Announcement

Collapse
No announcement yet.

Phoenix for Maya production buglist...

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

  • Phoenix for Maya production buglist...

    Phoenix Zoic Wiki
    *
    …a few of the Maya plugin TIPS, TRICKS, WARNINGS, and BUGS we’ve encountered on the front lines of production at Zoic on the TV series “Once Upon a Time”. *Hoping others can benefit from this and/or help us solve a few of these problems.
    *
    TIPS
    (tips n tricks)
    *
    PARTICLES – Use particles to drive your sim instead of trying to push around fluid with fields. *Note that velocity may need to be set at a high level for smoke to inherit motion. *Note that low grid resolution may not even register particle emission.
    *
    SPF LOWER LIMIT (Ardy) Typically set to 1, a lower number may give you additional detail in your sim.
    *
    SMOKE – using temperature (as opposed to smoke or fuel) as a basis for smoke will allow you to control emission opacity and give you more detail, but uncheck gravity from the dynamics section of your sim.
    *
    WATER – using temperature as a basis for water can change the apparent gravity of falling water. *Use fuel instead.
    *
    MAYA FIELDS – are fairly slow. *If possible, use animated geometry to initiate movement and preview results with “velocity”.
    *
    VELOCITY – remember to always check velocity “on” in the output section of your sim. *This will allow you to add on to your sim later via “restore”.
    *
    *
    WARNINGS
    (some of this may be basic knowledge, but may not be readily apparent to first time users)
    *
    SCALING (Bob) Do not attempt to scale a sim twice its cached size when rendering foam. *It will break.
    *
    ADAPTIVE GRID (Bob) …will not adapt past the initial bounding box if the fluid is expanding on the “jammed” side. *It also does not seem to adapt to temperature alone.
    *
    TEMPERATURE (Bob) You can’t just emit temperature into a sim. *It must be accompanied by smoke to render, even though there is an option to render based on temperature.
    *
    PHOENIX HANGS (Dave) …when referencing files. *Use import instead.
    *
    VISIBILITY (Bob) Anything visible will be used by Phoenix. *This includes any geometry or fields. *To exclude geometry, check the sim node under “scene interaction”. *Even hiding the Phoenix source (oil drum) will keep your fluid from being emitted. *When rendering foam, the foam shader must be included in your render layer.
    *
    PHOENIX RESTART (Tom) Any sim that runs for over 5 hours will restart automatically, killing your previously cached frames.
    *
    MAYA FIELDS – tend to overwhelm and slow the Phoenix sim. *Try using the turbulence field listed in the Phoenix toolbar.
    *
    RENDER SETTINGS – Max VRAY subdivs other than one will make particles blotchy.
    *
    MAYA HANGS - …occasionally when *a scene contains references. *Use import instead.
    *
    MOTION BLUR – Particles rendered with the Phoenix foam shader have a “force on” toggle that gives a fast, but way overdone motion blur that you can’t seem to control.
    *
    *
    BUGS
    SUBSAMPLING WITH CACHED NPARTICLES (Bob) nParticles do not subsample after they’ve been cached. *For fast moving particles, it means that particles will jump from frame to frame with no subsampling interpolation. *Use regular particles instead if you need to cache.
    *
    MOVING CACHED PARTICLES …will not emit fluids from new position. *They continue to emit at the source.
    *
    ANIM INDEX (Bob) …under sim node will not work on the render farm. *Seems like a great way to animate a cache, but it might be getting hung up on frames that aren’t whole numbers.
    *
    PARTICLE GENERATION (Bob) Having Phoenix produce a particle system produces unwanted results if trying to use motion blur. *Particles are seen streaking from a central point as they are instanced to their respective positions.
    *
    MOTION BLUR (Bob) Latest version of Phoenix does not produce a correct motion blur for fire.
    *
    SPELLING (Bob) Someone please run a grammar & spellcheck on the software – at least in the GUI! *Nothing destroys confidence faster. *My fave is under the sim/grid section checkbox “No Bellow Initial Grid”. *You can also find the word “limeted” a few times under the grid option.
    *
    STATIC VOXELS (Bryce) If your fluid intersects the edge of your container on a jammed axis or an axis with no room to adapt, the fluid emitted will be static and remain there for the duration of the sim.
    *
    RENDER GIZMO …keeps particles from being rendered within the geo, even if the reverse checkbox is ticked. *It does not, however, hide particles like it’s supposed to. *Have not been able to get this to work at all.
    *
    CLIPPING PLANES – Phoenix doesn’t support camera clipping planes.
    *
    TEXTURE EMISSION - …will not work if geo has been added to the “sim ignore” set. *The geo will still emit particles, but it will ignore the geo’s texture UV’s. *We discovered this when we noticed particles getting trapped in the geo, which would occasionally form large blobs. *Removing the geo from the sim got rid of this, but presented other problems. *One workaround has been to modify the geos with “holes” to allow particles to escape.

  • #2
    ADAPTIVE GRID - yes, this is by design. If the side is jammed, it can't grow.

    TEMPERATURE - Not sure what you mean. The phx_candle example uses only temperature. The opacity graph by default is using smoke, so you must change that.

    PHOENIX HANGS - references should be fixed in the official version.

    VISIBILITY - this is again by design. Hiding the source probably should disable it, it really seems more intuitive.

    MAYA FIELDS - out of the box they are very slow. If the field is not using curves, the multithreading mode must be enabled. If the simulator do not change in size, the field can be calculated less often then the simulation. For example only once per frame or two.

