Smooth mesh

  • Angel Angel Comment actions Permalink

    A smoother simulation (less creasing) is based on the poly mesh faces being set to a smaller particle distance. However this can greatly (exponentially) slow down your cloth simulation.

     

    Method one: to get less faceting by the polygon side is to reduce the patterns mesh particle distance in the property editor for each pattern piece.

     

    Method two: Involves using your external modeling pipeline where the garment will be sent, as a sub-divided mesh in quads. This actually involves a different mesh topology for your entire garment cloth model. So you need to decide if this is the workflow you want to invest time into. So to get a quad based cloth garment mesh out of MD you would :

     

    Model all the cloth elements in tri-mesh (20mm mesh particle distance) under GPU for work speed to get a final garment assembly using the tri-mesh simulation as that gives the best folding and fidelity for drape

     

    After running the garment assembly with a CPU (best quality) beauty pass simulation at higher mesh particle distance (5 mm mesh particle distance) for the final pose stop and save the project (save state).  Then shift into the pattern (garment) retopology tools, creating a surface copy in quad mesh over your tri-mesh high poly count master.

    Export the quad mesh model with welding into an .obj or .fbx into your next software application and pipeline for texturing, sculpting, detailing, animation. Here is where you can now sub-divide the low poly quad garment (retopolgised) in MD onto the high poly tri-mesh garment.  And the subdivided qauds will now make all the surfaces smooth. But it relies on having a good topography and using the highpoly tri-mesh simulation as the guide/master.

     

    Option three: Don't export a retopo mesh model from MD - but instead generate all pattern pieces as quad mesh using the algorithm in MD to get a uniform qaud that can be subdivided. This is an algorithmic interpolation that can create poor loopline flow and numerous poles in the mesh topography such that the normal when rendered will give slight surface light distortion when rendered externally. So although this can be the fastest method to get quads and very smooth surfaces they are not necessarily in the right technical format for managing good sub-division should you be shifting the model to another pipeline such as gaming, animation or sculpting. So although this method might appear enticing - you really need to think about the quality of the quads embedded in the finished model. If in doubt do manually retopography (method2).

Please sign in to leave a comment.