Local Contrast Enhancement

[den]

...

As has previously mentioned, for the current PWP v7.0.19 Sharpen - Bilateral Sharpen transform only has a maximum Blur radius of 20, yet JS's white paper has examples with settings greater than this... ...the Blur radius range is to be increased in a future release.

The following is a suggested work around if you wish to explore 'effective' Blur ranges greater than 0 to 20:
...(a) downsize the intended Input image by 50% will provide an 'effective' 0 to 40 Blur Radius with a transform setting of 0 to 20...
...(b) downsize the intended Input image by 33% will provide an 'effective' 0 to 60 Blur Radius with a transform setting of 0 to 20...
...(c) downsize the intended Input image by 25% will provide an 'effective' 0 to 80 Blur Radius with a transform setting of 0 to 20...

For example, (1) the image: DSCF1289.tiff [4896x3264pixels, 48 bits] was downsized by 33% with the Resize transform and Bicubic method; (2) Sharpen-Bilateral Sharpen LCE was performed at a Blur Radius = 60 with other preference settings on the downsized image; (3) the LCE image version was upsized by 300% to its original dimensions using Resize and the Bicubic method...

An image area of the Starting image modified for posting...

ScrnView(22)_Ori.jpg (97.3 KiB) Viewed 2440 times

The LCE image area of the Final image with the same modifications for posting...

ScrnView(22)_BS-LCE_Radius60.jpg (128.12 KiB) Viewed 2441 times

...an apparent benefit to the downsize and upsizing seems to be smoothing of small noise artifacts allowing one to use higher values of Sharpen Factor's then is suggested in JS's white paper...

Something else that has merit is to 'over cook' the upsized LCE image version and then: preference Composite-Blend and/or 1:1 Clone image areas from it to the original Starting image...

...have fun experimenting...
...den...

[tomczak]

I've been doing something similar with AS, not knowing the name for it...

I always wandered why I usually liked the results using AS more than BS. It may have something to do with the 'un-thresholded' blurring step producing different results.

The advantage of BS is that it seems faster than it used to be, and also faster than AS for large radii.

[tomczak]

Just out of curiosity, 3 different blurs of the original crop of Jonathan's image, with radius 16. The radii may not mean the same thing between the blurring methods, I imagine, but the results are visually different because of the blur computations, not only the amount of blur.

c.f. AS and the original from the previous post.

[tomczak]

Two (semi-academic) questions re. Bilateral Sharpen:

1) Is sharpening done on each RBG channel separately, or on HSV-V?

2) Is there a way of 'softening' Sharpen Threshold?