[Bug]: PolyCollection shows white space between polygons

[veenstrajelmer]

Bug summary

When plotting unstructured meshes with matplotlib PolyCollection, we would like to avoid the edges/lines to be vislible. This can be done with edgecolor="face" but for fine meshes (many small polygons) the width of the edges can become significant and the image is incorrect. This is documented in Deltares/xugrid#203. A better solution would be to use edgecolor="none" or linewidths=0. However, in this case there is still white space showing up between the polygons, even though they should align perfectly.

Code for reproduction

import numpy as np
import matplotlib.pyplot as plt
plt.close("all")
from matplotlib.collections import PolyCollection

vertices = np.array(
    [[[ 3.925     , 52.0125    ],
        [ 3.925     , 52.01458333],
        [ 3.9234375 , 52.0125    ],
        [ 3.925     , 52.0125    ],
        [ 3.925     , 52.0125    ]],

       [[ 3.925     , 52.01041667],
        [ 3.925     , 52.01145833],
        [ 3.9234375 , 52.01145833],
        [ 3.9234375 , 52.01041667],
        [ 3.925     , 52.01041667]],

       [[ 3.925     , 52.0125    ],
        [ 3.9234375 , 52.0125    ],
        [ 3.9234375 , 52.01145833],
        [ 3.925     , 52.01145833],
        [ 3.925     , 52.0125    ]],

       [[ 3.921875  , 52.0125    ],
        [ 3.9234375 , 52.0125    ],
        [ 3.921875  , 52.01458333],
        [ 3.921875  , 52.0125    ],
        [ 3.921875  , 52.0125    ]],

       [[ 3.925     , 52.01041667],
        [ 3.9234375 , 52.01041667],
        [ 3.9234375 , 52.009375  ],
        [ 3.925     , 52.009375  ],
        [ 3.925     , 52.01041667]],

       [[ 3.925     , 52.00833333],
        [ 3.925     , 52.009375  ],
        [ 3.9234375 , 52.009375  ],
        [ 3.9234375 , 52.00833333],
        [ 3.925     , 52.00833333]],

       [[ 3.925     , 52.01458333],
        [ 3.921875  , 52.01458333],
        [ 3.9234375 , 52.0125    ],
        [ 3.925     , 52.01458333],
        [ 3.925     , 52.01458333]],

       [[ 3.921875  , 52.01041667],
        [ 3.9234375 , 52.01041667],
        [ 3.9234375 , 52.01145833],
        [ 3.921875  , 52.01145833],
        [ 3.921875  , 52.01041667]],

       [[ 3.921875  , 52.0125    ],
        [ 3.921875  , 52.01145833],
        [ 3.9234375 , 52.01145833],
        [ 3.9234375 , 52.0125    ],
        [ 3.921875  , 52.0125    ]],

       [[ 3.921875  , 52.0125    ],
        [ 3.921875  , 52.01458333],
        [ 3.9203125 , 52.0125    ],
        [ 3.921875  , 52.0125    ],
        [ 3.921875  , 52.0125    ]],

       [[ 3.921875  , 52.01041667],
        [ 3.921875  , 52.009375  ],
        [ 3.9234375 , 52.009375  ],
        [ 3.9234375 , 52.01041667],
        [ 3.921875  , 52.01041667]],

       [[ 3.921875  , 52.00833333],
        [ 3.9234375 , 52.00833333],
        [ 3.9234375 , 52.009375  ],
        [ 3.921875  , 52.009375  ],
        [ 3.921875  , 52.00833333]],

       [[ 3.925     , 52.01666667],
        [ 3.921875  , 52.01666667],
        [ 3.921875  , 52.01458333],
        [ 3.925     , 52.01458333],
        [ 3.925     , 52.01666667]],

       [[ 3.921875  , 52.01041667],
        [ 3.921875  , 52.01145833],
        [ 3.9203125 , 52.01145833],
        [ 3.9203125 , 52.01041667],
        [ 3.921875  , 52.01041667]],

       [[ 3.921875  , 52.0125    ],
        [ 3.9203125 , 52.0125    ],
        [ 3.9203125 , 52.01145833],
        [ 3.921875  , 52.01145833],
        [ 3.921875  , 52.0125    ]],

       [[ 3.91875   , 52.01458333],
        [ 3.9203125 , 52.0125    ],
        [ 3.921875  , 52.01458333],
        [ 3.91875   , 52.01458333],
        [ 3.91875   , 52.01458333]],

       [[ 3.921875  , 52.01041667],
        [ 3.9203125 , 52.01041667],
        [ 3.9203125 , 52.009375  ],
        [ 3.921875  , 52.009375  ],
        [ 3.921875  , 52.01041667]],

       [[ 3.921875  , 52.00833333],
        [ 3.921875  , 52.009375  ],
        [ 3.9203125 , 52.009375  ],
        [ 3.9203125 , 52.00833333],
        [ 3.921875  , 52.00833333]],

       [[ 3.91875   , 52.01666667],
        [ 3.91875   , 52.01458333],
        [ 3.921875  , 52.01458333],
        [ 3.921875  , 52.01666667],
        [ 3.91875   , 52.01666667]],

       [[ 3.91875   , 52.01041667],
        [ 3.9203125 , 52.01041667],
        [ 3.9203125 , 52.01145833],
        [ 3.91875   , 52.01145833],
        [ 3.91875   , 52.01041667]],

