I have been a very big advocate of capture sharpening for fine art workflow for more than a few years now (landscape fine art gallery enlargements). I was under the impression that in ACR (7.1) by using the Radius to the left (0.5) and Detail to the right (100) I was maximizing the deconvolution aspect of sharpening and maybe even drawing out a hair more real detail from the raw file.
Today I did a test on a high frequency raw image (trees) and had my settings at Amount 45, Radius 0.5, Detail 0, and masking at 20. But then I also did it on the same image, this way: I brought a second version of the raw file into PS with no capture sharpening, and saved it as a tiff, then I brought that tiff back into ACR 7.1 and then did the capture sharpening. In Photoshop CS6 I put them on top of each other as layers (one raw sharpened and one tiff sharpened) and at all viewing distances they were exactly the same to my eye (I did notice that the histograms of each were very, very slightly different).
But I could not see a single pixel or edge change anywhere in the image even at 1600% viewing distance. This seems to blow away my impression about deconvolution sharpening actually drawing out more real detail from a raw file. I totally understand that sharpening is not real detail, but on this forum years ago I came to believe that somehow a bit more real detail might be accessed in raw by the aforementioned settings. Any thoughts?
Second question: do you really think there is a quantitative quality difference in detail (or the illusion of) in an image that has capture sharpening applied in ACR (7.1) at its native size, then is enlarged substantially with further rounds of sharpening and grain simulation in the end (versus just sizing it up soft and doing all the sharpening at the end)? I did testing on this years ago and it seemed to be noticably better overall. I guess I'm just second guessing it again.
My basic workflow is capture sharpen Radius 05. / Detail 100, then upsize with Smoother (40, 50, 60, 70 inches...) then do some moderate high pass, then advanced use of unsharp mask (LAB - L channel - or Luminosity - RGB - also blending/blend if sliders for fall off if necessary) then ACR grain simulation (on a seperate layer - not to create grainy photos - but create the illusion of more detail and to camouflage artifacting). I believe after years of testing and practice this seems to be about as good as it gets for my content.
Lastly, smart sharpen... I have not used this much, but do you think this workflow might benefit from using it instead of USM (with the more deconvolution - lens blur/more accurate) type of sharpen near the end)?
Side note: for those involved with focus bracketing (for increased depth of field with the sharpest f/stop) it is common knowledge that capture sharpening at the raw stage (before the Auto Align and Auto Blend in PS) confuses the auto blend algorithm as to what are the real sharper pixels. So, we don't capture sharpen those images (this is common practice).
And after blending the images (Auto Blend) we usually just size up (for enlargement) and go. Now I am thinking that after the focus blending is finished and the file is flattened it might be a good idea to bring that tif file back into ACR and apply a little bit of capture sharpening before the upsize. Does that make sense to you?
Cheers for your time and feedback!
Message linebreaks added by: PECourtejoie
Next time, you might want to break down your post to simpler less massive qustions...I'll respond to one.
To the point, there is not much difference between capture sharpening a low ISO raw image or a rendered image brought back into Camera Raw. Not sure why you would think there would be a strong difference...the raw is linear and the rendered file brought back into ACR goes back into a linear gamma for processing and since the sharpening is done on luminance date, it's about 6 of one, 1/2 dozen of the other.
Where capture sharpening at the raw stage really comes into play is where you are mixing sharpening and significant noise reduction...that's where it's optimal to do both at once in a single process because sharpening and noise reduction are two sides of the same coin.
Thank you Jeff. I know, too many questions (sorry). I thought maybe a pick and choose situation...
With low ISO shots (100) I always try to use the least noise reduction possible in ACR (less is more approach) which means on correctly exposed images ("to the right") on Canon cameras, my ACR settings are often 0 on luminance and 0-5 on color noise. My theory is that I am trying to keep a much of the fine detail as possible. Also the reason I like the 0.5 radius and 100 detail.
Disclaimer: I did not entirely understand the original post, and maybe some of this does not apply in your multi-step process, so take with salt...
With low ISO shots (100) I always try to use the least noise reduction possible in ACR (less is more approach) which means on correctly exposed images ("to the right") on Canon cameras, my ACR settings are often 0 on luminance and 0-5 on color noise.
At ISO 100, I rarely use any luminance noise reduction, however my experience is that even under ideal circumstances including low ISO, a modicum of color noise reduction is still essential for optimum image quality. - it is almost never desirable to set color noise reduction to zero, IMO - YMMV. (Note: in NX2 - you don't even get a choice about it - (true) color noise reduction will be applied, as Nikon sees fit).
Note: ACR's color noise reduction algorithm is image adaptive (it's more "aggressive" on higher ISO shots). You shouldn't see much (if any real) detail loss with color noise reduction at 25 on ISO 100 shots, eh?
My theory is that I am trying to keep a much of the fine detail as possible. Also the reason I like the 0.5 radius and 100 detail.
In my opinion, ideally, the sharpen settings should depend on the photo - type, and inherent focus... (.5/100 may be great for ultra-clean, ultra-sharp landscapes, but may not be appropriate for portraits...)
But different strokes for different folks. (and I don't know what kinds of photos you mainly (or only) shoot).
Thanks Rob. Good feedback. I agree. However I do occasionally have an image (today I was looking at an old file, one shot with the 1DS Mark2) of some aspen trees and even at 0 for color noise there is NO noise in it whatsoever. So, less is more, is what I am hearing but also don't be afraid to use some color noise reduction even on nearly every image. Interesting point about it being adaptive to different ISO's. I did not know that.
I do landscape, and post production for other nature/landscape gallery photographers, so that is my basic context. Higher frequency images. Sharp... But, yes you are right, a lot of what we discuss depends on many factors especially the type of image/photography we are doing. I would NEVER apply these settings to a portrait...
One way to see exactly what color NR is doing, is to zoom in well beyond 1:1 - at least 4:1, even to 11:1 if your eyes are bad .
Then do before/after comparisons, or the equivalent.
Also, consider the color NR detail slider - sometimes you can preserve detail better (whilst still keeping color noise in check) by cranking this slider up than cranking the amount slider down, although truth be told, it's effect is very subtle at lowest ISO. But beware, the color NR detail slider works very differently than the luminance NR detail slider - I regularly crank lum. detail up to 100, but you may actually exacerbate noise by cranking color detail all the way up - maybe better to balance lowered color NR amount with raised color detail, when you are trying to wring out max detail.
Like Jeff, I got overwhelmed by your original post and quit reading.
1.— One question or topic per thread; and
2.— Don't type more than three or four lines without inserting a blank line. Solid blocks of text are tough to read on screen, especially for old geezers like yours truly.
(I may come back once I copy and paste your text into a text file and reformat it to read it properly. )
Europe, Middle East and Africa