Which is "better" - sharpening in ACR, in PSE, or both?

[xpatUSA]

Hello,

This question is specifically about ACR 5.4 vs PSE 6, with which I will be staying for the foreseeable future. In ACR there are four sharpening sliders but in PSE (unsharp option) only three. They both have "amount" and "radius". However, ACR has "detail" and "mask" sliders but PSE just has "threshold". I do have an understanding of what all these adjustments are for but I'm wondering how members view the ACR method versus the PS(E) method. Do y'all only use the ACR one if editing a RAW file, or do you add more in post? Or always have the ACR sharpness sliders at minimum and sharpen in post only? Which is the better method, what do you think?

[Dave Humphries]

Hi Ted,

Personally, (these days) I set all the sharpening sliders in ACR to minimum and do it later in my workflow, when I am in PSE or CS5.

The minimum radius in ACR is 0.5px, which is sometimes too wide - the UnSharp Mask in PSE goes smaller (my standard is 0.3px).

Also, if you shoot high iso, you don't want to sharpen the noise and third party noise reducers are better than what is in ACR, but these usually work as plug-ins to PSE, hence not wanting to sharpen until after I have noise reduced in PSE (or CS5).

What's right for you will depend upon what you are shooting, the iso use, how much you are cropping, what output media you intend (e.g. web or print).

There are many, many posts on this topic here at CiC see tag cloud, plus some tutorials.

Here is the standard reference thread we give people; When/How to Best Sharpen a Digital Photograph - although my own knowledge on the topic has improved somewhat since I posted there

Cheers,

[Geoff F]

I usually do a little Capture Sharpening in ACR then finish with normal Unsharp Mask, or alternative, after doing remaining editing (with CS5 in my case).

ACR around 50% and 0.5 radius. Noise reduction at 10 or 20. Everything else at basic levels because this is just the first Capture sharpen.

Occasionally, if noise in the shadows is likely to be a problem, I do a couple of Raw conversions with the shadow conversion having more noise reduction. Then hand merge the conversions with layers and masks.

[xpatUSA]

Thanks Dave,

Yes, a very good point about the minimum radius in ACR - I am also a 0.3px guy ;-). I shoot pics for screen viewing only but have lately found slightly better contrast in Sigma HI res images downsized as opposed to straight LO res (binned) images. I only use ISO 100, so noise is not an issue. Thanks for the links, especially the "reference" thread, I shoulda looked first . .

Since the Sigma cameras don't have 'blur' filters, I am leaning toward keeping my current practice of not sharpening in ACR but always happy to read any other views on the subject . .

[Colin Southern]

Sharpening in ACR doesn't really "cut the mustard" IMO - but - it can be turned off completely (if one does a Control + K whilst in ACR they should be able to set sharpening to "preview only") (works for Photoshop-hosted ACR - not sure about PSE-Hosted though).

The problems are:

1. (for Canon shooters anyway) - Canon recommend capture sharpening of 0.3px @ 250 or 300% (depending on which white paper you read) @ 1 threashold (base ISO), rising as the ISO rises). Regardless, as Dave points out, ACR does a minimum radius on 0.5px (which is kinda surprising as sharpening was put into ACR specifically for capture sharpening -- so I'm guessing that there's a setting that's functionally equivalent).

2. Images normally need 3 sharpening passes (minimum) - and so one application from within ACR can't apply 3 different types

3. If one has dust bunnies it's usually slightly easier to zap them in certain areas (like faces) before capture sharpening.

Personally, I just have a capture sharpening and content/creative sharpening buttons setup in my actions panel (for content/creative sharpening I often use 4px @ 40% for portraiture, and the same - but twice - for landscape) (lazy mode )

[bedantabd]

Thanks Ted for starting this discussion.

I am quite new to pp and currently I am using ACR / CS-5. As soon as I open a RAW image in ACR, the image looks blurry with the yellow caution mark at the right hand top corner. After a few seconds, the image becomes much sharper with the yellow mark gone (not quite sure what's happening). Can anybody please explain?

Then I apply 80-100% amount and 20-40% noise reduction. That's it, I'm done with sharpening/noise reduction (and I am quite satisfied with the result)!!! Guys, please don't laugh at me, I know this is not the best way for image sharpening / noise reduction. But at least this is better than not doing any sharpening at all.

I have found that for (1) well exposed images of (2) subjects (birds in most of the cases) not far away and (3) at low ISOs (100-400), the above workflow works well for (4) monitor viewing at (5) 50% or less. This will probably not work if any of 5 condition(s) is/are not met.

Lately, I have discovered (??) that using “mask” results in a better final image than using “noise reduction”.

Any suggestion on my current workflow to make it better?

