Another Capture Sharpening Question

[MarkIV]

Got a pretty straightforward question about Capture Sharpening.  I am attempting to produce a maximum size Fuji Flex, Lightjet, panorama landscape print.  Its native Res is 28 MP.  I already have a print of it at 80 inches that turned out really good (but I processed it before knowing about Capture Sharpening - and my ability to really take my images to their potential in the raw pipeline).  

Now I am going back to the drawing board and reprocessing it with much better ACR 5.3 skills.  I think that Capture Sharpening is going to bring more quality detail to the table this time.  Since I am really pushing the size on this one, is it better to push the capture sharpening amount a tad, or to just process it the same as is always recommended (just make it look good and not at all crunchy at 100%)?

So to be clear, the question could be asked this way: when making mural size, maximum sized gallery prints, should any more Capture Sharpening be applied than when making a smaller or moderate enlargement?

*Please assume that I also interpolate using the best methods and my Creative Sharpening and Sharpening for output are going to be maximized too.

Thanks for the help.

[digitaldog]

Capture sharpening is output agnostic. That's why there are two separate sharpening steps. If an output or size needs adjustment for sharpening, its done at the output sharpening stage, not capture.

Capture sharpening should be done at the desired capture resolution meaning, if you decide at the get-go to size up past native resolution, you do that then capture sharpen. IOW, the sharpening is specific to that pixel density.

[Schewe]

is it better to push the capture sharpening amount a tad, or to just process it the same as is always recommended (just make it look good and not at all crunchy at 100%)?

No...if you oversharpen at the capture stage there is nothing you can do about that down stream...photographers tend to have this habit of "if a little is good, more is better" and that is simply not the case with capture sharpening in Camera Raw. If, after you get your image at the final size with the final sharpening in place and you determine you need a tad more, THAT'S easy to do...but if you get to that stage and find out it's too much, you are screwed, right?

[MarkIV]

No...if you oversharpen at the capture stage there is nothing you can do about that down stream...photographers tend to have this habit if "if a little is good, more is better" and that is simply not the case with capture sharpening in Camera Raw. If, after you get your image at the final size with the final sharpening in place and you determine you need a tad more, THAT'S easy to do...but if you get to that stage and find out it's too much, you are screwed, right?

That is what I thought.  I must have come across some false info somewhere.  Thanks both of you.

[MarkIV]

if you decide at the get-go to size up past native resolution, you do that then capture sharpen. IOW, the sharpening is specific to that pixel density.

So, I'm a liitle confused here.  Are you saying that if I am going to push a print to the largest size I can get it, interpolate up in ACR to the largest size it will allow me, and Capture Sharpen there as well (Max interpolation plus Capture Sharpen in ACR) at the pixel density I will print at (300 ppi)?  Instead of capture sharpening in ACR but bringing the file into PS at its native size then using bicubic Smoother (or whatever)  to take the size up? I have thus far done the latter thinking it is my best option.

[PeterAit]

So, I'm a liitle confused here.  Are you saying that if I am going to push a print to the largest size I can get it, interpolate up in ACR to the largest size it will allow me, and Capture Sharpen there as well (Max interpolation plus Capture Sharpen in ACR) at the pixel density I will print at (300 ppi)?  Instead of capture sharpening in ACR but bringing the file into PS at its native size then using bicubic Smoother (or whatever)  to take the size up? I have thus far done the latter thinking it is my best option.

The way I look at it, capture sharpening is the sharpening you want on the image no matter what you will do with it in the future - big print, small print, web, etc etc. The amount depends on the resolution and the type of detail in the image. Then, for a specific use you apply output sharpening on top of the capture sharpening.

Peter

[MarkIV]

The way I look at it, capture sharpening is the sharpening you want on the image no matter what you will do with it in the future - big print, small print, web, etc etc. The amount depends on the resolution and the type of detail in the image. Then, for a specific use you apply output sharpening on top of the capture sharpening.

Peter

Thanks Peter very much for your feedback. That is basically what Shewe said too and what I thought.  Maybe case closed?

[Panopeeper]

I am attempting to produce a maximum size Fuji Flex, Lightjet, panorama landscape print

Mark,

are you talking about a stitched panorama (as in contrast to a panorama-style crop)?

