Raw video feed from the webcam.

Pixel colours are converted to HSV (Hue, Saturation, Value) and a threshold filter applied (if the value doesn't meet the threshold it is rendered as black). Most of the other filters work off this, adjusting colours only if the pixel is within the threshold range. The 'blur sample' checkbox samples from a larger square of pixels (3x3) to get an average then works off that instead of the centre pixel.

It's suble, but you can see it if you're close. If value is skin colour then average from surrounding pixels to smooth it a little.

Increase R value (Red) if pixel is skin colour. Skin is already a little red.

Increased G value (Green) if pixel is skin colour. Skin is red so we also decrease R & B values to accentuate the effect.

Increased B value (Blue) if pixel is skin colour. Skin is red so we also decrease R & G values to accentuate the effect.

Greyscale. R, G, B values are averaged, e.g. (R+G+B/3) if pixel is skin colour. Should make face greyscale.

Pixelate Skin. If skin colour then take colour from pixel at nearest 5 value, e.g. 5, 10, 15, 20...

If pixel is NOT skin colour then reduce R, G, B values to 40% of original.

If pixel is NOT skin colour then R, G, B values are averaged, e.g. (R+G+B/3).

If pixel is NOT skin colour then blur (average colour from surrounding 4 pixels).

If pixel is NOT skin colour then pixelate (sample from nearest rounded 5 coordinate on a grid).

This filter applies to the whole image regardless of skin colour. Yellow is #FF0 so we increase the B value and slightly reduce the R & G values.

This filter applies to the whole image regardless of skin colour. Similar to the correct yellow balance filter but with more emphasis on reducing the R component.

R, G, B values are averaged, e.g. (R+G+B/3). This filter applies to the the whole image regardless of skin colour.

Same as Greyscale filter but also darkens pixel to 40% of original if not skin colour .