 20, 2013, 01:08:49 pm
That's not unexpected...but the question is, is the result better or worse? You can always back off the presets and make your own new ones for your camera and the type of image you make.

I believe the results are better, but I suppose sharpening is somewhat like contrast in that more generally looks better.    I was surprised at the extent of the difference.    It seems that Photokit is very conservative in its capture application and I wonder why since its other functions seem to start out strong and allow the user to back off.   I started out this exercise to compare Photokit with my Camera Raw settings assuming that the automatic Photokit settings would be "correct" or at least close.    I had been happy with them for years but now think I might have been giving up something.
Title: Re: What will work better? Sharpening workflow
Post by: ErikKaffehr on May 20, 2013, 01:26:43 pm

Hi,

My view, capture sharpen on raw conversion, stitch, upscalen and output sharpen.

You could try using deconvolution type sharpening in ACR/Lightroom, try amont 40, radius 0.7, detail 100 and some masking/noise reduction and see if you like the result.

You could try using an adfitional step of sharpening, say radius 2 and amount 20%, that would increase contrast on low frequency detail.

Best regards
Erik
I've done a lot of experimenting, but just on-screen so far.

Equipment: d800 with 105mm

Software: cs5, photokit sharpener 2

I'm shooting a collection of artwork I own for reproduction. The camera is stationary and the painting moves around on a wall. I shoot in chunks, then stich together (if anyone has any photomerge tips, I could use them. I haven't figured that out 100% either). Anyway, the camera is about 36" away from the wall, which gives me a ~10x7" chunk of the painting. Once I finish shooting/stitching, I end up with a huge file obviously, and the goal is to make an original sized (usually 50x50" or so) print at 300dpi. So the actual question (I follow the pk sharpener recommended workflow... capture sharpen>creative>output):

What would be the correct way to go about this? Should I apply the capture sharpening to the original (over sized) file and then re-sample? Or, should I re-sample to the desired dimensions@300dpi and THEN apply capture sharpening? Or, a different way altogether?

The reason I don't shoot from further back is because the quality I'm looking for just isn't there.
It makes sense that the closer I shoot, the larger the image, the more pixels I end up dropping when re-sampling. But I've done some testing, and even after dropping all those pixels, the images still look great compared to shooting from say... 6' away or so (to get the 300dpi I need for print). Looking for that happy medium.

Thanks
Title: Re: What will work better? Sharpening workflow
Post by: Jack Hogan on May 20, 2013, 02:56:10 pm
I shoot in chunks, then stich together (if anyone has any photomerge tips, I could use them. I haven't figured that out 100% either).

As far as stitching is concerned, you may want to try free MS ICE (http://research.microsoft.com/en-us/um/redmond/groups/ivm/ICE/) to see how it compares to CS6.  It's pretty smart, way (way!) faster and has a few nifty options.  Just drag and drop the TIFFs into it in no particular order and it does the rest.  I use it almost exclusively to stitch my landscapes.

Jack
Title: Re: What will work better? Sharpening workflow
Post by: John V. on May 20, 2013, 09:01:21 pm
Schewe, yeah this is all raw. Joe S, noted. Erik, I'm actually DOWN sampling... which is why I'm confused (among other reasons).

What doesn't seem to be clicking in my brain... is why would I apply capture sharpening to an image that's going to have so many pixels dropped? Instead, drop all unneeded pixels first, and then capture sharpen... Am I looking at this wrong? I wouldn't be surprised. Like Bill said, only a print can really help clear this up for me. I'm just trying to get a workflow down that actually makes some kind of sense before I start throwing canvas into the trashcan.

And yeah, Jack, I spent the weekend playing around with every photo stitching app out there, and I can second Microsoft ICE in case anyone out there is in the same boat. ICE and ptgui are what I'll probably end up using.

Thanks for all the info/suggestions.
Title: Re: What will work better? Sharpening workflow
Post by: Schewe on May 21, 2013, 01:41:20 am
It seems that Photokit is very conservative in its capture application and I wonder why since its other functions seem to start out strong and allow the user to back off.

