Become a mesh!

Besides all its power to generate high quality mesh, Gmsh also can be used to generate high quality mesh of yours!

Mesh from picture

To generate a mesh like this one, all you will need is python and gmsh. In its webpage you will find all instructions to install it. The following .py file is the script to perform all tasks in gmsh. You can download this file here.

# pic2msh.py

#  ====================================
#  Credits to Prof. Christophe Geuzaine
#  ==================================== 

import sys
import subprocess

# Picture to be meshed
filename = str(sys.argv[1])
filename = f'"{filename}"'

# Write GEO file as input for Gmsh
header = f'If(PostProcessing.NbViews == 0) \n \
\t Merge {filename}; \n\
\t Plugin(ModifyComponents).Expression0 = "1 + v0^3 * 10"; \n\
\t Plugin(ModifyComponents).Run; \n\
EndIf\n'

body = """
Background Mesh View[0];
w = View[0].MaxX;
h = View[0].MaxY;

Point(1)={0,0,0,w};
Point(2)={w,0,0,w};
Point(3)={w,h,0,w};
Point(4)={0,h,0,w};
Line(1) = {1,2};
Line(2) = {2,3};
Line(3) = {3,4};
Line(4) = {4,1};
Line Loop(5) = {3,4,1,2};
Plane Surface(6) = {5};

DefineConstant[
  algo = {Mesh.Algorithm, AutoCheck 0, GmshOption "Mesh.Algorithm",
    Choices{1="MeshAdapt", 2="Automatic", 5="Delaunay", 6="Frontal-Delaunay", 
    7="BAMG", 8="Frontal-Delaunay for Quads", 9="Packing of Parallelograms",
    11="Quasi-structured Quad"},
    Name "Meshing parameters/Algorithm"},

  sizeFact = {Mesh.CharacteristicLengthFactor, AutoCheck 0,
    GmshOption "Mesh.CharacteristicLengthFactor", Min 0.1, Max 10, Step 0.1,
    Name "Meshing parameters/Element size factor"},

  sizeMin = {Mesh.CharacteristicLengthMin, AutoCheck 0,
    GmshOption "Mesh.CharacteristicLengthMin", Min w/100, Max w, Step 0.1,
    Name "Meshing parameters/Minimum element size"},

  save = {StrCat("View.ShowScale=0;Print '", CurrentDirectory, "outputFile.png';"),
    AutoCheck 0, Macro "GmshParseString",
    Name "Save PNG"}
];

Mesh.ColorCarousel = 0;
Mesh.Color.Triangles = Black;
Mesh.Color.Quadrangles = Black;
Mesh.RecombineAll = (algo == 8);
Solver.AutoMesh = 2;"""

geoFile = header + body

def write_input():
    f = open('input.geo', "w")
    f.write(geoFile)
    f.close()

def run_mesh_gui():
    subprocess.run(["gmsh", "input.geo"])
    subprocess.run(["gmsh", "input.msh"])

# Run
write_input()
run_mesh_gui()

To obtain an expected results, it is recommended to use a picture with a ‘plain background’. To remove the background from any picture, I recommend this website. In addition to removing the background, it is important to set a plain color background, preferably white. After tweaking the input picture, open a terminal and run the following command:

python pic2mesh.py path_to_picture

After running the command, you will see the Gmsh GUI as below (not in dark mode, if it is have not previously changed):

Mesh from picture

Firstly, press Alt + P and Alt + L to hide geometry’s points and curves. In the left side bar menu, you can select the meshing options Algorithm, Element size factor, and Minimum element size. Choose these parameters according to your preference (or left them as default values) and press Run on the bottom to generate the mesh. To outline the mesh elements, press Alt + E. To save, press Save PNG. A file named outputFile.png will be saved same in the same directory.

Mesh from picture

Besides the saved picture, a mesh file named pictureToMesh.msh will also be saved. This mesh file is generated from your picture’s discretization based on pixel variations. When you close the Gmsh GUI that is currently open, a new GUI will appear, displaying only the generated mesh:

Mesh from picture

Now you can have a better visualization of the mesh. If it presents the expected result, you can export it in a vector graphics format, such as SVG or PDF. Using the top menu, go to File > Export and select you preferred file format. For tweaking the vector image, I do recommend using Inkscape, for instance. From now on, your creativity is what matters most.

Good meshing!