[Sidefx-houdini-list] nonLinear fluid reTime

Szymon Kapeniak szymon.kapeniak at gmail.com
Fri Jul 8 06:06:15 EDT 2011


2011/7/8 Olex P <hoknamahn at gmail.com>:
> I don't think that blending of two volumes in substep makes any sense
> especially if a difference in state of the fluid in two frames is high
> enough.

Well, people are using this technique to retime their smoke sims,
Olex. I understand why blending volumes works worse than particles,
but still it seems to be enough for most retiming cases, since frames
tend to be similar.

Agree though, particles from volume might work better for high displacements.

 But we can jump from Eulerian coordinate system (grids with
> static voxels and quantities going through this grid) to Lagrangian
> (particles where every particle is a local coordinate system with some
> quantity, let's say density, on it).
> In other words we can advect particles using velocity field,
> interpolate particle positions in subframes and put density and other
> needed attributes back into a grid.
>
> On Fri, Jul 8, 2011 at 10:33 AM, Szymon Kapeniak
> <szymon.kapeniak at gmail.com> wrote:
>> Hey Ben,
>> fluids are volumes (SDF), so they could be interpolated by Timeblend,
>> Sequenceblend and Timeshift Sops, similarly to plain geometries. I've
>> never tried it on fluids, but theoretically this should work the same
>> as on smoke (unless interpolating SDF is more computational
>> problematic, than density fields).
>>
>> So I would try to increase a little su- stepping to provide good
>> samples for extreme interpolation, and retime my SDFs (taken straight
>> from DOPS, without meshing), after which meshing should be possible as
>> usual.
>>
>>
>> hth,
>> skk.
>>
>> 2011/7/8 Ben Frost <benfrost3d at gmail.com>:
>>> hi all..
>>>
>>> I'm looking for anyones opinion regarding a non-linear reTime of a fluid
>>> sim.. and possible approaches to take.. the reTime is quite severe.. going
>>> from 24fps to around 432fps.. so 18 times slower.. and is based on a
>>> particle sim..
>>>
>>> at the moment I'm using a brute force method for having the subSteps at 18,
>>> writing out the subFrame for each step.. and then read it back in, and
>>> reTiming.. this works.. but as you can imagine the sim times goes through
>>> the roof a bit.. and change the sim some what.. although it's more accurate
>>> it does create the problem of previewing the simulation at 18 subSteps..
>>> ideally I'd like to simulate at as fewer subSteps as possible.. until i'm
>>> happy with the sim.. and then upRess and bake out subFrames/subSteps..
>>>
>>> we've reTimed the particles using the velocities to interpolate and get us
>>> the subFrames needed.. not sure there is a way to interpret this to fluid
>>> though..
>>>
>>> and to make things worse.. I've been shown an animatable "sim scale" option
>>> in maya.. which works all too well.. I noticed houdini has the same option
>>> of sim scale, but this is only able to be set at sim start.. just wondering
>>> if this is possible to be manipulated and animate..?
>>>
>>> anyway I'd be great to hear your opinions or possible more efficient less
>>> brute force approach..
>>>
>>> cheers
>>>
>>> ben..
>>> _______________________________________________
>>> Sidefx-houdini-list mailing list
>>> Sidefx-houdini-list at sidefx.com
>>> https://lists.sidefx.com:443/mailman/listinfo/sidefx-houdini-list
>>>
>>
>>
>>
>> --
>> skk.
>> _______________________________________________
>> Sidefx-houdini-list mailing list
>> Sidefx-houdini-list at sidefx.com
>> https://lists.sidefx.com:443/mailman/listinfo/sidefx-houdini-list
>>
> _______________________________________________
> Sidefx-houdini-list mailing list
> Sidefx-houdini-list at sidefx.com
> https://lists.sidefx.com:443/mailman/listinfo/sidefx-houdini-list
>



-- 
skk.



More information about the Sidefx-houdini-list mailing list