A contour line within a quad is, by default, a straight line between points on two of its edges. The quad_as_tri option divides each quad into four triangles using a virtual point at the centre (mean x, y of the corner points) and extra contour points are inserted where the contour level intersects the diagonals so that contour lines are piecewise linear within the triangles.
The z value of the central point is calculated depending on the z_interp setting; for the default z_interp=ZInterp.Linear this is the mean of the z values of the corner points. Corner masked triangles are not affected by this setting, only full unmasked quads.
quad_as_tri is available for the serial and threaded algorithms. It is always disabled by default, so if required it must be explicitly requested:
>>> cont_gen = contour_generator(quad_as_tri=True, ...)
Note
quad_as_tri produces more detailed contours, but not necessarily smoother ones.
Here is an example of the difference between quad_as_tri=False and True for z values on a coarse grid that are given by z = np.sqrt(x**2 + y**2). On a finer grid the contours would be semicircular.
from contourpy import contour_generator from contourpy.util.mpl_renderer import MplRenderer as Renderer import numpy as np
x, y = np.meshgrid([-2, -1, 0, 1, 2], [0, 1, 2]) z = np.sqrt(x*2 + y*2)
renderer = Renderer(ncols=2, figsize=(8, 2.4)) for ax in range(2): quad_as_tri = bool(ax) renderer.grid(x, y, ax=ax, quad_as_tri_alpha=ax*0.1) renderer.title(f"quad_as_tri={quad_as_tri}", ax=ax)
cont_gen = contour_generator(x, y, z, quad_as_tri=quad_as_tri) for i, level in enumerate(np.arange(0.5, 2.51, 0.5)): lines = cont_gen.lines(level) renderer.lines(lines, cont_gen.line_type, ax=ax, color=f"C{i}", linewidth=2)
renderer.z_values(x, y, z, ax=ax, quad_as_tri=quad_as_tri, fmt="0.2f")
renderer.show()
Another situation in which quad_as_tri may be useful is shown below. The quad has three corners at the same z level so without quad_as_tri contour lines cut across diagonally.
from contourpy import contour_generator from contourpy.util.mpl_renderer import MplRenderer as Renderer import numpy as np
x = y = [0, 1] z = [[0, 0], [0, 4]]
renderer = Renderer(ncols=2, figsize=(6, 3)) for ax in range(2): quad_as_tri = bool(ax) renderer.grid(x, y, ax=ax, quad_as_tri_alpha=ax*0.1) renderer.title(f"quad_as_tri={quad_as_tri}", ax=ax)
cont_gen = contour_generator(x, y, z, quad_as_tri=quad_as_tri) for i, level in enumerate(np.arange(0.0, 4.0, 0.4)): lines = cont_gen.lines(level) renderer.lines(lines, cont_gen.line_type, ax=ax, color=f"C{i}", linewidth=2)
renderer.z_values(x, y, z, ax=ax, quad_as_tri=quad_as_tri)
renderer.show()
Note
quad_as_tri=True produces contour lines and filled contours typically containing about three times as many points as quad_as_tri=False.