import os
import subprocess
import numpy as np
import scipy.ndimage as nd
import skimage.measure as ms
from edt import edt
from skimage.morphology import ball
from porespy.filters import reduce_peaks
from porespy.networks import generate_voxel_image
from porespy.tools import sanitize_filename
[docs]
def dict_to_vtk(data, filename, voxel_size=1, origin=(0, 0, 0)):
r"""
Accepts multiple images as a dictionary and compiles them into a vtk file
Parameters
----------
data : dict
A dictionary of *key: value* pairs, where the *key* is the name of the
scalar property stored in each voxel of the array stored in the
corresponding *value*.
path : string
Path to output file
voxel_size : int
The side length of the voxels (voxels are cubic)
origin : float
data origin (according to selected voxel size)
Notes
-----
Outputs a vtk, vtp or vti file that can opened in ParaView
Examples
--------
`Click here
<https://porespy.org/examples/io/reference/dict_to_vtk.html>`_
to view online example.
"""
try:
from pyevtk.hl import imageToVTK
except ModuleNotFoundError:
msg = 'The pyevtk package can be installed with pip install pyevtk'
raise ModuleNotFoundError(msg)
vs = voxel_size
for entry in data:
if data[entry].dtype == bool:
data[entry] = data[entry].astype(np.int8)
if data[entry].flags["C_CONTIGUOUS"]:
data[entry] = np.ascontiguousarray(data[entry])
imageToVTK(filename, cellData=data, spacing=(vs, vs, vs), origin=origin)
[docs]
def to_vtk(im, filename, divide=False, downsample=False, voxel_size=1, vox=False):
r"""
Converts an array to a vtk file.
Parameters
----------
im : 3D image
The image of the porous material
path : string
Path to output file
divide : bool
vtk files can get very large, this option allows you for two output
files, divided at z = half. This allows for large data sets to be
imaged without loss of information
downsample : bool
very large images acan be downsampled to half the size in each
dimension, this doubles the effective voxel size
voxel_size : int
The side length of the voxels (voxels are cubic)
vox : bool
For an image that is binary (1's and 0's) this reduces the file size by
using int8 format (can also be used to reduce file size when accuracy
is not necessary ie: just visulization)
Notes
-----
Outputs a vtk, vtp or vti file that can opened in paraview
Examples
--------
`Click here
<https://porespy.org/examples/io/reference/to_vtk.html>`_
to view online example.
"""
try:
from pyevtk.hl import imageToVTK
except ModuleNotFoundError:
msg = 'The pyevtk package can be installed with pip install pyevtk'
raise ModuleNotFoundError(msg)
if len(im.shape) == 2:
im = im[:, :, np.newaxis]
if im.dtype == bool:
vox = True
if vox:
im = im.astype(np.int8)
vs = voxel_size
if divide:
split = np.round(im.shape[2] / 2).astype(np.int)
im1 = im[:, :, 0:split]
im2 = im[:, :, split:]
imageToVTK(f"{filename}_1",
cellData={"im": np.ascontiguousarray(im1)},
spacing=(vs, vs, vs),)
imageToVTK(f"{filename}_2",
origin=(0.0, 0.0, split * vs),
cellData={"im": np.ascontiguousarray(im2)},
spacing=(vs, vs, vs),)
elif downsample:
im = nd.interpolation.zoom(im, zoom=0.5, order=0, mode="reflect")
imageToVTK(filename,
cellData={"im": np.ascontiguousarray(im)},
spacing=(2 * vs, 2 * vs, 2 * vs),)
else:
imageToVTK(filename,
cellData={"im": np.ascontiguousarray(im)},
spacing=(vs, vs, vs))
[docs]
def to_palabos(im, filename, solid=0):
r"""
Converts an ndarray image to a text file that Palabos can read in as a
geometry for Lattice Boltzmann simulations. Uses a Euclidean distance
transform to identify solid voxels neighboring fluid voxels and labels
them as the interface.
Parameters
----------
im : ndarray
The image of the porous material
filename : string
Path to output file
solid : int
The value of the solid voxels in the image used to convert image to
binary with all other voxels assumed to be fluid.
Notes
-----
File produced contains 3 values: 2 = Solid, 1 = Interface, 0 = Pore
Palabos will run the simulation applying the specified pressure drop from
x = 0 to x = -1.
