OC Sharpening in Photoshop

Copying over one of my old reddit comments to see how well it transfers over (with formatting and such).

Sharpening workflow

Sharpening is generally split into three steps;

  1. Input sharpening
  2. Sharpening for effect
  3. Output sharpening

1. Input Sharpening

Input sharpening is done to compensate for the slight blur introduced by the capture device (camera/scanner). You should optimally do this when processing the raw file in CR/LR.
It will generally be a small radius and low strength (you don't want any halos). Zoom to at least 100% to see what you're doing.

2. Sharpening for Effect

Optional step. This is where you use sharpening creatively to enhance the image. Making important areas "pop" a bit more, or increasing local contrast. You can use Layer Masks to limit the effects to specific areas.

3. Output Sharpening

Output sharpening is done to compensate for the slight blur introduced by the output device and to make the final image appear as sharp and crisp as possible at the final resolution and view distance.


Methods for sharpening in Photoshop

There are several methods you can use for sharpening. There is no clear superior filter/method, and they generally give you very similar results (sometimes identical).
However, the different filters/methods usually have some different options available to you, and some might fit your workflow better than others. So at this point it's mostly down to personal taste and the workflow you use.

Filter: Unsharp Mask (USM)
This is the "basic" sharpening filter in Photoshop. Tried and trusted it performs essentially the opposite operation of a Gaussian Blur. It gives you strength/radius sliders, and also a special "Threshold"-slider. By increasing the Threshold slightly you are telling the filter to not sharpen the contrast between nearby pixels if they have very similar values. This will help reduce the sharpening of noise in the picture, and the sharpening will be more focused on edges.

The filter will have to be applied to a flattened/merged copy of your image, or you will have to convert your layers to a Smart Object first.

Filter: Smart Sharpen
A more "intelligent"/modern sharpening filter. I would recommend this one (if set to "Lens Blur" it is the most accurate, if set to "Gaussian blur" it is identical to USM). In addition to the basic radius/strength sliders it also has a "Reduce Noise" slider that serves a similar purpose to the Threshold-slider in Unsharp Mask (I personally never use it).

It also has options for selectively reducing the amount of sharpening in shadow or highlight areas that you can experiment with, although I've yet to really need those options. Usually it's not worth the time to start tweaking that since we're talking very minor differences at this point, but it's there if you need it (e.g. perhaps your shadow areas have a lot of noise and you need to dial back the sharpening).

The filter will have to be applied to a flattened/merged copy of your image, or you will have to convert your layers to a Smart Object first.

Filter: High Pass
This one is often hailed as super-amazing sharpening of the professionals (with no reason given). But in fact it is almost identical to, and sliiightly less accurate, than the basic USM. You are also limited to just a radius slider, and no easy way to limit the effects of noise. So what's the big deal?
I believe the reason why it is so often used (besides "someone told me it was good") is because of the different way you would add this into your layer stack. In other words, it's a workflow issue.
To use High Pass you add a merged copy of your image at the top of the layer stack, set the blending mode to Overlay (or Soft Light, or Linear Light) and run the High Pass filter. Adjust the layer Opacity/Fill to control the strength (allowing you to do this after the filter is applied).

This means that if you want to go back and do some further fine-tuning of your adjustment layers (brightness, color tint, etc.) you can do this on the adjustment layers below without having to redo the sharpening. That can be convenient!

Wrapping all your layers into a Smart Object and applying a regular sharpening filter (USM/SS) to that is theoretically more flexible and non-destructive, but having to open up the Smart Object if you need further adjustments can sometimes be a bit slow compared to just having the sharpening as a layer.
It's up to how you prefer to work. Note that when High Pass first became popular for sharpening, Smart Objects/Smart Filters did not exist, so this is likely a contributor to why it is so wide-spread today.

"Filter": Frequency separation
If you are familiar with frequency separation then you may know that it's technically more accurate than the High Pass filter. Simply use the high frequency layer from the separation the same way you would use a High Pass layer.
This will create identical sharpening results to Unsharp Mask (minus the Threshold option), but done as a blended layer similar to the HP method. Since the method includes performing a frequency separation this sharpening method is only useful if you create an action for doing it (otherwise it is too slow).

Edit: An awesome thing here is that you can use other blur filters, like Surface Blur, for other results. Surface Blur is great for allowing a stronger effect on surfaces without getting too strong halos on edges.

