Image Decomposer

destination after Beta: analyzers

Functionality

To get output from this node you must connect something to the first 2 output sockets (xya and rgb), polygons is optional and only outputs faces when Filter? is off.

Takes the rgba components of an image currently loaded in Blender and decomposes them into xya and rgb Vertex-style sockets. xy are inferred by the number of pixels in the image and the image width. z doesn’t make much sense in relation to a pixel and was replaced by the Alpha channel of the pixel (a).

If you don’t have images loaded in the UV editor, they can be imported from N panel into Blender and loaded from there.

Inputs & Parameters

name	function
Skip n pixels	allows to sample a reduced grid of the image, every nth pixel in either direction.
xy_spread	the xy component of the xya socket can be multiplied to get a wider spread.
z_spread	this amplifies rgb, not a (which you can amplify yourself if that was needed.)
Filter?	uses a restricted eval to drop pixels using a simple typed command : example r < 0.8 and g > 0.4 (more below)

Outputs

name	function
xya	the x and y of the pixel, combined with the Alpha channel. The value of x and y are multiplied by xy_spread.
rgb	each (unfiltered) pixel component is multiplied by z_spread
polygons	this output will generate sensible polygon index list for xya when pixels are unfiltered.

Examples

The development thread contains working examples of this Node used as preprocessor for game maps.

Notes

The loaded image gets a fake user automagically, tho perhaps this should be optional.