Announcement

Collapse
No announcement yet.

Motion blur with Bercon Metaballs

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

  • #31
    Hmm, so been doing some more testrenders, and both workarounds seem to have problems...

    the first (using MB samples 1 and motion vectors) seems to go randomly nuts on some frames... the sequence I rendered gets to frame 89 or so, then the particles seem to blur in random directions. When it goes this the particle count spikes too.

    (Plus the amount of memory used is nuts. With 16GB ram I am still getting memory errors)

    The second workaround, will leave a gap in the particle streams (the lead particle in the drips is bigger, as rivulets of water tend to be, but if the spawns are a frame later, there is a gap in the stream)

    R.

    Comment


    • #32
      Yeah I managed to get the particles go nuts too with Lock & Bond operator (mb samples 1).

      Using the pflow velocities obivously works in metaballs end since it doesn't care what events you have in your particle system, it just takes postion and velocity and blurs it. If it works in one case, it should work in all cases. With this reasoning the problem is in pflow.

      The motion blur works fine with creating/deleting particles mid frame (birth/delete ops) and works fine when particles are delayed 1 frame before rendered after spawn operator (this shouldn't change anything?). This also makes me to suspect there is something wrong with pflow because these cases should all be similar. Either all cases should fail or all should work if the problem was in the metaballs.

      Since all the problems seem to be in pflow I can just come up with workarounds that are not perfect.

      I can try asking from the developers of pflow to see what they think about these issues but since pflow falls under Autodesks responsibility I doubt you'll see any fixes in near future.

      Of course there is the very unlikely possibility that the problem is somehow in how I handle pflow in my code. However, it just should not work it like does now if that were the case.

      Sorry that I can't help you more than that.

      - Jerry
      http://www.ylilammi.com/

      Comment


      • #33
        Originally posted by MasterBercon View Post
        Yeah I managed to get the particles go nuts too with Lock & Bond operator (mb samples 1).

        Using the pflow velocities obivously works in metaballs end since it doesn't care what events you have in your particle system, it just takes postion and velocity and blurs it. If it works in one case, it should work in all cases. With this reasoning the problem is in pflow.

        The motion blur works fine with creating/deleting particles mid frame (birth/delete ops) and works fine when particles are delayed 1 frame before rendered after spawn operator (this shouldn't change anything?). This also makes me to suspect there is something wrong with pflow because these cases should all be similar. Either all cases should fail or all should work if the problem was in the metaballs.

        Since all the problems seem to be in pflow I can just come up with workarounds that are not perfect.

        I can try asking from the developers of pflow to see what they think about these issues but since pflow falls under Autodesks responsibility I doubt you'll see any fixes in near future.

        Of course there is the very unlikely possibility that the problem is somehow in how I handle pflow in my code. However, it just should not work it like does now if that were the case.

        Sorry that I can't help you more than that.

        - Jerry
        Would it not be possible to make the plugin ignore particles moving past a certain velocity threshold? (they would be so blurred you would barely see 'em anyway!

        Is this not practical? (I have no clue, but cant hurt to ask!)

        R.

        Comment


        • #34
          Yeah that is possible. Actually here is a version to do just that: http://www.ylilammi.com/BerconMetaba...0_64_BETA2.dlo

          When Pflow vel. thresh. is negative (-1.0 for example) everything work as before. However when its positive, anything with higher velocity is ignored. So if you set it to 0.0 all moving particles are ignored.

          The speed units its using is the movement in world space during 1 frame. So if your particle size is 1.0 then their maximum movement could be something like 5.0 or 10.0, above that they'd be very blurred.

          I also contacted somebody with deeper knowledge of pflow and he said that the problems with spawn and velocities might indeed be a bug in pflow. He'll take a look at it when he has some spare time.

          - Jerry
          http://www.ylilammi.com/

          Comment


          • #35
            Originally posted by MasterBercon View Post
            Yeah that is possible. Actually here is a version to do just that: http://www.ylilammi.com/BerconMetaba...0_64_BETA2.dlo

            When Pflow vel. thresh. is negative (-1.0 for example) everything work as before. However when its positive, anything with higher velocity is ignored. So if you set it to 0.0 all moving particles are ignored.

            The speed units its using is the movement in world space during 1 frame. So if your particle size is 1.0 then their maximum movement could be something like 5.0 or 10.0, above that they'd be very blurred.

            I also contacted somebody with deeper knowledge of pflow and he said that the problems with spawn and velocities might indeed be a bug in pflow. He'll take a look at it when he has some spare time.

            - Jerry
            Hey - testing now, this seems to work pretty well - Thank you!

            Comment


            • #36
              More problems!

              I am using cached particles, but different machines seem to render the motion blur differently - the motion jitters. They are motion blurring it differently. Really baffling (and frustrating, as you can imagine!)

              Any idea what could be causing this? And more importantly, how do I fix it!

              Best,
              R.

              Comment


              • #37
                could this be because they are different types of cpus? Or should they render the same because the particles are cached?

                Comment


                • #38
                  Any ideas? I've torn most of my hair out by now. Next I will be gouging out my eyeballs and stuffing them in my ears.

                  Comment


                  • #39
                    did you ever get a fix for this motion blur problem ??

                    claude
                    Originally posted by 3DMK
                    do I want to be a rich business man or a poor artist?

                    caddworkx

                    Comment


                    • #40
                      Originally posted by claude_tnt View Post
                      did you ever get a fix for this motion blur problem ??

                      claude
                      Yes its working now.

                      I believe *part* of the problem was to do with how pflow feeds motion vectors to bercon - unless you cache the particles using subframe samples, it doesnt know where to point them, and they end up going random crazy directions. This was one fix of several, and I'm not 100% sure exactly what was going on under the hood, so a lot of this feels like voodoo to me. So in the end, the only real solution was to build a small shrine and to sacrifice several chickens and small children, and dance around on hot coals drenched in their blood, screaming "hail satan!" at the top of my lungs. Then my motion blur issues mysteriously vanished. The only issue now is that I'm probably going to hell. Oh well. At least my renders are working.

                      R.

                      Comment

                      Working...
                      X