Digital Light & Color

Being totally new to 'doctoring' portraits, I have a question about the 'beautifying' skin technique: in several articles there is a description of a technique which takes the RGB-R channel and blends it with the rest of the layers in 'Luminance' blending mode (I'm not sure what it actually does), making the face brighter and arguably more beautiful. Does anyone knows what it actually does and why it seems to make faces look more glamorous and how to translate it to PWP?

The best effects I achieved so far is to use Composite on the Original, with RGB-R as an overlay and the blending set to either Soft Light or Subtractive Filter (in Composite or in Filter transformations), but I don't understand why or how it really works (or if it really does at all).

Cheers!

In the following illustration, the dark/light sRGB MacBeth skin colors/tones were Composite-Soft Light and Composite-Subtractive Filter with their extracted RGB-R channel as the Overlay... ...the resulting HSV parameters are indicated: Whether this is an improvment or not depends on ones goals...

The illustrated changes in the HSV parameters can also be accomplished with: (1) HSV-V and HSV-S curves using Color Curves transform or (2) changes of hue/saturation/tone of selected skin color/tone control points in the Color Correct transform... ...perhaps accompanied with a skin color/tone range mask to ensure limiting changes to skin image areas.

You can probably get better results using Transformation/Color Correction and adjusting the skin tones directly - either by changing hue and saturation or brightness.

I tried it, but had a problem: shredding the continuity: even with the max radius. Colour Remap seems to work better for me - but I still think that channel juggling could be the smoothest overall.

tomczak wrote:I tried it, but had a problem: shredding the continuity: even with the max radius. Colour Remap seems to work better for me - but I still think that channel juggling could be the smoothest overall.

One could use the Extract-ed RGB-R channel Convert-ed to 8-bit BW if applicable as a gradient mask when using either the Color Correction transform or Color Curves... ;-)

Maciej,

I don’t *know* what is meant by “beautifying”, but the process you described using a soft light blend with the red channel is a way of enhancing contrast. Indeed, experimenting with the other channels as overlays offers some selectivity for the final result.

While experimenting on the same theme, I discovered that the HDR/Stack Images transformation with a original image composited with one or more of its B&W channels (they have to be converted to 24/48 bit first) is quite flexible, thanks to its simultaneous brightness and density curves adjustments, plus masks capability on each component image.

As for 'beautifying' I don't really know - I see that the red channel for instance seems to have the least small-scale skin imperfections - so smoothing the other two may be the best way to go, while retaining some realism. Also, the Asian beauty canons seem to favour pinkish-white, 'porcelain-like' skin - i.e. as far away from 'reddish-brown' as possible... I'm not sure; I'm still learning - trying to gauge it by reactions...

Cheers!