#14 bug in g-stl for windows?

release 7.8

"g-stl" provides for specification of a conversion
tolerance "-D", yet when I modify the value over
several orders of magnitude the number of
facets/triangles stays constant.

The geometry is a sphere-cone, and so decreasing
tolerance should result in increasing number of facets.


  • ecbngc

    ecbngc - 2006-11-07

    copy of g-stl, test1 containing rv.r, and rv.stl

  • Sean Morrison

    Sean Morrison - 2006-11-07

    Logged In: YES

    If I'm understanding you correctly, this isn't a bug -- there are four different tessellation tolerances that come into play. Run g-stl without any options to see the list, but they are -a (absolute), -r (relative), -n (normal), and
    -D (distance). They have to all coincide otherwise you'll just run into one of the other tolerance limits. To simply increase the number of polys quickly, the absolute tolerance is usually the most straightforward way to
    increase the count.

  • Sean Morrison

    Sean Morrison - 2006-11-07
    • assigned_to: nobody --> brlcad
    • status: open --> pending-rejected
  • John Anderson

    John Anderson - 2006-11-08

    Logged In: YES

    The -D tolerance is a calculational tolerance. Any two
    vertices within this distance
    of each other are considered the same vertex. This tolerance
    does not affect facetization.
    The options that adjust the facetization are -a, -r, and -n.
    The "-a" option specifies an absolute tessellation tolerance
    - the maximum allowed distance (mm) between the real surface
    and the facets
    The "-r" option specifies a relative tessellation tolerance
    - an absolute tolerance is calculated as the relative
    tolerance times the "size" of the object being tessellated
    (for a sphere, the "size" is the radius).
    The "-n" option specifies the maximum surface normal error
    (in radians).
    By default, tessellations are performed using a relative
    tolerance of 0.01. Try using the -r option with values other
    than 0.01.

  • ecbngc

    ecbngc - 2006-11-08

    Logged In: YES

    Thank you both -- issue resolved.

  • ecbngc

    ecbngc - 2006-11-08
    • status: pending-rejected --> closed-rejected

Log in to post a comment.