Announcement

Collapse
No announcement yet.

Best way to render smooth liquids using Phoenix ver 2.25.01 ?

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

  • Best way to render smooth liquids using Phoenix ver 2.25.01 ?

    Hi All,

    I'm back using Phoenix now after a few months so am trying to get used to the new workflow with Nightly build version 2.25.01

    Using the new PhoenixFDLiquid001 simulator (rather than enabling 'liquids' in the official build) with Liquid source PHXSource001.

    My first question --- is the liquid specific simulator based on the FLIP solver? - I can only see FLIP in the Advection method options in the new 'FireSmokeSim' PhoenixFDFire001

    The reason for my question is that I used to render liquid particles (generated from the standard / universal grid simulator) by using PHXFoam, with the liquid particles set to 'splashes' mode. I found that it gave a smoother result than rendering the grid cells using the 'mesh' mode.

    With the new setup in ver 2.25.01 what is the best way in practice to acheive a smooth surface on the liquid surface? (I am currently rendering the PhoenixFDLiquid001 in Mesh mode using particles to smooth the result but it's a but 'wobbly')

    Thanks for any help!

    Thomas
    Director, Flowstorm Ltd

    www.flowstorm.co.uk

  • #2
    the liquid simulator is based only on flip solver, the reason that you can see flip in the fire simulator is because we used initially one simulator object for all kind of simulations (liquid and fire), andwe had to add a temporary way to access the flip solver, the most logical one is to add it as advection method, because technically it is exactly this.
    the usual way uo render liquid is in mesh mode, sometimes splash mode particle shader also can be applied, but this is rather an exception (for example this veido)
    the mash mode has tools for additional smooth, but first you must be sure that the simulation itself is smooth enough. the way you using, mesh + particle based smooth is the best one. can you show an image and to share more details what are trying to do?
    ______________________________________________
    VRScans developer

    Comment


    • #3
      Hi Ivaylo Katev,

      Thanks for your reply and explanations.
      Here are some images showing what I'm trying to do and the Phoenix settings being used.
      Basically I am trying to create an 'Oil Pour' sim showing engine oil pouring out of a bottle.
      The scene units are in cm, models are at correct physical size but I am using a 'scene scale' in dynamics of 10 to try and make the oilpour look a bit slower/smoother.
      I did try playing with the Timescale parameter but it didn't seem to help. (Sorry can't remember what happened but I reset that to 1.0)

      Key parameters I have being playing with are - Viscosity, Non-newtoniam, Surface tension.

      I also tried icreasing the number of cells but seemed to get more problems so returned to this fairly low-res solution.

      Do you have any advice on how I can improve this? - ie) make a nice smooth sim without the bumps / wobbles?
      I am particularly keen on improving the realism of the shape of the oil at the edge of the 'spout' - as this is what the camera is focusing on.

      Thanks again,
      Thomas

      Click image for larger version

Name:	031_oilpour_Ver017_PHX_settings.jpg
Views:	1
Size:	507.2 KB
ID:	859890Click image for larger version

Name:	frame_095.jpg
Views:	1
Size:	91.9 KB
ID:	859891Click image for larger version

Name:	frame_164.jpg
Views:	1
Size:	92.4 KB
ID:	859892Click image for larger version

Name:	frame_203.jpg
Views:	1
Size:	92.0 KB
ID:	859893Click image for larger version

Name:	frame_328.jpg
Views:	1
Size:	250.5 KB
ID:	859894
      Director, Flowstorm Ltd

      www.flowstorm.co.uk

      Comment


      • #4
        Phoenix settings again as that image looks hard to read :

        Click image for larger version

Name:	031_oilpour_Ver017_PHX_settings_crop.jpg
Views:	1
Size:	476.2 KB
ID:	859895
        Director, Flowstorm Ltd

        www.flowstorm.co.uk

        Comment


        • #5
          what kind of bumps you are trying to avoid, the mesh looks smooth for me
          ______________________________________________
          VRScans developer

          Comment


          • #6
            See images - the 'waves' and the spikes (this is not the only one).

            I guess these issues are physically correct in terms of the maths going on.... but I'm just wondering if I'm going about this the right way?....adjusting the correct parameters in order to get the right look.
            My aim is to produce a smooth oily liquid that 'looks' like oil pouring.
            At real scale the liquid seemed to pour too fast so I increased the scene scale to 10 in the hope it would calm things down a bit.

            Do you think the waves on the top liquid surface are due to the the large scene scale?

            Are the spikes most likely caused by a wrong surface tension or viscosity?

            Do you have any guidance on the relationship between real materials viscosity and the viscosity?
            Any pointers much appreciated.

            Click image for larger version

Name:	frame_203_text.jpg
Views:	1
Size:	233.8 KB
ID:	859900Click image for larger version

Name:	frame_328_text.jpg
Views:	1
Size:	538.7 KB
ID:	859901
            Director, Flowstorm Ltd

            www.flowstorm.co.uk

            Comment


            • #7
              now it's clear what you mean
              about the waves -it may be simulation related, if you use initial fill up with a shape geometry object, the particles are created ordered, it is good when the liquid will be disturbed suddenly (splash or fast movement), but is bad for slow changes, it's visible how the order disappears. you can shake the object and wait until the order is destroyed, and then start the animation.
              about the leak - it's almost sure that is geometry related. i can't see the geometry, how can we see it together with the liquid?
              about the spike - have no idea, it may be also geometry related. perhaps is a good idea to see the movement of the particles in this spike
              ______________________________________________
              VRScans developer

              Comment


              • #8
                Thanks again for your help Ivaylo.

                FYI the initial fill up is created by dropping a large 'box' of fluid inside of the pouring geometry and allowing it to almost settle before animating the pour geometry - slow, gradual rotation to one side.
                The pour geometry is shown in the first set of images I posted - shown in blue wireframe. (A simple editable poly with turbosmooth - but no thickness so maybe that's the cause of the leak.)
                I hadn't had much trouble recently with leaks when using the FLIP solver so hadn't bothered here with thickness but will try and see if that helps.

                I get the feeling I still have some playing with parameters to do in order to learn better how they all interact.
                Will run some tests in the background whilst I work on other shots.

                Here is a render :

                Click image for larger version

Name:	031_shot_04_oilpour_AND_LID_0146.jpg
Views:	1
Size:	445.9 KB
ID:	859922
                Director, Flowstorm Ltd

                www.flowstorm.co.uk

                Comment

                Working...
                X