[Soc-2017-dev] Weekly Report #10 - Mantaflow Particles and Fluid Guiding

[Nils Thuerey]

Hello Sebastian,

nice! Cool to see a first version of the guiding working in blender. I think the possibility to rasterize curves and use them to define velocities would be neat. That would also be very useful to add directed forces to sims without guiding.

Apart from that I'm still looking into the inflow velocity problems. I don't think I've found all of the problems, but here are already a few insights from that setup:

- I think it'll be important to always set the velocity of inflow regions, e.g., by extending the level set by 1, and writing the velocities. in the current setup, the unconstrained vels do weird things over time...

- in "addForceField" , shouldnt this be scaled by solver.timestep? you could add a scaling factor param, and then set the factor in the python script

- just wondering, does the obstacle level set created on the blender side have outside values? that's needed for pushOutofObs plugin , and the box in the example didn't have a proper outside (but that was the mantaflow box SDF)

As mentioned by Gottfried, the particle system UI is probably a good starting point for the inflow & particle source velocities.

Cheers,
-> Nils


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




> On 4. Aug 2017, at 14:25, Sebastián Barschkis <sebbas at sebbas.org> wrote:
> 
> Hello everyone,
> 
> here is my progress report for week 10. Enjoy and have a nice weekend!
> 
> https://wiki.blender.org/index.php/User:Sebbas/GSoC_2017/Reports/Week_10 <https://wiki.blender.org/index.php/User:Sebbas/GSoC_2017/Reports/Week_10>
> 
> = Week 10 =
> 
> This week I was busy with exams. However, as promised, I created something "visually pleasing" with fluid guiding :)
> 
> === What's been done ===
> 
> * Mantaflow has a built in test function that sets up a 2D velocity grid describing a spiral flow. My idea for a first fluid guiding test was to make use of this function and create a 3D smoke tornado (Gooseberry! :)
> 
> * Therefore I first had to adapt the spiral velocity function. I needed a 3D instead of a 2D target velocity grid for Mantaflow to work in Blender.
> 
> * I then fitted all the necessary Python fluid guiding functions into the smoke script in Blender. The most important part here is to use the dedicated fluid guiding pressure function (it takes care of guiding and also the pressure solve). With this setup in place, the basic (tornado) guiding was working.
> 
> * To control the guiding just a bit I added a "Fluid guiding" tab to the UI. It lets you toggle guiding and also set a strength value. Note that this UI setup is meant to be just a temporary solution though.
> 
> === Next up ===
> 
> * Next week I want to get back on track. Before fully focusing on guiding though I need to take a look at particles again. There are still some bugs when multiple particle types are enabled.
> 
> * The final fluid guiding setup should allow users to define arbitrary velocity fields (and not just tornados :). So a possible solution, which I'll try out, might be to use curves along which a velocity can flow.   
> 
> === 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/20170804/83ce90a0/attachment.htm