#204 Tessellation problem for Extrusions mixing Arcs and Beziers

unexpected behavior

Tessellation routines seems to present problems for sketches
mixing arcs with Beziers and arcs with arcs.

Attached herein there's an example of correct behavior
(file correct.g) consisting in a sketch composed of closed
beziers and circles that extrudes and converts well to (for
example) STL format:

$ g-stl -o correct.stl correct.g ext
428 triangles written

In addition there's the problematic example mixing arcs
and beziers in a topological correct closed sketch (see incorrect.asc
for details about correctness of vertex linking) that extrudes well on mged
and can be raytraced without problems but makes g-stl complain
about some issue regarding the fact that sketch loop is not closed:

$ g-stl -o incorrect.stl incorrect.g ext
rt_extrude_tess: A loop is not closed in sketch sk
tessellation failed!!
nmg_booltree_leaf_tess(ext): tessellation failure
db_walk_subtree() FAIL on '/ext'
0 triangles written



  • aegio

    aegio - 2009-02-03

    Two examples of tessellation correct Vs. incorrect behavior

  • John Anderson

    John Anderson - 2009-02-05

    The problem with the sketch in "incorrect.g" is that the segments are not consistent in their direction. Every segment (line, circular arc, bezier) has a start vertex and an end vertex. For a loop to be closed, each end vertex must be the start vertex of another segment. In "incorrect.g" there are segments that start at a vertex that is the start vertex of another segment, and segments with an end vertex that is the end vertex of another segment. Reversing the direction of two of the segments in "incorrect.g" will allow g-stl to succeed. This change can be accomplished using the command:
    db adjust sk SL { { bezier D 3 P { 3 2 1 0 } } { bezier D 3 P { 4 5 6 7 } } { carc S 0 E 4 R 0.8669825810741810245829697 L 1 O 0 } { carc S 7 E 3 R 0.8617633860033965254032751 L 1 O 0 } }


Log in to post a comment.

Get latest updates about Open Source Projects, Conferences and News.

Sign up for the SourceForge newsletter:

No, thanks