Examples
--------
`Click here
<https://porespy.org/examples/io/reference/to_palabos.html>`_
to view online example.
"""
# Create binary image for fluid and solid phases
bin_im = im == solid
# Transform to integer for distance transform
bin_im = bin_im.astype(int)
# Distance Transform computes Euclidean distance in lattice units to
# Nearest fluid for every solid voxel
dt = nd.distance_transform_edt(bin_im)
dt[dt > np.sqrt(2)] = 2
dt[(dt > 0) * (dt <= np.sqrt(2))] = 1
dt = dt.astype(int)
# Write out data
with open(filename, "w") as f:
out_data = dt.flatten().tolist()
f.write("\n".join(map(repr, out_data)))
[docs]
def openpnm_to_im(
network,
pore_shape="sphere",
throat_shape="cylinder",
max_dim=None,
rtol=0.1,
):
r"""
Generates voxel image from an OpenPNM network object.
Parameters
----------
network : OpenPNM GenericNetwork
Network from which voxel image is to be generated
pore_shape : str
Shape of pores in the network, valid choices are "sphere", "cube"
throat_shape : str
Shape of throats in the network, valid choices are "cylinder", "cuboid"
max_dim : int
Number of voxels in the largest dimension of the network
rtol : float
Stopping criteria for finding the smallest voxel image such that
further increasing the number of voxels in each dimension by 25% would
improve the predicted porosity of the image by less that ``rtol``
Returns
-------
im : ndarray
Voxelated image corresponding to the given pore network model
Notes
-----
(1) The generated voxelated image is labeled with 0s, 1s and 2s signifying
solid phase, pores, and throats respectively.
(2) If max_dim is not provided, the method calculates it such that the
further increasing it doesn't change porosity by much.
"""
return generate_voxel_image(
network,
pore_shape=pore_shape,
throat_shape=throat_shape,
max_dim=max_dim,
rtol=rtol,
)
[docs]
def to_stl(im, filename, divide=False, downsample=False, voxel_size=1, vox=False):
r"""
Converts an array to an STL file.
Parameters
----------
im : 3D image
The image of the porous material
path : string
Path to output file
divide : bool
vtk files can get very large, this option allows you for two output
files, divided at z = half. This allows for large data sets to be
imaged without loss of information
downsample : bool
very large images acan be downsampled to half the size in each
dimension, this doubles the effective voxel size
voxel_size : int
The side length of the voxels (voxels are cubic)
vox : bool
For an image that is binary (1's and 0's) this reduces the file size by
using int8 format (can also be used to reduce file size when accuracy
is not necessary ie: just visulization)
Notes
-----
Outputs an STL file that can opened in Paraview
Examples
--------
`Click here
<https://porespy.org/examples/io/reference/to_stl.html>`_
to view online example.
"""
filename = sanitize_filename(filename, ext="stl", exclude_ext=True)
if len(im.shape) == 2:
im = im[:, :, np.newaxis]
if im.dtype == bool:
vox = True
if vox:
im = im.astype(np.int8)
vs = voxel_size
if divide:
split = np.round(im.shape[2] / 2).astype(np.int)
im1 = im[:, :, 0:split]
im2 = im[:, :, split:]
_save_stl(im1, vs, f"{filename}_1")
_save_stl(im2, vs, f"{filename}_2")
elif downsample:
im = nd.interpolation.zoom(im, zoom=0.5, order=0, mode="reflect")
_save_stl(im, vs * 2, filename)
else:
_save_stl(im, vs, filename)
def _save_stl(im, vs, filename):
r"""
Helper method to convert an array to an STL file.
Parameters
----------
im : 3D image
The image of the porous material
voxel_size : int
The side length of the voxels (voxels are cubic)
filename : string
Path to output file
"""
try:
from stl import mesh
except ModuleNotFoundError:
msg = 'numpy-stl can be installed with pip install numpy-stl'
raise ModuleNotFoundError(msg)
im = np.pad(im, pad_width=10, mode="constant", constant_values=True)
vertices, faces, norms, values = ms.marching_cubes(im)
vertices *= vs
# Export the STL file
export = mesh.Mesh(np.zeros(faces.shape[0], dtype=mesh.Mesh.dtype))
for i, f in enumerate(faces):
for j in range(3):
export.vectors[i][j] = vertices[f[j], :]
export.save(f"{filename}.stl")
[docs]
def to_paraview(im, filename, phase=2):
r"""
Converts an array to a paraview state file.
