Page 1 of 2 12 LastLast
Results 1 to 10 of 13

Thread: DMM Objects Intersecting

  1. #1

    DMM Objects Intersecting

    Hi there,

    Anyone know why my DMM Object are intersecting each other after a straight forward break?

    I have a floor and a wall and am simply allowing the wall to fall onto the floor and break but there seems to be a fair bit of intersecting going on with pieces of the wall with other pieces and also the floor - which obviously looks bad and not very realistic.

    Any ideas?

  2. #2

  3. #3
    It has to do with the way we calculate collision forces. We use tet-tet overlap to determine how much force to apply to slow down or stop interpenetration. You can increase the accuracy of this by increasing the number of simulation steps per frame. Increasing it to 5, 20 or even 100 can yield good results, though it will result in a slower simulation. We are working on a better contact model that will be fast enough to use interactively on an upcoming release.


  4. #4
    i routinely crank the sub steps up to usually around 128... although it does take a bit longer to calculate the results are far better and sim time is still really fast.

    I think the key to getting predicable results is to establish scene scales. Do some test at various sizes and find the one that returns the best results.

  5. #5
    I worked with Scene Scale 1, .1 and .001. Also substep as high as 128.
    So I came up with an idea.
    playback by 0.5 or 0.2. So I think we are multiplying subsetp (Correct me if I am wrong). I never tried 256 substeps, if it is possible then it is equal to playback by 0.5.
    (Again please correct me if I am wrong, I am telling by my own understanding).
    Also I thought if I can model a Tet Meshe dense at the area of contect. But I think for that I need more in depth Idea about its structure.

  6. #6
    Substeps have nothing to do with the frames per second being played back. Substeps are how many times the solver iterates per frame. Not sure why you are mentioning Scene Scale - that doesn't have anything to do with steps or fps.

    You also do want to avoid having tetrahedra that are several orders of magnitude different in size interact with each other, keep your tets close to the same size in interacting objects.


  7. #7
    Thanks. For clearing my doubt.
    Is there any limit for substeps ?

  8. #8
    to be clear I didnt mean the internal scene scale in DMM I meant the global scale of your maya scene that you are simulating. When I found the right Maya scale to build my DMM sims at everything started to look better behavior wise.
    Another thing worth a mention that has helped me out a bunch is when I convert my DMM sim to Geo Cache I can choose any substep in evalution and the calculation is smoothly interpolated. This is great for me because i retime that cache in maya and it is super smooth even when making the sim say ten times longer

  9. #9
    That is a good trick! How do you handle exposed faces with fracture in the GeoCache? I think it works quite well if you don't have fracture, but when things start breaking, inner faces don't show. That's why we created our own cache. In 2013, Alembic seems to work pretty good, but that's because it creates a snapshot of all geometry per frame.


  10. #10
    There isn't a set limit, but you will likely reach a point of diminishing returns after hitting 500 in that the sim will run much slower.


Posting Permissions

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