Digital Dream Studio V2

Blending Modes

In image processing, amongst others, there is a problem which very often arises: let there be a pixel B which needs to be overimposed upon an underlying pixel A. How can this be done? There are literary tens if not hundreds of solutions to this problem, each one best fitting a certain case.

The most common solution is opacity blending: it means considering an opacity in the [0;1] range and obtaining a blend value by the ponderate arithmetic mean a*(1-o)+b*o. Of course, here opacity can be either constant or variable. However this method is often insufficient.

So a suplementary blending function has been introduced. Thus the blending can be considered not between the original pixel and the new pixel but between the original pixel and a function of the new and the original pixel. Consequently, the following equation arises: a*(1-o)+f(a,b)*o. The only problem now is picking a function f.

Here are the blending functions (a,b shall be considered in the range [0,1])

Normal - the result is the new pixel

Darken - chooses the darkest of pixels

Lighten - chooses the lightest of pixels

Hue - takes the hue from the new color and the luminance and saturation from the original color

Saturation - takes the saturation from the new color and the luminance and hue from the original color

Color - takes the hue and saturation from the new color and the luminance from the original color

Luminance - takes the luminance from the new color and the hue and saturation from the original color

Multiply - f(a,b)=a*b; the result is always darker than each color; if one color is white, the other is left unchanged; if one color is black, the result is black

Screen - f(a,b)=1-(1-a)*(1-b); the result is always lighter than each color; if one color is black, the other is left unchanged; if one color is white, the result is white

Multiply (soft) - f(a,b)=sqrt(a*b); the result is the geometric mean of the colors; if one color is white, the other is lightened; if one color is black, the result is black

Screen (soft) - f(a,b)=sqr(1-(1-a)*(1-b)); if one color is black, the other is darkened; if one color is white, the result is white

Dodge (NL) - increases gamma of the original image, in regard with the brightness of the second image

Burn (NL) - decreases gamma of the original image, in regard with the brightness of the second image

Dodge (L) - increases the brightness of the original image in regard with the brightness of the second image

Burn (L) - increases the brightness of the original image in regard with the brightness of the second image

Vivid Light

Pin Light

Hard Light - combination of screen and dodge, depending on the new color

Soft Light - similar to overlay mode

Overlay - combination of screen and dodge, depending on the old color

Contrast - increases contrast of the first color in regard the the second one's brightness

XOR - performs bitwise XOR of the colors, normalized as [0;255] integers

OR - performs bitwise OR of the colors, normalized as [0;255] integers

AND - performs bitwise AND of the colors, normalized as [0;255] integers

Additive - f(a,b)=a+b

Differential - f(a,b)=|a-b|

Complementary - f(a,b)=|a+b-255|

Negate - f(a,b)=255-|255-a-b|

Exclusion - f(a,b)=a+b-a*b

Reflect - f(a,b)=a^2/(1-b)

Glow - f(a,b)=b^2/(1-a)

Heat - f(a,b)=1-(1-b)^2/a

Freeze - f(a,b)=1-(1-a)^2/b

Burn (Natural) - f(a,b)=1-(1-a)/b

Dodge (Natural) - f(a,b)=a/(1-b)

Burn (Inversat) - f(a,b)=1-(1-b)/a

Dodge (Inversat) - f(a,b)=b/(1-a)

Burn (Soft) - This mode is a combination of color burn and inverse color dodge mode, but a lot smoother than both of them.

Dodge (Soft) - This mode is a combination of color dodge and inverse color burn mode, but a lot smoother than both of them

Soft Light (PS) - the version of soft light used in Photoshop

Soft Light (XF) - the version of soft light used in XFader

Geometric Mean - this changes the blend equation to power(a,1-o))*power(b,o)

The remaining blend modes are slightly different versions of the ones presented above (channeled, natural, grey etc).

The dissolve mode chooses at random the old or the new pixel, with the probability depending on global opacity.

The Invisible mode does not consider the new color in at all.

The Visible mode does not consider the old color at all, disregarding opacity.