Because Bruce was a pretty conservative guy and expected people to do creative sharpening and final output sharpening as well. All of theses must be factored in...
Title: Re: What will work better? Sharpening workflow
Post by: Joe S on May 21, 2013, 02:35:57 am
Because Bruce was a pretty conservative guy and expected people to do creative sharpening and final output sharpening as well. All of theses must be factored in...

Sure, it's understood that there would be creative sharpening and output sharpening but I was just comparing capture sharpening to capture sharpening...oranges to oranges and was surprised at the big difference.    As I said, I was using it as a comparison tool to help with my Camera Raw capture sharpening since it can be difficult for me to dial it in.
Title: Re: What will work better? Sharpening workflow
Post by: Bart_van_der_Wolf on May 21, 2013, 03:51:37 am
Erik, I'm actually DOWN sampling... which is why I'm confused (among other reasons).

What doesn't seem to be clicking in my brain... is why would I apply capture sharpening to an image that's going to have so many pixels dropped? Instead, drop all unneeded pixels first, and then capture sharpen...

Hi,

It is not necessary to (capture) sharpen, before down-sampling by a significant percentage (say 50% of the original size or less). However, you mentioned that there is a stitching step first.

Depending on the stitching method, your image quality can benefit from Capture sharpening. This is especially the case when the stitching combines distortion correction, squaring, and rotation. Again, when down-sampling significantly that is less of an issue, although the stitching itself could be slightly more accurate with properly Capture sharpened image tiles.

I'm not exactly sure, but it seems like you are down-sampling the stitched result to achieve a certain output size at 300 PPI. Depending on the printer you use, it can help slightly (depends on the printer driver settings) if you aim for 600 PPI instead. While I understand that the huge file size will quadruple, it will also mean that you have to down-sample less and retain more of the original captured pixels, and you can apply better output sharpening with less risk of small sharpening artifacts showing up.

Cheers,
Bart
Title: Re: What will work better? Sharpening workflow
Post by: John V. on May 21, 2013, 05:51:37 pm
awesome. So,  I really need to settle on at which point the stitching will take place during this process.

I have no problem printing at a high dpi.. (Or should I have a problem with that?). Aside from the printer having to handle the larger file size...) I'm not using a rip. Will the printer driver have issues with the big files? Epson 11880..
Title: Re: What will work better? Sharpening workflow
Post by: Bart_van_der_Wolf on May 21, 2013, 07:47:23 pm
awesome. So,  I really need to settle on at which point the stitching will take place during this process.

I have no problem printing at a high dpi.. (Or should I have a problem with that?). Aside from the printer having to handle the larger file size...) I'm not using a rip. Will the printer driver have issues with the big files? Epson 11880..

Hi,

Some printer drivers will have issues with such large print files if they are sent as one large output file, but if you e.g. work on a Windows OS (or on a Mac with Parallels to run Windows applications), e.g. Qimage Ultimate should offer several benefits.

Not only does it automatically resample your images up (or down) to the printer driver's preferred resolution (e.g. 720 PPI in case of Epsons, and it also applies halo free DFS output sharpening), but it also sends the printer/spool data in chunks, to avoid crashing the printer driver. It also allows to recall older print jobs, with the exact same settings.

Cheers,
Bart
Title: Re: What will work better? Sharpening workflow
Post by: Stephen G on May 22, 2013, 12:57:30 am
Photomerge tip: On the shots around the edge of the painting, you should crop off the bits of wall around the painting. The wall image is static from shot to shot and I've seen Photomerge get confused by this. It does not care if the shots are not all the same pixel dimensions. Just use 'auto'. Sometimes I've seen Photomerge blend the images badly, loosing small sections of painting. When this happens I use PTAssembler with careful control points to stitch (have to set the lens to 2 degree FOV and the optimization to Euclidean plane).

When I stitch large artworks, much in the same way you are, I just capture sharpen slightly conservatively in LR. Then I stitch and crop as needed. Print sizing and sharpening is handled by Qimage.
Title: Re: What will work better? Sharpening workflow
Post by: Gulag on May 22, 2013, 08:30:00 pm
I normally disable sharpening when output to tiff from C1Pro, and use Photoshop for sharpening.  From my own experience, I have found Photoshop's High Pass Filter is quite good,  if you can fix Adobe's handicapped implementation (aka clipping) first before you apply it. Besides, you can also add a mask to control exactly the area and amount of sharpening needs to apply and
Title: Re: What will work better? Sharpening workflow
Post by: Peter Le on May 22, 2013, 11:16:33 pm
Hi,

Some printer drivers will have issues with such large print files if they are sent as one large output file, but if you e.g. work on a Windows OS (or on a Mac with Parallels to run Windows applications), e.g. Qimage Ultimate should offer several benefits.