[MarkIV]

Mark,

are you talking about a stitched panorama (as in contrast to a panorama-style crop)?

This image is 6 shots processed near completion in ACR 5.3, then Photomerged in PS CS4.  I just did it today, and zooming up on the image, with the new and improved ACR 5.3 skills it kills the image that had no Capture Sharpening originally and was adjusted a lot more in PS vs the new one done almost exclusively in ACR.

Sometimes I do this on non-panoramas.  Just recompose and shoot in 2,3,4 or whatever to pump up the M pixels for mural size prints. Right now I'm working on a 2:3 image of Peto Lake, Canada that I shot in one, but then also recomposed and shot it in 4.  With the 1DS Mark II, stitched, it is about 35-40 MP.

[Ben Rubinstein]

For the life of me I've never understood the concept of doing 'capture sharpening' right at the beginning for the reasons outlined above. I do capture/local/output (seperately and in that order of course) when the image is finished and I already have my 20 odd layers or whatever. Never understood why you would want to do something that is both destructive and non reversable at the beginning of working on an image. I find the 'contrast' argument rather weak. Of course if all your layers are adjustment layers then do the capture on the original layer as a smart object if you wish. Opening a RAW file with the capture sharpening as a smart object for use as your first layer is a good example of this. I very rarely end up with adjustment only layers though.

[michael]

Ben,

Most photographers have multiple uses for their images. They may end up as a fine art print, a web gallery image, prepared for commercial press printing in a book, and more.

Since all of these will require the same input sharpening and different output sharpening, why not do the input sharpening once, up front? Do it on a layer if you wish, but why do it over and over each time?

Michael

[Ben Rubinstein]

Ben,

Most photographers have multiple uses for their images. They may end up as a fine art print, a web gallery image, prepared for commercial press printing in a book, and more.

Since all of these will require the same input sharpening and different output sharpening, why not do the input sharpening once, up front? Do it on a layer if you wish, but why do it over and over each time?

Michael

I'll ask you the same thing Michael, why not just do it as the last layer then? Then apply the output dependant on what you are doing with it? That way you are not doing such an important step at a stage that could cause serious grief should the capture need changing. For example should you want to use a more advanced capture method some time in the future. I've started using ACR for capture finding the level of control very much to my taste. Should I have decided to change with the capture as the first layer of many complicated files I would have either had to give up the idea of redoing with (IMO) a better method or be tearing my hair out at the amount of work involved.

[Schewe]

I'll ask you the same thing Michael, why not just do it as the last layer then?

Well, if you waited till the end and spent a lot of time working on your image only to find out that it's "soft", you would feel pretty foolish, huh? The reason to do it in the beginning is so you are working with a good, sharp image from the beginning. A lot of subsequent imaging will be dictated upon what the image looks like from the beginning so you want a good "use-neutral" starting point and that's a "sharp" image (which is what the capture sharpening is all about).

[MarkIV]

For the life of me I've never understood the concept of doing 'capture sharpening' right at the beginning for the reasons outlined above. Never understood why you would want to do something that is both destructive and non reversible at the beginning of working on an image. I find the 'contrast' argument rather weak.

I'm curious, have you done a lot of test printing to see what the end results are in the real world?

I have. I am currently doing a ton of it right now in prep for the gallery scene.  

My large (40-80+ inches) Fuji Flex (Poly) Lightjet (or Chromira or Lambda) gallery prints end up better and with more discernible quality by moderate Capture Sharpening at the raw stage, interpolating up to max size (with bicubic Smoother) and then pushing the sharpening (custom) there to its max.  Versus not Capture Sharpening, then interpolating up to the same size and then pushing the sharpening (custom) to its max.  Very noticeable differences to me.  I am also able to get the detail I need while pushing the size limitations a little more.

As far as "destructive" IMO that depends on how far you push the Capture Sharpening in ACR.  The general guidelines have been to tighten the images look up a little but don't do discernible damage to the image at 100% view.  IMO, non-discernible "damage" at 100% is even much less discernible as a finished print...

