#219 Inconsistent alignment of vertices between wireframe and rt

closed-invalid
Raytrace (49)
5
2010-04-09
2009-07-05
No

I have 6 arb8s lined up, all equal height, tapering from 300 to 0 in the Y axis at their tops, and I'm trying to do similar from 200 at the bottom. Alignment in wire frame is fine, and all numbers in the primitive editor are correct, however, when I ray trace, the vertices of the resulting solids are not all correct. Attached is a screen shot of the render, and I'll also post (or provide on request if I can't) the database.

I think this is a bug, however, I'm new to mged, so if I've missed something, please let me know, and file this issue appropriately.

Thanks for your time,

James.

Discussion

  • James Hudson

    James Hudson - 2009-07-05

    Incorrect alignment of rendered vertices

     
  • James Hudson

    James Hudson - 2009-07-05

    The database. Offending shapes are hbu[1-6].s

     
  • John Anderson

    John Anderson - 2009-07-11

    The problem is with some of your ARBs. The ARB primitives are required to have planar faces. When the ARB's are ray traced, the following errors are printed:
    arb(hbu1.s): face 1485[3] non-planar, dot=-0.0122724
    arb(hbu2.s): face 1485[3] non-planar, dot=-0.0242372
    arb(hbu4.s): face 1485[3] non-planar, dot=-0.449104
    arb(hbu4.s): face 2673[3] non-planar, dot=-0.150787
    Raytracing an ARB is based on the planes of the faces, so the above faces are raytraced using the plane formed by the first three points of the face (ignoring the fourth point), resulting in an image that does not match the wireframe.

     
  • Sean Morrison

    Sean Morrison - 2010-04-09

    An interesting exercise for someone might be to pass non-planar arbs through the new brep code as it handles them just fine, but as things presently stand, this is a case of invalid geometry. Thus, you're getting inconsistent behavior. Make sure your arbs have planar faces and you should be good.

     
  • Sean Morrison

    Sean Morrison - 2010-04-09
    • milestone: 387263 --> other bug / workaround
    • assigned_to: nobody --> johnranderson
    • status: open --> closed-invalid
     

Log in to post a comment.