Announcement

Collapse
No announcement yet.

Round corners on intersecting meshes,

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Round corners on intersecting meshes,

    As we all know, booleans and bevels are two of Maya absolute weakest points when it comes to modelling. Any boolean work has a tendency to send maya Polygon models to a very bad place, and Rhino is overkill for some projects.

    While VRay round edges is very useful in some circumstances it really misses out on the option to work on intersecting meshes. Mentalray has this, and it's proven a god-send on so many projects (all the micro welds on small components)

    eg: Intersecting a large amount of cylinders into another tubular surface causes obscene amounts of normals problems after a boolean operation (even on high fidelity meshes). In mental ray, one doesn't need to perform a boolean to make them *look* connected. Just round round edges.

    First "and hopefully last" wish list for me. Vray has surprised me at every turn with how much it does have covered.

    -RH

  • #2
    agreed it also works really well on wet part of the eyes... where it meets the eyelids.

    i'd like that as well!

    A

    Comment


    • #3
      I've mentioned this before, I think... Master Zap (the creator of the mental ray shader) has a patent on pretty much the only feasible way to implement this with a shader (without doing an actual boolean mesh operation). So even though it is possible to write a similar shader for V-Ray, we won't be doing it.

      There are some other approaches to achieving a similar result f.e. based on the VRayDistanceTex algorithm in 3ds Max, but they are more complicated to set up.

      Best regards,
      Vlado
      Last edited by vlado; 06-04-2011, 01:37 PM.
      I only act like I know everything, Rogers.

      Comment


      • #4
        Just out of curiousity, what is that method ? Is there a paper available or alike ?

        Regards,
        Thorsten

        Comment


        • #5
          im guessing since its not described anywhere that - maybe- it works a similar way to master-zaps method, but hasnt been automated so no one can get sued


          ive almost cracked it using a distance tex in the bump slot and a vray edged tex in the far slot of that. then a white colour in the close slot. set the distance tex to the same radius as the edges tex, and link the distance tex to the second object. since circular references are not allowed in max you have to create a hidden copy of the first object if you want the second object to have the radius too. link the distance tex of the second object to the dummy copy of the first object.

          now if its in the diffuse slot, i get, as expected a white line along all the mesh edged from the edges tex, plus a white line of the same thickness along the intersection between the two objects.. to my mind this means it works, but when i put in the bump slot i only get fillets from the edges tex.. so something is not quite right.

          somehow the output of the edges tex (a white line along edges but -transparent elsewhere, using the base diffuse colour swatch - if in the diffuse slot) gives a different result to the distance tex (white edge along intersection but black elsewhere)

          maybe someone could help me crack the final issue and get it working...?

          Comment


          • #6
            Originally posted by instinct View Post
            Just out of curiousity, what is that method ? Is there a paper available or alike ?
            There is the patent itself, of course: http://www.google.com/patents/about?...son+H%C3%A5kan

            Best regards,
            Vlado
            I only act like I know everything, Rogers.

            Comment


            • #7
              Originally posted by super gnu View Post
              im guessing since its not described anywhere that - maybe- it works a similar way to master-zaps method, but hasnt been automated so no one can get sued
              No, not at all. The mental ray shader uses raytracing to sample the normals around the shaded point, whereas VRayDistanceTex does not involve any raytracing (VRayEdgesTex does not involve raytracing either, it is based on static analysis of the mesh shape, but has no information about other intersecting meshes).

              Best regards,
              Vlado
              I only act like I know everything, Rogers.

              Comment


              • #8
                Originally posted by super gnu View Post
                maybe someone could help me crack the final issue and get it working...?
                Yes, will try to do that a bit later...

                Best regards,
                Vlado
                I only act like I know everything, Rogers.

                Comment


                • #9
                  doh.... patents!

                  A

                  Comment


                  • #10
                    How about Round Edges added to the Vray Displacement property set?

                    Comment


                    • #11
                      Originally posted by super gnu View Post
                      maybe someone could help me crack the final issue and get it working...?

                      Originally posted by vlado View Post
                      Yes, will try to do that a bit later...

                      Best regards,
                      Vlado
                      Hey again
                      -Super Gnu, did you make any progress with this one ? The problem is rearing it's ugly head again for me, so was just wondering. (Currently the thought of having to convert the project to MR makes me itch... And you seemed to have a good grip on it.)

                      /RH

                      Comment


                      • #12
                        hmm.. i didnt play with it since.. was waiting for vlado to offer his suggestion


                        as far as i know my system should have worked, but there appears to be some difference between using the distance tex to put a thin white line along the intersection edge, and using the edges tex to put a thin white line along the object edges, when used in bump. the edges tex produces a radius, the distance tex does nothing i can see... dunno why.

                        seems the edges tex produces a white line on the edges and -nothing- everywhere else.

                        distance tex puts white edge and black every where else...

                        dunno how it works basically. thought i did// appears i dont :/

                        Comment


                        • #13
                          Originally posted by super gnu View Post
                          hmm.. i didnt play with it since.. was waiting for vlado to offer his suggestion


                          as far as i know my system should have worked, [SNIP]
                          distance tex puts white edge and black every where else...

                          dunno how it works basically. thought i did// appears i dont :/
                          Ahh, but it was a valiant effort, nonetheless! Thanks for the fast reply on this one at least. I'll take stock of things a little and see whether I have to switch to another solution.

                          Much appreciate that someone else has had a look as well though
                          /RH

                          Comment


                          • #14
                            This is what I have so far; will post the scene in a moment. The lighting doesn't show the effect too well, will try a better scene too.

                            Best regards,
                            Vlado
                            Attached Files
                            Last edited by vlado; 24-05-2011, 01:54 AM.
                            I only act like I know everything, Rogers.

                            Comment


                            • #15
                              Well good grief that's a huge step in the right direction!. Certainly useable for what I need it for at the moment (micro-welds).

                              Very excited, and thank you Vlado for the heads up!

                              Comment

                              Working...
                              X