[Panopeeper]

This image is 6 shots processed near completion in ACR 5.3, then Photomerged in PS CS4.  I just did it today, and zooming up on the image, with the new and improved ACR 5.3 skills it kills the image that had no Capture Sharpening originally and was adjusted a lot more in PS vs the new one done almost exclusively in ACR

IMO one *must* perform capture sharpening before stitching. Capture sharpening should occur on the original pixels, in their original placements. Stitching includes warping (like "free transform"), which means that pixels vanish, other will be interpolated, neighbours "go apart", etc. No talk can be about "capture sharpening" after the stitching.

[Ben Rubinstein]

I'm curious, have you done a lot of test printing to see what the end results are in the real world?

I have. I am currently doing a ton of it right now in prep for the gallery scene.  

My large (40-80+ inches) Fuji Flex (Poly) Lightjet (or Chromira or Lambda) gallery prints end up better and with more discernible quality by moderate Capture Sharpening at the raw stage, interpolating up to max size (with bicubic Smoother) and then pushing the sharpening (custom) there to its max.  Versus not Capture Sharpening, then interpolating up to the same size and then pushing the sharpening (custom) to its max.  Very noticeable differences to me.  I am also able to get the detail I need while pushing the size limitations a little more.

As far as "destructive" IMO that depends on how far you push the Capture Sharpening in ACR.  The general guidelines have been to tighten the images look up a little but don't do discernible damage to the image at 100% view.  IMO, non-discernible "damage" at 100% is even much less discernible as a finished print...

Who said I didn't do capture sharpening? Of course I do it, just at the end of editing not the beginning as I said so that if I change my mind I can do it again rather than having to start from scratch as you people will have to. Finish image, apply capture sharpening, after that local sharpening if needed and save. Output sharpening after resize and relative to output media. I don't and never have denied Capture Sharpening, just still waiting for a single good reason for doing it as the first stage of a process which by definition means that if you screwed up it's irreversible.

Sorry Schewe, that's laughable. Either use ACR with sharpening 'just for preview' or half a second to open USM and check it's sharp. The concept of needing to APPLY permanent and unchangeable capture sharpening at the beginning of a process just to check sharpness is insulting the intelligence of users. Given that I'm certainly not applying layers while viewing at 100% or even usually a regular fraction of that (no open GLU and it certianly never existed when the concept was formed) when working with 35+ megapixel images I can't even begin to see the sharpening while working on the image. So again, why at the beginning?

Panopeeper? You want to do capture before all the warping and messing of the pixels? First time I've heard it suggested that you would want to sharpen prior to distortion rather than capture sharpen when you have the final finished product. If you sharpen prior to stitching then all your sharpening will be messed up completely by the distortion and the finished stitch will have some sharp, some smearing, some weird stuff and a million artifacts.

[MarkIV]

Who said I didn't do capture sharpening? Of course I do it, just at the end of editing not the beginning as I said so that if I change my mind I can do it again rather than having to start from scratch as you people will have to. Finish image, apply capture sharpening, after that local sharpening if needed and save. Output sharpening after resize and relative to output media. I don't and never have denied Capture Sharpening, just still waiting for a single good reason for doing it as the first stage of a process which by definition means that if you screwed up it's irreversible.

Sorry Schewe, that's laughable. Either use ACR with sharpening 'just for preview' or half a second to open USM and check it's sharp. The concept of needing to APPLY permanent and unchangeable capture sharpening at the beginning of a process just to check sharpness is insulting the intelligence of users. Given that I'm certainly not applying layers while viewing at 100% or even usually a regular fraction of that (no open GLU and it certianly never existed when the concept was formed) when working with 35+ megapixel images I can't even begin to see the sharpening while working on the image. So again, why at the beginning?

Panopeeper? You want to do capture before all the warping and messing of the pixels? First time I've heard it suggested that you would want to sharpen prior to distortion rather than capture sharpen when you have the final finished product. If you sharpen prior to stitching then all your sharpening will be messed up completely by the distortion and the finished stitch will have some sharp, some smearing, some weird stuff and a million artifacts.

Hmm, interesting.  Maybe I'm wrong here, but is your argument that you see no qualitative benefit of capture sharpening at the raw stage in the raw pipeline?  

