Nice result! There is a somehow dirty emboss approach that is based on the union or intersection of skewed copies. Although it is not perfect and F6 render may take its time, I sometimes use it because it is so straight forward. Note that *intersection_for* has convexity problems on F5. I removed all logic, to show you the bare principle: <http://forum.openscad.org/file/t887/taper.png> //taper_down(.2, .2) linear_extrude(1) text("hello world"); taper_up(.5, .5) linear_extrude(1) text("test text", font="Times New Roman:style=Regular"); module taper_down(x=0, y=0) rotate([0, 0, 180]) for(i=[-x, 0, x], j=[-y, 0, y]) skew(x=i, y=j) scale(-1) children(); module taper_up(x=0, y=0) intersection_for(i=[-x, 0, x], j=[-y, 0, y]) skew(x=i, y=j) children(); module skew(x=0, y=0, z=0) multmatrix([[1, z, x], [0, 1, y]]) children(); -- Sent from: http://forum.openscad.org/

Parkinbot wrote > Nice result! > > There is a somehow dirty emboss approach that is based on the union or > intersection of skewed copies. Although it is not perfect and F6 render > may > take its time, I sometimes use it because it is so straight forward. I > removed all logic, > to show you the bare principle: > ... Yeah, that looks way easier (and smarter), but isn't going to work well with diagonal intersections like the one in the x. -- Sent from: http://forum.openscad.org/