PP2014 render (smoothing?) artefacts on a morphed prop

  • I have a base OBJ mesh that I import into Poser and it renders fine.
    I have an OBJ morph target for it that I can import into Poser as aseparate object and that too renders fine.
    However, when I load the MT OBJ as a morph target for the original imported base and apply it with value of 1.0 I get artefacts.

    Anybody know why ?

    Anybody know howto get round this ?


  • @3dcheapskate Hi, looks like a normals issue. If you export the morphed version do the normals look weird in the modeling app?

  • Just gone back to Blender to check the meshes and I find I have a couple of duplicate vertices. Need to fix that and retest to make sure that's not the culprit.

  • @amethystpendant Normals look correct in Blender.


    Also got rid of the two duplicate vertices and redid the morph.Same problem still, so it wasn't that either.

    By the way - the vertices of each quad are coplanar for both the base OBJ and MT, and all internal angles of the quads are less than 180

    (Importing the exported OBJs back into Blender (just in case) the normals look correct too)

  • Poser Ambassadors

    I don't know if this is important or not, but the original statement regarding both A and B render fine independently is not what I see. I see an artifact on B.


  • I may be wrong, but the model B seems to have some degenerate triangles. You move the vertice from the Sword edge well behind the next edge (row of vertices). So you have some vertices very close, but unwelded.

    I would try it in a way that the edges do not cross:

  • Poser Ambassadors

    My understanding is that all render engines triangulate all geometry at render; maybe this is having an impact? A hard-surface object like this is going to be utterly unforgiving, & that morph is a big change in the geo so I'm wondering if the renderer is struggling when it's triangulating the mesh with the morph. Are there any polys that are even slightly non-planar? Maybe use some tri's instead of all quads? Just a thought.

  • Thanks for the responses:

    @bagginsbill Every oddity's worth double checking. Nothing else visible in the scene for that to be a shadow. Artefact only appears in my Poser 6 render, not the PP2014 render of the same scene. The original objects A and B used for that scene/render had two duplicate vertices, as I mentioned in a subsequent post. Using the corrected object B the Poser 6 artefact disappears. Checking the duplicate vertices in Blender they seem to be the culprit. Screenshot of first duplicate vertex attached - the other duplicate is of the corresponding vertex on the bottom edge.

  • @bantha Just double checked 'degenerate triangle' - means all three vertices are colinear. Depending on which way Poser splits the quads (as Caisson mentioned) around the notch/nick into tris the triangles there are several close calls - BDE, ABC and BCD on the attached screenshot


    "I would try it in a way that the edges do not cross:" the vertices/edges in the second column (orange in the screenshot below) are intended to remain vertical and colinear as they are the edge of the 'sharp bit' (technical term). Pulling the vertices from the sharp edge back past them was the only way I found to make the notch and (a) keep this vertical line, and (b) avoid coincident vertices, and (c) avoid quads with internal angles greater than 180°.
    Your example (the green screenshot in your post) is an approach I didn't of, moving the central vertex to the side, so I may give that a try. The bottom face on the inside of the notch could end up non-planar (as it appears to be on your screenshot), and it would be easy to get interior angles of quads on the vertical surface greater than 180°.But forewarned is forearmed ! :o)
    0_1494296050408_vertical edge.jpg

  • This post is deleted!

  • @caisson All the quads are planar to best of my knowledge, but there's always rounding errors. Bantha's comment about degenerate triangles is interesting and fits your comment about using a few triangles - maybe if I manually triangulate the specific problem quads to avoid the potential near-degenerate triangles ?

  • Test 3 - Manually triangulating the potential problem quads in the base mesh doesn't seem to help.
    (The morphed lower notch still gets those long thin triangles that Poser doesn't like. But on the unmorphed Obj B3 the long thin triangles are the same but the artefacts aren't there in the render. It seems to me that it's more to do with "that morph is a big change in the geo" (as caisson said, with the specific word highlighted by me). Maybe something to do with parts of the smoothing being done in accordance with the base mesh paramaeters ? (just a wild guess)


    Edit:just confirmed same render artefacts in PP2014

  • Removing smoothing and setting crease angle nice and low to 5 (started with test 3 scene with the manually triangulated mesh and made the adjustment noted)

  • Question: the Firefly renderer. The only references to it I can find are in relation to Poser. Was it developed specifically for Poser, or was it taken from somewhere?

  • This is IIRC, I did not find a reference any more. Firefly was licensed from another company, but I do not remember the name. When I'm not mistaken it was Mac only before. But that was FireFly as in Poser 5, no idea how much it still behaves like the original beast.

    What happens with the morph dialed in at 0.9? When do the artifacts start?

  • @bantha Up to 0.5 they're not really noticeable.

  • I've been slogging away at this, trying to get to the root of the problem.
    Artefacts show up on a simple OBJ (no MTs) in the preview pane with smoothing disabled.
    If I reduce the crease angle from default 80 I find a value below which a particular artefact will disappear.
    The artefact covers the whole of the face adjacent to the edge inquestion, although the lighting setup often makes it appear as if it's only one triangular half of the quad.
    Adding a control edge close to the problem edge limits the artefact to the face between the edge and control edge.


  • Yep, it needs control edges. This is just a quick test slapping in a couple of extra edge loops, but I think that's nailed the culprit and a possible solution/workaround:
    I may have to rethink the modelling of the notches to get better control edges.

  • But of course those control edges will also restrict the curvature of the surface to the area between the edge and the control edge.
    And I've also got the sharp edge above the notch (red arrow) breaking open when I render with Smooth Polys.Reminds me of this https://forum.smithmicro.com/topic/746/why-does-this-mesh-split-on-one-edge-with-smooth-polys-pp2014?page=1
    (Edit: but that red arrow is pointing to the old lower notch,which I didn'tfix with control edges! The problems probably still in the upper notch but restricted tothe control edge and thus too small to see)

  • @3dcheapskate aRtBee researched the history of the the plugins including firefly in his Missing Manuals page and it is very interesting reading here: