Announcement

Collapse
No announcement yet.

Simulate liquid over sphere - How get rid of grid artefact ??

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

  • Simulate liquid over sphere - How get rid of grid artefact ??

    Hi,

    my question is probably really simple. I got a pretty complexe scene with a lot of round geometry that water simply goes over and then kind of dry. My problem is when the liquid particle disappear it let the liquid geometry with strange shape and artefact. I already increase my resolution to about 9 000 000.

    I made a simplify 3d scene of what is the problem. I link an image and there is the 3d Scene too.

    Thanks.

    Click image for larger version

Name:	Phoenix_FD_Liquid_Test_01.jpg
Views:	1
Size:	233.1 KB
ID:	882883

    Click image for larger version

Name:	Phoenix_FD_Liquid_Test_02.jpg
Views:	1
Size:	192.9 KB
ID:	882884

    Click image for larger version

Name:	Phoenix_FD_Liquid_Test_03.jpg
Views:	1
Size:	194.6 KB
ID:	882885


    https://www.dropbox.com/s/3gr8kvvwg8...ne_01.max?dl=0


    PS: I got another question after I install the nightly build. What is exactly "FLIP" in the Dynamic rollout under Material transfer ?

  • #2
    Originally posted by alexish657 View Post
    What is exactly "FLIP" in the Dynamic rollout under Material transfer ?
    (houdini but explains the differences a little - it's particle based, faster and more stable)
    https://www.sidefx.com/docs/houdini1...dop/flipsolver

    It's whats going to fix your problem.

    Comment


    • #3
      You are going to love the FLIP solver in comparison.. enjoy!
      Adam Trowers

      Comment


      • #4
        Thank you guys for the fast answer.

        So basically, to switch to FLIP mode and get rid of my problem I only have to switch the rollout menu to FLIP or do I have to change some setting to get a right simulation ?

        Because I already made a fast test before posting here, whit the same scene that I post picture below, where I only change the "material transfer" to FLIP and my simulation doesn't look right at all.

        Comment


        • #5
          in general yes, you have just to switch to flip, can you give an image to see what is the problem?
          ______________________________________________
          VRScans developer

          Comment


          • #6
            Hi,

            here you will find the comparaison pictures between (Left : Classic and Right : FLIP) the two "Material transfer method". Everything it the same except for the Material transfer method.

            Like you can see, the two simulation is not quite similar. The FLIP method seems to get rid of the grid artefact, however the liquid seem to not flow on the sphere. Even at the exit of the source cylindre, the FLIP method seem to act strangely. I don't have the wetting option tick.

            Click image for larger version

Name:	Phoenix_FD_Liquid_Test_Comparaison_01.jpg
Views:	1
Size:	152.1 KB
ID:	858464

            Click image for larger version

Name:	Phoenix_FD_Liquid_Test_Comparaison_02.jpg
Views:	1
Size:	149.1 KB
ID:	858465

            Click image for larger version

Name:	Phoenix_FD_Liquid_Test_Comparaison_03.jpg
Views:	1
Size:	152.9 KB
ID:	858466

            Click image for larger version

Name:	Phoenix_FD_Liquid_Test_Comparaison_04.jpg
Views:	1
Size:	169.3 KB
ID:	858467


            Thanks

            Comment


            • #7
              well, when you turn to flip, it automatically sets spf=1, because the classic advection requires in general higher spf and it's better to adjust it to new value.
              this is what can say from this images. seeing the images, i would suggest spf=4, try it and tell us the result
              ______________________________________________
              VRScans developer

              Comment


              • #8
                Hi,

                actually the Step by frame was at 8 in the last pictures. I try to do another simulation at a higher value and I will post the result here.

                Comment


                • #9
                  So here is the comparaison pictures between the two simulation with FLIP method.

                  -Left : Conservation = direct smooth : 30 Method FLIP = Step by frame : 4
                  -Right : Conservation = direct smooth : 40 Method FLIP = Step by frame : 40

                  Click image for larger version

Name:	Phoenix_FD_Liquid_Test_Comparaison_02_01.jpg
Views:	1
Size:	165.6 KB
ID:	858470

                  Click image for larger version

Name:	Phoenix_FD_Liquid_Test_Comparaison_02_02.jpg
Views:	1
Size:	174.4 KB
ID:	858471

                  Click image for larger version

Name:	Phoenix_FD_Liquid_Test_Comparaison_02_03.jpg
Views:	1
Size:	182.0 KB
ID:	858472

                  Click image for larger version

Name:	Phoenix_FD_Liquid_Test_Comparaison_02_04.jpg
Views:	1
Size:	184.1 KB
ID:	858473

                  Here is the link to the 3d scene

                  https://www.dropbox.com/s/lsgummfxvo...ne_02.max?dl=0
                  Last edited by alexish657; 01-10-2015, 10:37 AM.

                  Comment


                  • #10
                    I try the FLIP method to my actual 3d project and I got another problem...

                    Like you can see in the picture below, there is some particules created at the exit of the source emiter that appears like some random "splash". I need a constant flow of water like a hose, I don't want to see the water spray everywhere. The splash appears when I increase the SPF.

                    Click image for larger version

Name:	Phoenix_FD_Liquid_Test_Splash_Geometry_01.jpg
Views:	1
Size:	504.1 KB
ID:	858474


                    PS: I got another question. I try to create a river flowing through the rock. With the "Classic" Material transfer method I created a box and use it as initial fill up geometry to save some simulation time, but with the FLIP method all my liquid created by the box dissapear quickly avec spread everywhere.

                    Comment


                    • #11
                      let resolve the problems sequentially, jumping form one to another will give us no progress at all.
                      let start with the first scene, the one with the grid artifacts over the object's surface, can you send it to us? my suggestion was low spf, because the liquid has visible segments, each segment is corrseponding to single simulation step, but it may be something more complicated. so, can you send us the scene, and describe what final look you are pursuing, what it must be, water, or chocolate, or milk.
                      ______________________________________________
                      VRScans developer

                      Comment


                      • #12
                        Hi Ivaylo,

                        you will find a link to the first 3D scene down below. There is 3 simulation in the scene that is perfectly identical except the Method and the SFP.

                        I want a water aspect material that flow over the rocks.

                        In the last picture of the river scene that I send, I presently use the Classic method wich give a ok result but there is some problems at some places where the geometry is more complex (waterfall, round rock, etc) so this is why I start to do the test with the sphere.


                        https://www.dropbox.com/s/lsgummfxvo...ne_02.max?dl=0

                        Thanks.

                        Comment


                        • #13
                          well, i played with your scene, see no problems, it behaves as expected. the previews of the flip and the classic sims are attached, if you find the behavior not satisfying tell us what exactly is not like you need to be.
                          Attached Files
                          ______________________________________________
                          VRScans developer

                          Comment


                          • #14
                            Hi,

                            it's been a while since I reply, we got a little rush at work so I was unable to take the test further.

                            Like you have noticed it's seem that the Flip method is better and ok for what I wanted to do.

                            So basically, to put everybody in context, my main 3D scene is a natural river into wich some water flows. Then, when my river flow constantly, some blood is mixed in the river to transform the water almost completely into blood.

                            So there is my problem.

                            Like you can see in the picture below, when I use Flip method and I increased the outgoing velocity to be abble the fill my river geometry correctly, I got some particle spraying around the exit of my source. When I do some troobleshooting, it's seem that is created when I increased the SFP paramter. How can I get rid of it and keep my velocity ? The problem also occurs in the test scene that you have already take a look at. I link a video of at the left SFP 4 and on the right SFP 40

                            Click image for larger version

Name:	Phoenix_Splash_Exit_01.jpg
Views:	1
Size:	494.4 KB
ID:	858666

                            https://www.dropbox.com/s/46x89st8tm...40_01.avi?dl=0


                            EDIT:

                            It seems that I got another problem. Like you can see in the picture below the "edge" of the liquid that end on the side of the river appears to be round like if my liquid have viscosity properties . I need that the liquid end realisticly like a lake should do on the rock.

                            Click image for larger version

Name:	Phoenix_Round_Liquid_01.jpg
Views:	1
Size:	373.0 KB
ID:	858678

                            Click image for larger version

Name:	Phoenix_Round_Liquid_Wire_01.jpg
Views:	1
Size:	492.8 KB
ID:	858679

                            Thanks !
                            Last edited by alexish657; 22-10-2015, 08:49 AM.

                            Comment


                            • #15
                              can you tell the build revision? it's shown in the about box. i'm asking because we had similar issue that is already fixed, perhaps the build is old?
                              ______________________________________________
                              VRScans developer

                              Comment

                              Working...
                              X