Digital Light & Color

Ref 1: retouch_images_with_frequency_separation
http://www.computerarts.co.uk/tutorials ... separation

Ref 2: HighPass Sucks (+ solution)
http://www.modelmayhem.com/po.php?threa ... ost9585669

Ref 3: Frequency Separation Technique [7:34 minute video]
http://www.youtube.com/watch?v=dMyaeZmkZD8

There seems to be an interest in low/high spatial frequency imaging on many of the other message boards... to smooth skin colors/texture and sharpening... just Google/Bing: "frequency seperation photoshop"

To date, I have yet to find a PWP workflow that would include smoothing textures/colors using an image's low/high spatial frequencies as is suggested in the Photoshop references [actually, BilateralSharpen can do this with the right settings] ...but there is a possible PWP manual sequence to perform sharpening with the high spatial frequencies. This sharpening is more subtle and less prone to excessive light/dark halos than the normally used HighPass-SoftLight/HardLight sequences that have an equal blur radius as that used during the construction of the high spatial frequency image version... the following is good for both 24/8-bit and 48/16-bit images...

Basic Sequence: conservative light/dark tone changes with possible objectionable changes in colors [large radius Local Contrast Enhancement]...
(1) Click on the 'starting image' and open Blur-Gaussian method. Set the Radius to preference [perhaps 3->10 for sharpening or 25->75 for Local Contrast Enhancement]; Threshold and Amount = 100; and click OK.
(2) Click Transformation\Gray\Negative... a negative of the 'resulting (1) image' should be created.
(3) Click on the 'starting image' and open the Composite - Blend Operation. Set Overlay = 'resulting (2) image' and Overlay Amount = 50. Click OK, creating the High Spatial Frequency data image.
(4) Click on the 'starting image' and open the Composite - SoftLight or HardLight Operation. Set Overlay = 'resulting (3) image' and adjust either the Input or Overlay Amount to a preference. Click OK, creating the sharpened or local contrast enhanced image.

Additional Comments: Photoshop users use 'Linear Light' for (4). As this operation is not available in PWP, Soft Light or Hard Light needs to be used. Soft Light tends to have a softer effect and retain colors more so than Hard Light which may provide an hard edged low-saturated desiccated effect at its extreme.
A more Advanced Sequence will use an 'active' contrast mask of the 'resulting (3) image' [the High Spatial Frequence data image] as the Overlay and Overlay Mask of the (4) Composite - SoftLight or HardLight operation. This mask will produce a more 'dramatic' effect without changes in colors [i.e., effects luminance only] and allows one to preference the amount of light and/or dark tone changes for the sharpening or local contrast enhancement. Ask if more detail is needed...

Enjoy!!!...

Den,

Although the links (and some of your comments) indicate this sequence is specific to portrait/skin, you use a non portrait example, so I assume this may be used on other type images such as landscape? If so, a couple questions:

For sharpening, is this intended to replace the usual capture sharpening that comes either during a RAW conversion or immediately after, or is intended to be a creative sharpening process?

The radius range suggestions you quote in step 1, are they related to an amount or to image size? My images are about 6000 X 4000 and am wondering if I should be using the higher end of the range for the larger files.

What is the reason for the 50% Overlay amount in step 3, rather than 100% or other number?

I followed the sequence for the local contrast enhancement (using Soft Light in the final step), and believe I reached the intended conclusion on an image that contained a full dynamic range (in fact, the highlights are just hanging on the edge of clipping) and found the highlights had become a bit blown. Is a highlight specific mask to perhaps composite the resultant image with the initial image, the most appropriate way to manage this, or is the amount slider in one of the steps a better way (see question above re 50% recommendation)? I tried looking up Soft Light and Hard Light in the Composite Transform in the Help Manual, but neither are mentioned (although the other blend modes are described), so I am not entirely clear on what these two specifically act on. At the very least, it appears Soft Light is quite a bit less contrasty, depending on the type of image.

Thanks in advance

Marv

Marv... ...Hope the following is of help!!!...

As prefaced, I chose to only use the 'high spatial frequency image data' as an alternative for HighPass sharpening and local contrast enhancement... the Photoshop skin smoothing technique described in the references is done with three layers that have specific blending modes between each layer. PWP only has two layers [the Composite transform] with specific blending modes... so simultaneous changes to smooth colors and alter textures cannot be done in parallel as in Photoshop but potentially could be done serially in PWP... but PWP has other tools to smooth skin that are easier to implement than trying to emulate this particular Photoshop technique.

(1) Sharpening/LCE... ...this is intended to be a 'creative sharpening' sequence occuring near the end of the workflow where the 'starting image' [could be a landscape or a portrait for a 'grunge', 'goth', or 'dragan' ambiance] that has near preference tone/colors and can be used with a small radius blur to increase edge contrasts (sharpening), followed perhaps with another application with a larger radius blur for Local Contrast Enhancement... the same as if you were using HighPass sharpening (small radius blur) and HighPass local contrast enhancement (large radius blur).

(2) Blur Radius... ... for me and 3456x2304 image resolution, the 3 for sharpening and 25 for LCE work well, so you may be correct to use the middle to upper range values...

(3) Step(3) Overlay amount... In Photoshop, a scale setting of 2 is used when they Add the 'starting image' with its 'inverted' [negative] blurred image version to obtain a primarily 50% gray image result. In PWP, Composite-Blend adds and 50% (1/2) Overlay is needed for an Input image at 100, so that the blended image would will be primarily 50% gray... ...if the Overlay is blended at 100%, the resulting image would be the same as the Overlay image. Primary 50% gray images, except for the blur radius changes are needed for the Soft Light and Hard Light filter Operation modes in Step(4).

[Note: try creating a High Pass blurred image version with the same blur radius used in Step(1) with the 'resulting Step(3) image... they should look very similar except that the 'high spatial frequency' 'resulting Step(3) image' will be more refined than the High Pass image... but both will be primarily 50% gray... also compare the 'high resolution histograms' of these two images... the 'resulting Step(3) image' will have a more concentrated histogram about the 50% vertical... ...more so than the High Pass image where both have the same blur radius.]

(4) Composite - SoftLight or HardLight. These two filters act the same way, HardLight having a stronger effect than SoftLight. Both when used as the Overlay image: will 'lighten' when its tones are greater than 50% gray; cause no change when equal to 50% gray; and 'darken' when its tones are less than 50% gray; the tones/colors of the Input image. The easiest way to control loss of highlight and/or shadow detail with the suggested "Basic Sequence" above is to simply decrease the Overlay image Amount in Step(4) with the Input Amount remaining at 100... ...and/or switch from Hard Light to Soft Light for the Composite transform Operation mode.

Luminance-Only SoftLight & HardLight illustration/example: (1) Composite Input & Overlay @ 0, 25, 50, 75, 100% tones and (2) Preview - SoftLight & Preview - HardLight...

Den,

Thanks for the extensive reply. As I am not as conversant in many of the terms you, and other forum members, use, it will take me a while to make my way through your explanation to ensure I understand it!

Thanks again,

Marv