Defining obstacles via a numpy array

[Luca Massidda]

Many thanks to all the creators and developers of the project. Really impressive.

My question as a beginner concerns the setting of obstacles in the grid.

Is it possible to define obstacles "by filling the cells" using the python script? I would like to define the obtacles using a numpy array. 

I suppose it can be done, but I did not find an example of this in the documentation.

Thanks,

Luca

[Nils Thuerey]

Hi Luca, yeah - there's no example for that at the moment, but you can initialize any grid as numpy first, and then copy it into a mantaflow grid (e.g., as described here https://ge.in.tum.de/research/mantaflow-tensorflow-tutorial-1/).

Just make sure to use the right integer values as defined in grid.h:

  

    enum CellType {

        TypeNone     = 0,

        TypeFluid    = 1,

        TypeObstacle = 2,

        TypeEmpty    = 4,

        TypeInflow   = 8,

        TypeOutflow  = 16,

...

    };

Cheers, Nils

[Luca Massidda]

ok!

thanks a lot,

Luca

[Jorge]

I'm sorry, I dont want to ask dumb questions (I have been trying to do this by myself), but can you please explicitely show me how to do this? how to initialize a FlagGrid from a numpy array. There is no copyArrayToGridFlag(source=npFlags, target=flags) function.

Again, many thanks in advance

[Jorge]

For anyone having trouble doing this.

In the file "numpyconvert.cpp"  in source/plugin add:

PYTHON() void copyArrayToFlagGrid(const PyArrayContainer source, FlagGrid& target) {

copyArrayToGridScalar<FlagGrid>(source, target);

}

Then build manta.

This allowed me to initialize the flag grid with a numpy array.

[Nils Thuerey]

Hello Jorge, good to hear you solved the problem - and thanks for posting a fix. This looks like a very good addition, we'll include it in the main branch...

Cheers, Nils