[Dave Humphries]

Any suggestion on my current workflow to make it better?

Have you seen this tutorial?
DIGITAL PHOTO EDITING WORKFLOW

Also, here's mine (it hasn't really changed since 2010 if you substitute "CS5" for "Elements")

Cheers,

[Andrew76]

There's some great advice, and I never really realized how close my actual (self-taught) protocol is to what's already been mentioned.

I do one major thing differently though. Cropping is MUCH higher on my list. Often times, I see the image in my head, the way I want it to be, and so when it shows up on my computer, if there isn't a 'shot' already there, or something that I can crop down into what I had envisioned, then I pass over it, and move onto the next.

To me, there's no sense going through all the work, and making the image pretty, and sharp, if there's nothing there to look at.

[xpatUSA]

Thanks Ted for starting this discussion.

I am quite new to pp and currently I am using ACR / CS-5. As soon as I open a RAW image in ACR, the image looks blurry with the yellow caution mark at the right hand top corner. After a few seconds, the image becomes much sharper with the yellow mark gone (not quite sure what's happening). Can anybody please explain?

Namaste, Bedanta-ji

I can only speak for some converters which, for Sigma .X3F (RAW) files, extract the JPEG thumbnail which is embedded in the file as well as the RAW data. The JPEG thumbnail (1/4 size for SD10 images) is expanded to fit the window and and can look quite blurry until it is replaced by the loaded and decoded RAW data. The effect is very noticeable in Sigma PhotoPro, Picasa, and also ACR 5.4 but not all the time. It seems to happen more when a number of files are opened that don't yet have sidecar files. That is to say, if I open a single previously saved file in ACR, the yellow triangle does not always appear.

[dje]

I wonder sometimes about this 0.3 vs 0.5 pixel radius argument. I found this article recently which gives quite a good insight into the US mask method. Basically it says

• The US mask is produced by using a Gaussian distribution to assign contribution weighting to the surrounding pixels
• The pixel radius translates to the Standard Deviation (SD) of the distribution
• For values of SD of 1 to about 3, the number of contributing pixels out from the centre roughly equates to the SD ( with rapidly diminuishing contributions as you go out).
• For values of SD less than 1, it is only the first set of pixels out from the centre which contribute. The smaller the SD, the lower the contribution. In other words, the "radius" stays at 1 but the amplitude of the halo reduces as you reduce SD below 1.

If this guy has it right, then it suggests to me that whether you use 0.3 or 0.5 shouldn't matter much, provided you use a lower "Amount" setting for 0.5 than for 0.3.

Something to ponder !!

PS : I may not have explained that very clearly but if you read the linked article you will get a more detailed explanation.

[xpatUSA]

The minimum radius in ACR is 0.5px, which is sometimes too wide - the UnSharp Mask in PSE goes smaller (my standard is 0.3px).

Prompted by the re-appearance of the sidecar file hacking thread, I changed the radius tag in one such file to 0.3mm. However, ACR was not fooled :-(

Dave (dje) - thanks for the link which is the best description of un-sharp masking I've ever read and, based on the description, I am inclined to agree re: 0.3 vs 0.5.

It occurs to me that ACR's sharpening could be used to restore scene sharpness (as in edge-spread, not appearance) but that could sink us into a morass of magnification, diffraction, DOF, atmospheric diffusion, lens abberations, etc. ad naus. But, for my simple photographic needs (table top stuff), i.e. same lens, same distance, same lighting - a little ACR might be a good thing. Some recent lens tests have already shown no overshoot in the edge spread function with ACR 5.4's default settings of 25% amount, 1px, 25% detail, 0% masking. Which probably shows how bad my lenses are ;-)

Worthy of an experiment, I reckon . . .

[xpatUSA]

Worthy of an experiment, I reckon . . .

I had a RAW file handy to play with: slant edge shot with SD10/70mm macro at about 1 meter distance, LED floodlit.

Opened in ACR a) with no sharpening, and b) with Adobe's default sharpening 25/1/25/0. Saved to .tif with no further PP and opened in QuickMTF:

None:


Default:


As can seen, no overshoot with the default ACR sharpening (apart from the abberant blue channel which I blame on the lighting ;-). The MTF at 30 lp/mm improved from 35% to 55%. The 10-90% edge-rise fell from 2.51px to 1.76px (less is better).

I conclude that ACR's sharpening with it's greater flexibility (more sliders), moderately applied, is indeed useful (for my purposes) as a "scene sharpness restorative" before more artistic processing in post.

[DanK]

Dave, very helpful link and post. This clarified a lot for me. A digital image is discrete, not continuous, which would seem to imply that none can't really have a radius < 1 pixel. Either a pixel is changed, or it is not. So it makes sense that fractional radius settings are a matter of amplitude, not actual pixel width.

