Add boundary mode to cycles/eevee Voronoi node
voronoi node is missing a very important feature that people have been trying to recreate for a long time. most commonly i think people settle with mixing two voronoi nodes with one slightly offset or rotated but it gives uneven results that aren't easily modified. other times i've seen very complex node setups try to recreate the boundary lines when it really should be something that's available from the node itself.
i'm suggesting adding a boundary mode similar to the current cells and intensity modes it already gives.
the closest thing i think we've got in blender right now to what i'm suggesting would be the brick texture node's mortar size input and factor output.
have you ever strung together a musgrave texture factor to a color ramp (for fine tuning darken the white) to a math multiply x .125 to the brick mortar size input, from the factor out (or color out is neat too) to the color input of a diffuse bsdf?
if you do and then picture voronoi boundries instead of uniform brick lines you'll probably start to come up with some quick and simple ideas as to why what i'm suggesting would be useful.
other than that this also might give you some ideas as to why it could be useful. especially when mixed with the other two voronoi modes.
3 Comments
I second that request, was looking for exactly this just last week. The rest of my Voronoi wish list (basically, stippling requirements): an output of each cell center's coordinate as a dot, size from input value/cell size relaxation from local color/grey scale input
@netgrazer: i've tried to recreate this in various ways a number of times over the years. never with satisfactory results.
likely for the same reason you can't really get your center dots with the node as it is now, the intensity mode doesn't always have a full 0-1 range from center points to edge boundary.
some of the cells are a strange blur where if the edge is 1 the center point would sometimes somehow instead of having a value of 0 have a value of .5 or .4 or something other than 0. it really breaks the expected functionality you should be able to get with just the factor out and a color ramp node.
I third the motion, specifically, I believe Cycles' implementation of Voronoi should include the F2 output as well, which would give the node much more flexibility (and easy cell boundary detection) without adding much complexity in the typical use case.