Results 1 to 9 of 9

Thread: initial state - rest position

  1. #1

    initial state - rest position


    What is the best approach for setting up objects that stand on the ground before being fractured/broken? It seems like objects need some time on the ground before they come to rest (depending on he material of course). Also in all tutorials walls, the chairs etc. seem to float on the ground a little.

    Is the best approach to include passive-regions close to the ground? Or use enough preroll??

  2. #2

    Re: initial state - rest position

    The thing to remember about objects is that DMM does something called "freezing". When something doesn't move for a number of frames, DMM stops simulating it automatically. If you leave something right on the ground, and it doesn't move, you should not have to do anything, as DMM will simulate for a few frames and as the object won't change, it will remain absolutely still. You can also use passive regions if you like, to force freezing.

    BTW, objects are automatically "woken up" as they are touched or our connectivity trace notices something that shouldn't be floating because all its support has been removed. For the real-time version, we expose something called "Wake-Up Radius" which is a parameter that lets you control how many tets around an active region to wake up. Wake-up radius doesn't really make sense to have in the Maya version as the vast majority of the time you can use keyed passive regions to achieve the same result.

    Hope this helps.

  3. #3

    Re: initial state - rest position

    Thanks - yes that helps.

    But the objects are moving for a few frames (sinking into the ground a little etc.). Is that normal because of the material-settings (I tried wood and concrete)? It seems like everything needs some time to settle (and shrink slightly). Maybe it has to accepted as preroll (kind of like softbodies)?

    Thanks again

  4. #4

    Re: initial state - rest position

    It can happen. Try increasing the number of substeps in the scene setting, that can help, though it will make the sim run slower. I would try 20 and possibly 100 substeps on the finicky shot frames.

    The whole sinking issue is being worked on now in an update. We have a better collision response model that will get perfect collision response. We hope to have it in an upcoming Maya Hot Fix.

  5. #5
    I can see on your Trestle example, that there is no sinking at all. It looks perfect. Is there a way to let the objects be passive until first hit or how was this done?

  6. #6
    We do something called "freezing" in a scene to minimize simulation overhead. When something moves very slowly, we stop simulating it. Most simulations work out so that they are frozen in their initial state because they are not (a) floating (we do a connectivity trace to ensure things don't float...) or (b) There are no forces being applied to them. The sinking issue is being worked on, we may be able to provide a quick solution with some fixes we did for the high end VFX work we have been doing. A longer-term, more comprehensive solution is being worked on as well, but that is going to take a little longer to implement. So far it is looking really really good though

    To answer your specific question about how to keep things from sinking - you need to:
    1) Avoid stacking lots of stuff. Keep everything as a single mesh
    2) Use passive regions to keep tets you want to be still from moving
    3) Try and "weld" meshes together where you want absolute solidity. Welding involves forcing a tet cage to share verts. Its great when you can do it because it saves the overhead of glue (which is currently kinda expensive to use). I will post an article today of how to do that.
    4) Increase the number of substeps. I know this is a pain, but decreasing the simulation time allows the solver to more accurately determine when something should stop moving.

    I'll see if we can get some of the sinking fixes out in our next update.


  7. #7
    well, the freezing approach sounds pretty perfect to me. Is that custom-made or available with every DMM-version?

    So far I'm using passive regions and quickly animate them before impact .. not the best workaround but it works for now

    BTW, I worked at MPC and saw Kali in production.. pretty awesome! Unfortunately I had to work on water-splashes not rbds

    Thanks for your help.

  8. #8
    The freezing is automatic, so you shouldn't have to do anything. We used to allow people to set a freezing radius in the plug-in, but that has been taken out. People tended to start using it in weird ways which ended up with nonphysical adjustments. The freeze radius is still a setting in DMM and is often used in the real-time version.

    The approach you are using is probably fine. I wish I could offer you a better approach. We have some things I think we can do with the force pushback in the contact model, so I'll see if we can add those to the next version.

    Those MPC guys have done a great job with Kali. The shots I have been seeing come out of that have really been fantastic.

  9. #9
    ok, but then maybe I'm doing something wrong .. maybe it's also a matter of scale/volume/density?

    Lets take the most simple example: On a passive ground I'm creating an elongated cube (like a standing column). Depending on the material I'm giving the cube starts to collapse under its own weight - or at least shrinks and bend a little.

    With the freezing, are you saying that this cube shouldnt move at all? Or is the freezing-algorithm something that is activated once the object is fractured (and colliding with the ground)?

Similar Threads

  1. Initial resting postions Probem
    By lozo3d in forum Discussion
    Replies: 1
    Last Post: 11-04-2010, 05:16 PM


Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts