Idaho Airships, Inc.

Let’s explore the Soft Light Blending Mode on a Fill Layer, as applied to an area of profound haze in an aerial image.

In researching the previous Soft Light Blending Mode post I found myself producing more questions than answers, and decided to run a brief experiment in order to resolve a few of the fundamental issues I was struggling with.

In review:

Symbology:

* Resulting Color = A
* Blend Color = B
* Background Color = C

If the Blend Layer Channel value exceeds .5, the application per Channel is:

• A = (2B - 1) * (SQRT(C) - C) +C

If the Blend Layer Channel value <= .5, the application per Channel is:

• A = (2B - 1) * (C - C^2) + C

This calculation will limit the “Gamma” change to between .5 and 2, and will not force any aggregate mean or Channel value to the margins (0 or 1).

Our test image is an aerial photograph of Tamarack Ski Resort in central Idaho, 7/10/07. This is the image as recorded on a 5D in “Neutral” Mode and aRGB. There were a number of wildfires in the region, accounting for the smoke/haze.

(Click to Enlarge). Our test image has received very limited preparations of:

• A simple Levels Adjustment Layer for Global Contrast, with the Heel and Shoulder adjusted as shown in the graphic
• A Curves/Screen Adjustment Layer (to lighten the image), Opacity 15%
• A Curves/Hard Light Adjustment Layer (to produce additional Contrast in the image), Opacity 25%.

Our test Pixels (x, y) were chosen primarily in areas of significant haze where Contrast and Saturation were profoundly affected.

The original values from the partially prepared image shown directly above were:

• Red (50, 25): 241, 246, 254
• Green (1400, 200): 205, 225, 249
• Blue (2400, 600): 214, 227, 248
• Yellow (1400/1800): 97, 116, 135

Changing the Fill Layer in Soft Light Mode (see the above “Click to Enlarge” image) to a value of 80, 80, 80 produces the following result:

• Red (50, 25): 236, 242, 254
• Green (1400, 200): 190, 215, 247
• Blue (2400, 600): 201, 218, 245
• Yellow (1400/1800): 75, 92, 111

Changing the Fill Layer in Soft Light Mode to a value of 10, 10, 10 produces the following result:

• Red (50, 25): 228, 238, 254
• Green (1400, 200): 168, 200, 243
• Blue (2400, 600): 183, 204, 241
• Yellow (1400/1800): 42, 57, 76

Changing the Fill Layer in Soft Light Mode to a value of 45, 10, 10 produces the following result:

• Red (50, 25): 232, 238, 254
• Green (1400, 200): 179, 200, 243
• Blue (2400, 600): 192, 204, 241
• Yellow (1400/1800): 58, 57, 76

While these results (posted so that they may be criticized-so please do!) are limited in number and scope, they do suggest or corroborate that:

1. Soft Light does not induce clipping. It’s effects moderate toward, and are nonexistent at the extremes. This is also obvious in the math.
2. Soft Light does not introduce an appreciable bias when applied to its neutral color (gray) or any luminance variation of gray (even R, G & B values).
3. Soft Light does allow a per-Channel customization of bias when desired.
4. Soft Light provides additional control in an image that has modest, previously applied Contrast manipulations.
5. Soft Light provides therapeutic benefit to the instance and amount of haze represented by our test image.

L