Shows only an image’s edges.
Channels
— Which combination of image channels to find the edges of.Radius
— The distance, in points, to blur the image before finding its edges. Larger values are helpful for reducing noise.Intensity
— How bright (or dark) the edges are.Threshold
— The cutoff point when deciding between edges and non-edges. Larger values hide softer edges.Range
— How to convert the edges into a visible image. The edge colors depend on whether an object is lighter or darker than its background:
Unipolar
— Against a darker background, interior edges are dark, exterior edges are bright, and non-edges are grey, producing a nimbus effect. Negating Intensity
reverses the edges, and produces an effect like embossing or relief printing.Bipolar
— Against a darker background, exterior edges are bright, and interior edges and non-edges are dark.Absolute
— Both interior and exterior edges are bright, and non-edges are dark.This node applies a mathematical transformation to the image to find its edges — it blurs the image and finds its 2D partial derivative (the “Laplacian of Gaussian”).
Keywords: Laplacian of Gaussian, LoG, Marr, convolution, edge detection, emboss, filter, relief