[Soc-2017-dev] Weekly Report #07 - Mantaflow Particles and Fluid Guiding

[Nils Thuerey]

Hello Sebastian,

yes, looks like nice progress overall. Good you caugt the discrepancy between obstacle level set and flag grids for the open domain settings. I think that was important to fix for the practial usage of the solver in Blender.

Btw., for the bubbles: it would be good to additionally advect them with the flow velocity, so that they don't only rise straight up. Otherwise, the animated GIF is nice to show the behavior. I can imagine, that the initial velocity damping of bubbles would be a good UI parameter in the secondary particle panel (some scenes might require more so that the bubble particles dont overshoot).

And to quickly re-cap our discussion from Friday: it's an interesting question how to best design and implement an inflow feature for liquids. A UI widget to show and set the inflow direction is important, and we discussed using planes (as currently done for smoke). The problem there is that planes don't have a volume, but the volume is crucial for generating the FLIP particles of a liquid sim. This topic is not directly related to the secondary particle milestone, but it's something we should keep in mind.

Cheers,
-> Nils 


=== Technical University of Munich , http://ge.in.tum.de/ ===





> On 15. Jul 2017, at 02:26, Sebastián Barschkis <sebbas at sebbas.org> wrote:
> 
> Hello everyone,
> 
> here are a few notes on what happened this week:
> 
> https://wiki.blender.org/index.php/User:Sebbas/GSoC_2017/Reports/Week_7 <https://wiki.blender.org/index.php/User:Sebbas/GSoC_2017/Reports/Week_7>
> 
> = Week 7 =
> 
> This week I extended the secondary particle sampling and added support for float particles. I also took care of a bug and added inital velocity support for smoke.
> 
> === What's been done ===
> 
> * So far I hadn't really touched the "Initial velocity" settings from the flow source menu. So I started off with smoke and implemented a first version. Results look promising and similar to the one from the master branch. Initial velocity for liquids is slightly more complicated. (For liquids you'd have to set the initial velocity to a whole volume of particles, however, the smoke UI is more suited for planes with just one normal). I will come back to this later.
> 
> * There was (and still is to some extent) a problem with domains that have outflow cells. Liquid particles weren't deleted correctly and in some situations liquid simulations appeared to have too much velocity. With the help of Nils and many example scenes (thanks guys!) I managed to get the outflow problem under control. What's left are velocities. I experimented a bit and it seems there is too much velocity extrapolation before and after the pressure solve.
> 
> * I added float particle support to the seconday particle sampling routine. Tracers remain a todo, however, there is now also "Bubble" particles! For now, here is a simulation of those floats and bubbles in Mantaflow. (In Blender these new files are unstable, I am still hunting the bug).
> 
> * On the Blender side, everything is now set for float, tracer, bubble particles. That includes UI, automatic particle system setup and caching. I just have to tranfer the Mantaflow code and uncomment some lines.
> 
> === Next up ===
> 
> * Get my Mantaflow code working in Blender again.
> 
> * Add support for tracer particles.
> 
> * Take more care of the Github bug. Again, outflow should be fixed, velocities are still off. Fixing this bug has high priority. 
> 
> === Questions ===
> 
> * No
> 
> Best wishes,
> Sebastián
> _______________________________________________
> Soc-2017-dev mailing list
> Soc-2017-dev at blender.org
> https://lists.blender.org/mailman/listinfo/soc-2017-dev

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.blender.org/pipermail/soc-2017-dev/attachments/20170716/33e2e422/attachment.htm