I have not ever even thought of pre sharpening an image that is not in the raw stage (except when reading Bruce Frasers Sharpening book).  If I did not believe in the superior qualitative difference of capture sharpening in the raw stage, and I had a finished, non sharpened tiff that was at native size and I needed to enlarge it, I would just do all the sharpening after the full interpolation.  It is the fact that it can be done in the raw pipeline on raw data and with some very savvy tools designed for it there, that I believe I end up seeing a substantial benefit on the finished enlarged print.

[MarkIV]

IMO one *must* perform capture sharpening before stitching. Capture sharpening should occur on the original pixels, in their original placements. Stitching includes warping (like "free transform"), which means that pixels vanish, other will be interpolated, neighbours "go apart", etc. No talk can be about "capture sharpening" after the stitching.

Panopeeper? You want to do capture before all the warping and messing of the pixels? First time I've heard it suggested that you would want to sharpen prior to distortion rather than capture sharpen when you have the final finished product. If you sharpen prior to stitching then all your sharpening will be messed up completely by the distortion and the finished stitch will have some sharp, some smearing, some weird stuff and a million artifacts.

I do quite a bit of stitching to make mural size gallery laser prints.  Here are my thoughts on this (and I'm here to mull this stuff over and learn so I don't necessarily think I'm always right):

When software stitches, it does warp and localized interpolation (more so on images not shot well).  This averaging or softening occurs irregardless of capture sharpening at the raw stage or not.  So, some localized areas can be slightly softer regardless of what pre sharpening method is used.  If these areas are detectable and are not desirable then a touch of additional custom sharpening (in those specific areas) before printing should suffice to give the perfect illusion of consistent sharp detail.  

I (both for stitches, or non stitched images) do the capture sharpening at the raw stage in ACR 5.3 because of what I see as the overall superior benefits of raw capture sharpening.  This is the very reason I went back to the drawing board on the aforementioned pano image.  I see (maybe I'm crazy) a big benefit in discernible detail in mural sized prints when the images are capture sharpened in ACR at the raw stage.  

The comment:

If you sharpen prior to stitching then all your sharpening will be messed up completely by the distortion and the finished stitch will have some sharp, some smearing, some weird stuff and a million artifacts.

[/b]
 may be true, but I have yet to see this in my stitched maximally enlarged prints.  

As a counter theoretical argument, maybe it is possible that the stitching software (like PS CS4 Photomerge) will do a better and more accurate stitching job when the pixels it is trying to locate, compare, overlap and work with are more defined?

I don't know.  

Either way,  I have yet to see any problems related to raw ACR capture sharpening on stiches or otherwise, in a finished print.

[Panopeeper]

Panopeeper? You want to do capture before all the warping and messing of the pixels? First time I've heard it suggested that you would want to sharpen prior to distortion rather than capture sharpen when you have the final finished product

Sharpening of the final finished product is output sharpening. I keep my panos multileyered, unsharpened until I create a specific size for online or print.

Capture sharpening is about restoration of the image to the state it would be without the AA filter. It's effect is very different from the output sharpening. I suggest you to try the settings in ACR in 300% view. It does not create halos like the sharpening is PS; instead, it restores the contours where they belong. My experience is, that this is done best on the original image, together with the raw processing.

If you sharpen prior to stitching then all your sharpening will be messed up completely by the distortion and the finished stitch will have some sharp, some smearing, some weird stuff and a million artifacts

This is not so. That sharpening will be carried on.

[Panopeeper]

As a counter theoretical argument, maybe it is possible that the stitching software (like PS CS4 Photomerge) will do a better and more accurate stitching job when the pixels it is trying to locate, compare, overlap and work with are more defined?

That not; however, if you were using a decent stitcher (as opposed to Photomerge etc.), then an extra sharpening step before stitching could be helpful.

In the first step I am sharpening the pano frames in ACR stronger than I want to have it in the pano. I am using those frames when defining the control points (the matchings between the frames), for I am doing that manually. The stronger sharpening makes this easier. When I am happy with the control points and with the adjustment, sizing of the pano, I regenerate the frames now with less sharpening (like amount 80, radius 0.8 to 1, detail 25), now 16bit ProPhoto (the 16bit is too slow and useless for the pano "development"), and then I start the final stitching, using the parameters created and recorded previously.