I am much less experienced than many of the people who post about a multi-stage sharpening process (capture, creative, output), but logically, sharpening seems to me to have one or two stages, depending on how one thinks of it and how one is outputting the image. The first of these stages entails sharpening so that it looks as you want on the monitor on which you are doing the editing. This, it seems to me, encompasses both capture and creative sharpening. The second entails modifying that to account for an outputted image that will look different--either an on-screen image of a very different size, or a printed image. This is output sharpening.

If you use a pixel editor, I think it could make a difference if you sharpened more than once in the workflow. With ACR and LR, AFAIK, it doesn't. All you are doing with ACR and LR is adding to the xml file another mathematical adjustment, and I don't think the order in which you add the parameters determines the order in which LR actually performs them. Perhaps someone who knows the software better than I can either confirm this or debunk it.

Ted--back to the original question: for what I do, I find the sharpening in LR is adequate most of the time. I sharpen to taste, usually toward the end of the editing, although I may revisit this a few times. I then use LR output sharpening for both printing and exporting to the web. I use Photoshop only when I need something that LR won't do, or won't do adequately. The thing that pushes me to do that is rarely sharpening. When I do edit in Photoshop, I sometimes sharpen a few times using different methods (USM, smart sharpen, high-pass sharpening) because they sometimes produce very different effects.

In LR, I usually leave the detail slider in the range from default to 45 or so. I use the masking slider a great deal. To see their effects, hold down the alt key while you adjust one of them. That will give you a black-and-white image showing what is and is not sharpened based on the position of the slider.

[bedantabd]

Have you seen this tutorial?
DIGITAL PHOTO EDITING WORKFLOW

Yes Dave, I have gone through that link more than once. And as I have already said, most of my editing ends in ACR, including sharpening and noise reduction. Then I open the image in photoshop for converting it to JPEG. For the time being, I have kept it quite simple not to get overwhelmed by all those layers/masks/USM/LCE etc.

Also, here's mine (it hasn't really changed since 2010 if you substitute "CS5" for "Elements")

I think I should give it a try now. Thanks

[DanK]

Then I open the image in photoshop for converting it to JPEG.

No reason to do that. LR has an export to jpeg function that works flawlessly, even adding output sharpening if you want. Saves a large amount of time and does not create another huge file.

[jcuknz]

There are a thousand and one ways to sharpen and a thousand and one answers. I gave up unsharp mask because I can never remember what figures to enter in the boxes and simply use 'sharpen' or 'sharpen more' one button clicks or maybe I will do it twice. I used to always be on the watch for fringing but it never seems to happen these days ... occasionally for a change I use high pass .... but I am using a different editor to you, thank goodness, but I hope you follow my drift
I sharpen just before printing , a rare occassion these days, or posting to the web else making final projection file.

[bedantabd]

No reason to do that. LR has an export to jpeg function that works flawlessly, even adding output sharpening if you want. Saves a large amount of time and does not create another huge file.

Hi Dan, I'm not using LR. After exporting from ACR to Photoshop, I simply save it as JPEG and discard the PSD file. In case I want to re-edit, the original RAW file is always there.

[Colin Southern]

Hi Dan, I'm not using LR. After exporting from ACR to Photoshop, I simply save it as JPEG and discard the PSD file. In case I want to re-edit, the original RAW file is always there.

Hi Bedanta,

Do you realise that you can save to JPEG directly from ACR?

[bedantabd]

Hi Bedanta,

Do you realise that you can save to JPEG directly from ACR?

I'm afraid, I don't Colin. Anyway, in Photoshop I normally add a border to the image before saving it as JPEG. Now don't tell me that ACR can add borders................

[xpatUSA]

I'm afraid, I don't Colin. Anyway, in Photoshop I normally add a border to the image before saving it as JPEG. Now don't tell me that ACR can add borders................

In my version of PhotoShop (Elements 6) the ACR is a separate window which opens up from PSE. In that window, only basic editing is possible which does not include borders - as I'm sure you know already. Now when you click on the "Open Image" button at the bottom right of the ACR window, ACR converts your file into the Adobe working space, using whatever ACR settings you've chosen, and your file re-appears in the PhotoShop window (behind the scenes, what you see is no longer a RAW image, even though it says .NEF, or .X3F, etc, at the top). At this point, you are able to save the file as any available "save as" option (jpg, gif, png, tif and whatever else it says) - whether you post-process the image or not, makes no diff.

That is why there is no necessity to take the intermediate step of saving as a .psd - UNLESS you've used layers and might want to go back and change them for some reason.

This should help you understand the previous one-liner, assuming our software to be similar.