    SUBSAMPLING WITH CACHED NPARTICLES - the problem is that Maya do not do this subsampling and currently we have some work-around in VRay which will be eventually ported. The only solution is to decrease "Evalauete every" of the cache. However this may completely change the particle simulation because of the way Maya works. The best option is not to use cache at all...

    ANIM INDEX - this should be fixed in the official release

    MOTION BLUR - this was changed a couple of times. If you have issues with the official build, a scene will be helpful.

    TEXTURE EMISSION - this is by design. If you ignore the geo from the simulator, it is treated as non-solid object. Such objects emit from the whole volume, not only their surface. They only support 3d textures.
    V-Ray/PhoenixFD for Maya developer

    Comment


    • #3
      MOVING CACHED PARTICLES - I think this is also fixed in the official build
      V-Ray/PhoenixFD for Maya developer

      Comment


      • #4
        Thanks for getting back to us Ivaylo! We appreciate any tips you can give us. Our post production schedule is extremely tight and we often don't have time to figure out workarounds for any issues we come across.

        ADAPTIVE GRID - Let me clarify. Say the y axis is jammed, but x and z are not. if you have smoke spreading across the floor, it should be allowed to spread. Instead, the fluid stays in the confines of the x and z box. To test this, try jamming the y side, and then opening the y side with a ground plane as a collision object.

        TEXTURE EMISSION - Not sure how this is by design. You're allowed to emit from a texture, and it would be beneficial if you could emit from the texture without having the geo as a collision object. That way you could emit from a textured fan waving back and forth without the fan disturbing the fluid.

        CACHED PARTICLES - just letting folks know that subsampling does seem to work for basic maya cached particles.

        Comment


        • #5
          As for the motion blur problem, simply take a sphere in a default fluid and animate the sphere very quickly across the grid (10 frames or so). Run the sim and render one of the middle frames with motion blur. You can see the sphere correctly blurred, but the fluid is not.

          Comment


          • #6
            Have you tried moblur in geometry mode ? This is I suppose would an example of the correct moblur. IMHO.
            I just can't seem to trust myself
            So what chance does that leave, for anyone else?
            ---------------------------------------------------------
            CG Artist

            Comment


            • #7
              The point is, we're trying to get fire to correctly motion blur. We've got fireballs all over this show and we can blur them in Nuke with Oflow, but it would be great if we could get actual motion blur - especially since Phoenix suggests this is possible.

              Comment


              • #8
                When using geometry mode - all the blur done entirely by vray.
                I just can't seem to trust myself
                So what chance does that leave, for anyone else?
                ---------------------------------------------------------
                CG Artist

                Comment


                • #9
                  How does that help with fire?

                  Comment


                  • #10
                    paul means just geometry mode, non solid. i doubt that there will be difference compared to the atmosphere mode.
                    what you mean "not working" can we see images?
                    ______________________________________________
                    VRScans developer

                    Comment


                    • #11
                      It's not really possible to make actually proper motion blur for fire. The problem is that you don't have object/vertices that deform, but a whole simulation that is going on. And of course, we can't do it, it will be very slow. In Phoenix we use simple and fast method, which purpose is to add some sense for motion, but you will not be able to match a fast moving geometry. You will not be able to match properly even the next frame. I'm fighting with the same problem when rendering Maya fluids. AFAIK, Mental Ray shows only one frame, which is at the end of the motion blur interval (probably to hide the geometry). What we did there was to add 2 other fake motion blur methods, which purpose is to be able to match moving geometry. But we must get more feedback if they are really better.

                      I think the older version was ignoring the length of the motion blur interval, and was doing a little more. Which was compensating this method in some way, because it seems to detect less movement. Here is the 3ds Max version

                      Click image for larger version

Name:	mblur0.png
Views:	1
Size:	38.3 KB
ID:	846430Click image for larger version

Name:	mblur1.png
Views:	1
Size:	63.5 KB
ID:	846431

                      I was even thinking to add manual option to extend the length of the fluid motion, but it's too fake, can't work in a stable way. The methods implemented recently for the Maya fluids are more stable. But we will see ...
                      V-Ray/PhoenixFD for Maya developer

                      Comment


                      • #12
                        About the emission from non-solid object, you mean the trapping of particles inside the geo that emits them? It should not do this. I'm not able to reproduce it. You can emit from texture, but it must be 3d texture. Here is an example:

                        phx_simple.zip
                        V-Ray/PhoenixFD for Maya developer

                        Comment


                        • #13
                          If you have jammed y axis, it will spread along x and z. I didn't understand what is the role of the ground plane. Something like this?

                          phx_adapt.zip
                          V-Ray/PhoenixFD for Maya developer

                          Comment


                          • #14
                            Unfortunately, like most studios in LA, we no longer have internet access other than an iphone or laptop, so getting images or scene files across can be a pain. Yea, I know it kinda crazy. But you can do the test I outlined above in less than 2 minutes.

                            Comment


                            • #15
                              Thanks for the response. I'm glad you guys were able to see exactly what I was talking about with the motion blur.

                              Have you considered a simple stepping option for renders? Since you have velocity info, you could subsample frames. It would be slow, but at least it would give you valid motion blur.

                              As for the clamped ground plane bug, here's a quick test. Set a sphere at the bottom of a simulation grid jammed in -y and set adaptive grid to smoke with a threshold at .1. Set the temperature emission to 0, discharge to 1, and set vorticity to 0. You will see cool smoke crawl across the ground, but stop at the edge of the grid, even though the grid is expanding. The fix for this is to open the -y on the grid and just use a collision plane for the ground.

                              Comment

                              Working...
                              X