Parameters
----------
im : ndarray
The image of the porous material.
filename : str
Path to output file.
phase : str
The desired phase of output image where phase = 0 represent the
pore phase, phase = 1 represents the solid phase, and phase= 2 is
the whole domain. The default value is 2.
Notes
-----
Outputs an pvsm file that can opened in Paraview.
Examples
--------
`Click here
<https://porespy.org/examples/io/reference/to_paraview.html>`_
to view online example.
"""
try:
import imageio
except ModuleNotFoundError:
msg = "The imageio package can be installed with pip install imageio"
raise ModuleNotFoundError(msg)
try:
import paraview.simple
except ModuleNotFoundError:
msg = ("The paraview python bindings must be installed using conda"
" install -c conda-forge paraview, however this may require"
" using a virtualenv since conflicts with other packages are"
" common. This is why it is not explicitly included as a"
" dependency in porespy.")
raise ModuleNotFoundError(msg)
data = im.astype("uint8")
file = os.path.splitext(filename)[0]
path = file + ".tiff"
if len(im.shape) == 2:
imageio.imwrite(path, np.array(data))
view = "Slice"
zshape = 0
xshape = im.shape[1]
yshape = im.shape[0]
elif len(im.shape) == 3:
imageio.volsave(path, np.array(data))
view = "Volume"
zshape = im.shape[0]
xshape = im.shape[2]
yshape = im.shape[1]
maxshape = max(xshape, yshape)
paraview.simple._DisableFirstRenderCameraReset()
# Create a new 'TIFF Series Reader'
dtiff = paraview.simple.TIFFSeriesReader(FileNames=[path])
# Get active view
renderView1 = paraview.simple.GetActiveViewOrCreate("RenderView")
# Uncomment following to set a specific view size
# renderView1.ViewSize = [1612, 552]
# Get layout
_ = paraview.simple.GetLayout()
# Show data in view
dtiffDisplay = paraview.simple.Show(dtiff, renderView1, "UniformGridRepresentation")
# Get color transfer function/color map for 'TiffScalars'
tiffScalarsLUT = paraview.simple.GetColorTransferFunction("TiffScalars")
# Get opacity transfer function/opacity map for 'TiffScalars'
tiffScalarsPWF = paraview.simple.GetOpacityTransferFunction("TiffScalars")
# Trace defaults for the display properties.
dtiffDisplay.Representation = view
dtiffDisplay.ColorArrayName = ["POINTS", "Tiff Scalars"]
dtiffDisplay.LookupTable = tiffScalarsLUT
dtiffDisplay.OSPRayScaleArray = "Tiff Scalars"
dtiffDisplay.OSPRayScaleFunction = "PiecewiseFunction"
dtiffDisplay.SelectOrientationVectors = "None"
dtiffDisplay.ScaleFactor = maxshape / 10 - 0.1
dtiffDisplay.SelectScaleArray = "Tiff Scalars"
dtiffDisplay.GlyphType = "Arrow"
dtiffDisplay.GlyphTableIndexArray = "Tiff Scalars"
dtiffDisplay.GaussianRadius = maxshape / 200 - 0.005
dtiffDisplay.SetScaleArray = ["POINTS", "Tiff Scalars"]
dtiffDisplay.ScaleTransferFunction = "PiecewiseFunction"
dtiffDisplay.OpacityArray = ["POINTS", "Tiff Scalars"]
dtiffDisplay.OpacityTransferFunction = "PiecewiseFunction"
dtiffDisplay.DataAxesGrid = "GridAxesRepresentation"
dtiffDisplay.PolarAxes = "PolarAxesRepresentation"
dtiffDisplay.ScalarOpacityUnitDistance = 8.256564094912507
dtiffDisplay.ScalarOpacityFunction = tiffScalarsPWF
dtiffDisplay.IsosurfaceValues = [0.5]
dtiffDisplay.SliceFunction = "Plane"
shape = np.array([xshape, yshape, zshape])
# Init the 'Plane' selected for 'SliceFunction'
dtiffDisplay.SliceFunction.Origin = [xi / 2 - 0.5 for xi in shape]
# Reset view to fit data
renderView1.ResetCamera()
# Changing interaction mode based on data extents
# renderView1.InteractionMode = mode
renderView1.CameraPosition = [
xshape / 2 - 0.5,
yshape / 2 - 0.5,
4.6 * np.sqrt(np.sum(shape / 2 - 0.5)**2)
]
renderView1.CameraFocalPoint = [xi / 2 - 0.5 for xi in shape]
# Get the material library
_ = paraview.simple.GetMaterialLibrary()
# Show color bar/color legend
dtiffDisplay.SetScalarBarVisibility(renderView1, True)
# Update the view to ensure updated data information
renderView1.Update()
# Saving camera placements for all active views
# Current camera placement for renderView1
# renderView1.InteractionMode = mode
renderView1.CameraPosition = [
xshape / 2 - 0.5,
yshape / 2 - 0.5,
4.6 * np.sqrt(np.sum(shape / 2 - 0.5)**2)
]
renderView1.CameraFocalPoint = [xi / 2 - 0.5 for xi in shape]
renderView1.CameraParallelScale = np.sqrt(np.sum(shape / 2 - 0.5)**2)