Not only does it automatically resample your images up (or down) to the printer driver's preferred resolution (e.g. 720 PPI in case of Epsons, and it also applies halo free DFS output sharpening), but it also sends the printer/spool data in chunks, to avoid crashing the printer driver. It also allows to recall older print jobs, with the exact same settings.

Cheers,
Bart
 Bart quick question.....I am thinking of trying Qimage. You say it applies output sharpening.....does this mean if you have a Photoshop work flow with output sharpening part of it....do you drop your output sharpening in Photoshop and let Qimage do it....and you say it has better results ?
   Thanks Peter
Title: Re: What will work better? Sharpening workflow
Post by: Schewe on May 22, 2013, 11:37:51 pm
 Bart quick question.....I am thinking of trying Qimage. You say it applies output sharpening.....does this mean if you have a Photoshop work flow with output sharpening part of it....do you drop your output sharpening in Photoshop and let Qimage do it....and you say it has better results ?

I'll let Bart answer in more depth but...you only want to do one round of output sharpening, ever! I use LR's output sharpening even for images sent through Photoshop, so when printing from LR I do NOT use output sharpening in Photoshop. The deal would be the same for Qimage...it's an either/or situation. You'll need to test for yourself what's the optimal approach, but in any event, never output sharpen twice–that would result in over-sharpening!
Title: Re: What will work better? Sharpening workflow
Post by: Peter Le on May 23, 2013, 12:41:57 am
I'll let Bart answer in more depth but...you only want to do one round of output sharpening, ever! I use LR's output sharpening even for images sent through Photoshop, so when printing from LR I do NOT use output sharpening in Photoshop. The deal would be the same for Qimage...it's an either/or situation. You'll need to test for yourself what's the optimal approach, but in any event, never output sharpen twice–that would result in over-sharpening!
  Thanks Jeff......that's what I thought......but have always printed from Photoshop with my own action for output sharpening. Have never used Qimage and was not aware they had very good output sharpening.
  thanks
  Peter
Title: Re: What will work better? Sharpening workflow
Post by: ErikKaffehr on May 23, 2013, 12:44:25 am
Hi,

You should scale the image to printer native resolution, which would be 360 PPI on the Epson (or 720 PPI in high res mode), else the printing pipeline would do the scaling using an inferior method.

Best regards
Erik


Hi,

It is not necessary to (capture) sharpen, before down-sampling by a significant percentage (say 50% of the original size or less). However, you mentioned that there is a stitching step first.

Depending on the stitching method, your image quality can benefit from Capture sharpening. This is especially the case when the stitching combines distortion correction, squaring, and rotation. Again, when down-sampling significantly that is less of an issue, although the stitching itself could be slightly more accurate with properly Capture sharpened image tiles.

I'm not exactly sure, but it seems like you are down-sampling the stitched result to achieve a certain output size at 300 PPI. Depending on the printer you use, it can help slightly (depends on the printer driver settings) if you aim for 600 PPI instead. While I understand that the huge file size will quadruple, it will also mean that you have to down-sample less and retain more of the original captured pixels, and you can apply better output sharpening with less risk of small sharpening artifacts showing up.

Cheers,
Bart
Title: Re: What will work better? Sharpening workflow
Post by: Bart_van_der_Wolf on May 23, 2013, 06:16:33 am
 Bart quick question.....I am thinking of trying Qimage. You say it applies output sharpening.....does this mean if you have a Photoshop work flow with output sharpening part of it....do you drop your output sharpening in Photoshop and let Qimage do it....and you say it has better results ?