       [[ 3.91875   , 52.0125    ],
        [ 3.91875   , 52.01145833],
        [ 3.9203125 , 52.01145833],
        [ 3.9203125 , 52.0125    ],
        [ 3.91875   , 52.0125    ]],

       [[ 3.91875   , 52.0125    ],
        [ 3.9203125 , 52.0125    ],
        [ 3.91875   , 52.01458333],
        [ 3.91875   , 52.0125    ],
        [ 3.91875   , 52.0125    ]],

       [[ 3.91875   , 52.01041667],
        [ 3.91875   , 52.009375  ],
        [ 3.9203125 , 52.009375  ],
        [ 3.9203125 , 52.01041667],
        [ 3.91875   , 52.01041667]],

       [[ 3.91875   , 52.00833333],
        [ 3.9203125 , 52.00833333],
        [ 3.9203125 , 52.009375  ],
        [ 3.91875   , 52.009375  ],
        [ 3.91875   , 52.00833333]],

       [[ 3.91875   , 52.01666667],
        [ 3.915625  , 52.01666667],
        [ 3.915625  , 52.01458333],
        [ 3.91875   , 52.01458333],
        [ 3.91875   , 52.01666667]],

       [[ 3.91875   , 52.01041667],
        [ 3.91875   , 52.01145833],
        [ 3.915625  , 52.01041667],
        [ 3.91875   , 52.01041667],
        [ 3.91875   , 52.01041667]],

       [[ 3.91875   , 52.0125    ],
        [ 3.915625  , 52.0125    ],
        [ 3.91875   , 52.01145833],
        [ 3.91875   , 52.0125    ],
        [ 3.91875   , 52.0125    ]],

       [[ 3.91875   , 52.0125    ],
        [ 3.91875   , 52.01458333],
        [ 3.915625  , 52.01458333],
        [ 3.915625  , 52.0125    ],
        [ 3.91875   , 52.0125    ]],

       [[ 3.91875   , 52.01041667],
        [ 3.915625  , 52.01041667],
        [ 3.91875   , 52.009375  ],
        [ 3.91875   , 52.01041667],
        [ 3.91875   , 52.01041667]],

       [[ 3.91875   , 52.00833333],
        [ 3.91875   , 52.009375  ],
        [ 3.915625  , 52.00833333],
        [ 3.91875   , 52.00833333],
        [ 3.91875   , 52.00833333]],

       [[ 3.9125    , 52.01666667],
        [ 3.9125    , 52.01458333],
        [ 3.915625  , 52.01458333],
        [ 3.915625  , 52.01666667],
        [ 3.9125    , 52.01666667]],

       [[ 3.915625  , 52.0125    ],
        [ 3.915625  , 52.01041667],
        [ 3.91875   , 52.01145833],
        [ 3.915625  , 52.0125    ],
        [ 3.915625  , 52.0125    ]],

       [[ 3.9125    , 52.0125    ],
        [ 3.915625  , 52.0125    ],
        [ 3.915625  , 52.01458333],
        [ 3.9125    , 52.01458333],
        [ 3.9125    , 52.0125    ]],

       [[ 3.915625  , 52.00833333],
        [ 3.91875   , 52.009375  ],
        [ 3.915625  , 52.01041667],
        [ 3.915625  , 52.00833333],
        [ 3.915625  , 52.00833333]],

       [[ 3.9125    , 52.0125    ],
        [ 3.9125    , 52.01041667],
        [ 3.915625  , 52.01041667],
        [ 3.915625  , 52.0125    ],
        [ 3.9125    , 52.0125    ]],

       [[ 3.9125    , 52.00833333],
        [ 3.915625  , 52.00833333],
        [ 3.915625  , 52.01041667],
        [ 3.9125    , 52.01041667],
        [ 3.9125    , 52.00833333]]])

kwargs = dict(edgecolors="none")

fig, ax = plt.subplots()
collection = PolyCollection(vertices, **kwargs)
collection.set_array(np.ones(len(vertices)))
primitive = ax.add_collection(collection)

xmin = vertices[:, :, 0].min()
ymin = vertices[:, :, 1].min()
xmax = vertices[:, :, 0].max()
ymax = vertices[:, :, 1].max()
ax.set_xlim(xmin, xmax)
ax.set_ylim(ymin, ymax)

Actual outcome

Expected outcome

This second figure is created with edgecolors="face". However, keep in mind that the original xugrid issue linked in the summary also shows that in this specific application this is not desireable. I would expect that this figure without lines would also be the result of edgecolors="none" since all polygons are exactly next to each other.

Additional information

No response

Operating system

Windows 10

Matplotlib Version

3.8.2

Matplotlib Backend

Qt5Agg

Python version

3.11.6

Jupyter version

No response

Installation

pip

[timhoffm]

Thanks for the report. This is a fundamental rendering limitation. See #9574 (comment) for a good explanation. Unfortunately there's nothing we can realistically do about it.

Variants of this have been discussed multiple times, e.g. #10115, #1188, #5694.

[veenstrajelmer]

Thanks for the quick reply and the links, good to know. I searched for open issues but not for closed ones, although I am not sure if I would have found all of this. Either way, we will see how to work around this in that case.