# Uncomment the following to render all views
# RenderAllViews()
# Alternatively, if you want to write images, you can use SaveScreenshot(...).
threshold1 = paraview.simple.Threshold(Input=dtiff)
threshold1.Scalars = ["POINTS", "Tiff Scalars"]
if phase == 0:
threshold_range = [0.5, 1]
elif phase == 1:
threshold_range = [0, 0.5]
else:
threshold_range = [0, 1]
threshold1.ThresholdRange = threshold_range
# Show data in view
_ = paraview.simple.Show(
threshold1, renderView1, "UnstructuredGridRepresentation"
)
# Hide data in view
paraview.simple.Hide(dtiff, renderView1)
paraview.simple.SaveState(file + ".pvsm")
[docs]
def open_paraview(filename=None, im=None, **kwargs):
r"""
Open a paraview state file or image directly in paraview.
Parameters
----------
filename : str
Path to input state file.
im : ndarray
An image to open directly. If no filename given, then this image is
sent to ``to_paraview`` and a state file is created with a random name.
Any additional keyword arguments are sent to ``to_paraview``.
Examples
--------
`Click here
<https://porespy.org/examples/io/reference/open_paraview.html>`_
to view online example.
"""
if filename is None:
from datetime import datetime
now = datetime.now()
filename = now.strftime("%d-%m-%Y_%H-%M-%S")
to_paraview(im=im, filename=filename, **kwargs)
file = os.path.splitext(filename)[0]
statefile = file + ".pvsm"
# paraview_path = "paraview.exe"
paraview_path = "paraview"
subprocess.Popen([paraview_path, statefile])
[docs]
def spheres_to_comsol(filename, im=None, centers=None, radii=None):
r"""
Exports a sphere pack into a Comsol geometry file.
An image containing spheres can be specified. Alternatively as list of
``centers`` and ``radii`` can be given if known.
Parameters
----------
filename : string or path object
Location and namge to output file
im : ndarray (optional)
A voxel image containing spheres indicated by non-zeros values.
Spheres can be generated using a variety of methods and can overlap.
The sphere centers and radii are found as the peaks in the
distance transform. If ``im`` is not supplied, then ``centers`` and
``radii`` must be given.
centers : array_like (optional)
An array (Ns, 3) of the spheres centers where Ns is the number of
spheres. This must be specified if ``im`` is not suppplied.
radii : array_like (optional)
An Ns length array of the spheres's. This must be specified if ``im``
is not suppplied.
Notes
-----
If ``im`` is given then some image analysis is performed to find sphere
centers so it may not perfectly represent the spheres in the original
image. This is especially true for overlapping sphere and spheres extending
beyond the edge of the image.
Examples
--------
`Click here
<https://porespy.org/examples/io/reference/spheres_to_comsol.html>`_
to view online example.
"""
from ._comsol import _save_to_comsol
if im is not None:
if im.ndim != 3:
raise Exception('Image must be 3D.')
dt = edt(im > 0)
dt2 = nd.gaussian_filter(dt, sigma=0.1)
peaks = (im > 0)*(nd.maximum_filter(dt2, footprint=ball(3)) == dt)
peaks = reduce_peaks(peaks)
centers = np.vstack(np.where(peaks)).T
radii = dt[tuple(centers.T)].astype(int)
_save_to_comsol(filename, centers, radii)