Hi Peter,

As Jeff said, you'll want to avoid double output sharpening. Therefore you must either resample to output size in Photoshop and apply output sharpening there, or let Qimage handle the up-sampling of your Capture sharpened image plus output sharpening. IMO the Qimage path is more efficient because you do not need to produce different files for different output sizes, and Qimage uses better resampling than the current Photoshop algorithms. Qimage does offer the possibility to adjust or even switch off the output sharpening, in case the file was already output sharpened.

Which route gives better results, depends on the up-sampling and sharpening steps in Photoshop. The Qimage quality is very good, and the output sharpening is halo free, so it can be pushed quite far. I usually use the default setting of 5, but it depends on the specific media one uses.

Cheers,
Bart
Title: Re: What will work better? Sharpening workflow
Post by: John V. on July 05, 2013, 10:19:41 am
Ok, so what would be advised given this scenario? Suppose I want to use PK sharpener all the way around (capture/creative/output) Do I want to tell Camera Raw to not apply any sharpening? I notice it defaults to 25,1.0,25. I'm guessing yes.
Title: Re: What will work better? Sharpening workflow
Post by: Bart_van_der_Wolf on July 05, 2013, 11:37:57 am
Ok, so what would be advised given this scenario? Suppose I want to use PK sharpener all the way around (capture/creative/output) Do I want to tell Camera Raw to not apply any sharpening? I notice it defaults to 25,1.0,25. I'm guessing yes.

The Camera Raw sharpening defaults are just that, defaults (no intelligence based on file content or aperture used,  none whatsoever). In fact, analysis has shown (http://www.luminous-landscape.com/forum/index.php?topic=68089.msg538932#msg538932), they are sub-optimal in most cases so you may want to skip any in ACR sharpening at that stage, and use PKS all the way if you like.

Cheers,
Bart

P.S. One reason to use ACR sharpening is when you also require ACR noise reduction, because they are somewhat linked.  Of course, when you have a better noise reduction application, e.g. Topaz Denoise, you can do it separately after Raw conversion and before sharpening.
Title: Re: What will work better? Sharpening workflow
Post by: AFairley on July 05, 2013, 01:28:00 pm
Bart, this thought occurred to me the other day:  Is there any advantage to doing capture sharpening in the raw conversion as opposed to doing it on an already "baked" demosaiced file?  I believe Eric Chan has said that it is preferable to do lens corrections in the demosaic process as opposed to after the fact (though I may have misunderstood what he was saying).
Title: Re: What will work better? Sharpening workflow
Post by: Bart_van_der_Wolf on July 05, 2013, 01:59:39 pm
Bart, this thought occurred to me the other day:  Is there any advantage to doing capture sharpening in the raw conversion as opposed to doing it on an already "baked" demosaiced file?  I believe Eric Chan has said that it is preferable to do lens corrections in the demosaic process as opposed to after the fact (though I may have misunderstood what he was saying).

Hi Alan,

That depends on at which stage the corrections take place. It is e.g. possible to 'do' Chromatic Aberration correction before actual demosaicing, which would not only improve the color accuracy of the demosaicing itself, but can also help (luminosity) resolution coming out of the demosaicing stage, even before Capture sharpening.

When a relatively mediocre resampling algorithm is used for small resampling adjustments, e.g. lens distortions (especially near the optical center) or rotation/keystone correction, then proper (!) Capture sharpening before the resampling may help to preserve micro detail a bit better.

Here (http://pixinsight.com/doc/docs/InterpolationAlgorithms/InterpolationAlgorithms.html#__section004__) (Rotation examples) is a nice demonstration of the devastation at the micro-detail level that can result from mediocre resampling methods. It's visually even better to make a relatively soft conversion and use deconvolution sharpening to recover some of the detail because it looks more homogenous instead of fading in and out of sharp detail (sharpening that can make it look even worse). Of course it also depends on the actual image detail, because e.g. out-of-focus (OOF/DOF) zones have no micro-detail other than noise.

The key concept is proper Capture sharpening. Without that, all bets are off. That's why I'm stressing the need for proper Capture sharpening so much, it's at the source of what follows. And garbage-in-garbage-out (GIGO) still rules ...

Without such automatic resampling, it's always better to use a superior deconvolution sharpening method after e.g. ACR than as part of the postprocessing in ACR. It complicates the workflow by adding an additional step, but such is the fate of those striving for perfection.

Cheers,
Bart