Ps: This is the method I usually use (unless I'm applying sharpening to a Smart Object; then I use Smart Sharpen).

Filter: 3rd part filters
There are 3rd party applications that specialize in Sharpening. They may or may not give better sharpening results than the new Smart Sharpen in Photoshop CC, I have not tested (as I personally don't have the need). The differences on the final output image are likely to be negligible in the vast majority of cases though.

However, if you are willing to pay it might be worth trying out. They are likely to have convenient and easy-to-use built-in functions for handling noise, edge masking, reducing harsh halos, etc. as well as convenient predefined presets for different outputs. Nothing you can't do manually (or using Actions) in Photoshop, but it's a question of simplicity/convenience and speed (and how it would fit into your workflow).

Additional techniques:

Sharpen Luminosity
To prevent enhancing color noise, and to avoid halos getting a color tint you can limit your sharpening effect to the "Luminosity" (brightness) of the image.

If you are using the HP/FS method this means desaturating your sharpening layer.
If you are using a regular sharpening filter, just blend it using the "Luminosity" blending mode.

Edge Mask
Edge masks are masks used for limiting the sharpening effect to edges (very useful together with HP or FS method to limit the effect on smooth surfaces like skin, sky, etc.). For input sharpening in CR/LR you have sliders for this, but in Photoshop you have to generate your own (usually this would be something like Find Edges + Invert + Curves + Maximum + Median + Gaussian Blur).
Make an action for it.

Surface Mask
An inverted Edge Mask. Hiding/reducing the sharpening near edges.

Blend If
The "Blend If"-sliders in the layer blending options can be used to reduce the strongest halos (assuming your sharpening is done as a separate layer), effectively allow you to use a stronger sharpening strength to bring out the "texture" a bit more, while dialing back on the effect in areas that already has high-contrast edges. Remember to hold down Opt/Alt to split the sliders so you don't get a sharp cut-off.
Just throwing it out there as an option for you guys to play with...


Sharpening Settings:

When sharpening, what radius should you use? There are no "perfect" values to use that will work everywhere, so any recommended values you see should be considered as good default values, and then you go up/down from there depending on what you want.
The contents of your image, personal taste, and how it will be output and viewed all affects the settings.

With higher resolutions or view distances you need higher radii. Double the view distance or PPI = double the sharpening radius.

A good rule of thumb is that for screen use on a typical desktop monitor (resampled to the correct size and viewed at 100% size) you need 0.3-1 px. See what suits your image - it's easy since you (usually) see the final result live.
For most print scenarios 1.2-3 px is a good choice; my personal default is 2 px. If you are delivering a high-resolution image to someone who will be printing it but you don't have control over the final output size, I'd go for 2 px. It should look a bit over-sharpened when viewed at 100% on screen (if you want a better idea of how the sharpening will be on print, zoom to 50%)!

HERE is a great website with more information on sharpening, and a calculator you can use to calculate an recommended sharpening radius (just input PPI and view distance).

Some numbers from the calculator in table form:

View distance/Output PPI 100 PPI 150 PPI 200 PPI 300 PPI 600 PPI
25 cm 0.4 px 0.6 px 0.8 px 1.2 px 2.4 px
50 cm 0.8 px 1.2 px 1.6 px 2.4 px 4.7 px
1 m 1.6 px 2.4 px 3.1 px 4.7 px 9.4 px
5 m 7.9 px 11.8 px 15.7 px 23.6 px 47.2 px
10 m 15.7 px 23.6 px 31.5 px 47 px 94 px
100 m 157 px 236 px 315 px 472 px 945 px

[Wow, the table worked!]

Note: Not covered in this (sligtly dated) post is new Ai-based sharpening techniques that has the ability to generate new details, or the latest update to sharpening in ACR. Also, some images are imgur-hosted and might break in the future.

  • All
  • Subscribed
  • Moderated
  • Favorites
  • photoshop
  • DreamBathrooms
  • magazineikmin
  • InstantRegret
  • mdbf
  • Youngstown
  • slotface
  • Durango
  • khanakhh
  • thenastyranch
  • rosin
  • kavyap
  • cubers
  • GTA5RPClips
  • Backrooms
  • JUstTest
  • ethstaker
  • tacticalgear
  • everett
  • osvaldo12
  • hgfsjryuu7
  • tester
  • modclub
  • anitta
  • normalnudes
  • cisconetworking
  • Leos
  • provamag3
  • lostlight
  • All magazines