spheres_from_coords: Add option to define shape of output image

[heinsimon]

Hi,

It would be nice, if the spheres_from_coords function would allow to define a image shape, where the spheres given in df are added into.
This could be done as additional mode.

Possible application:

• Add spheres using pseudo_gravity_packing
• Extract the coordinates of the spheres
• Option A:
  • Remove some of the spheres from this list
  • Create new structure using spheres_from_coords
• Option B:
  • Modify the radius of the spheres
  • Create new structure using spheres_from_coords
  • Combine structure from pseudo_gravity_packing with new structure to obtain particle shells

[jgostick]

The main idea of spheres_from_coords is that the coords are coming from some other package or source. I like the suggestion you made elsewhere that the pseudo-packing functions should return (optionally?) as list of coords that could be used in the spheres_from_coords function. If this were the case, would it address your situation?

Btw, it's probably relatively easy to write a 'spheres_to_coords' function. You could use a watershed to find each sphere, find the coords from the center of mass of each spheres, and a local thickness to find their radius, then combine all the results into a data frame.

[heinsimon]

I think it was not clear what I meant:

• If spheres in the input of the spheres_from_coords do not reach x, y or z == 0, I would like to have the option to pad the resulting image to a given shape or add the spheres into a given box.

The following example is not as simple as possible, since ...

• .. there is one bug in pseudo_gravity_packing preventing it from adding all possible spheres (3 in this case, instead only adds one) ...
• and spheres_from_coords has a bug, that it does not allow for adding just one sphere.
• The positions of the spheres in df are identical to the positions of the spheres of packing_bnd.
• The idea is, that one could define the shape of the image, such, that the result of spheres_from_coords would be identical to the output packing_bnd

import porespy as ps
import numpy as np
import matplotlib.pyplot as plt

im_box_bnd=np.full((100,50,50),True)
im_box_bnd[0:10,:,:]=False
im_box_bnd[:,0:10,:]=False
im_box_bnd[:,-10:,:]=False
im_box_bnd[:,:,0:10]=False
im_box_bnd[:,:,-10:]=False

im_box_nobnd=np.full((83,30,30),True)

# Bug in pseudo_gravity_packing: To get spheres with radius R one needs to input radius R+1
packing_bnd=ps.generators.pseudo_gravity_packing(im_box_bnd,r=15,clearance=0,maxiter=3,seed=42)
packing_nobnd=ps.generators.pseudo_gravity_packing(im_box_nobnd,r=15,clearance=0,maxiter=3,seed=42)

# Bug in pseudo_gravity_packing: There is room for 3 spheres, but only one is added.
# df={}
# df['X']=np.array([23,51,79])
# df['Y']=np.array([24,23,26])
# df['Z']=np.array([25,26,23])
# df['R']=np.array([14,14,14])

# Bug in spheres_from_coords: Adding only one sphere is not possible
df={}
df['X']=np.array([23,51])
df['Y']=np.array([24,23])
df['Z']=np.array([25,26])
df['R']=np.array([14,14])

packing_sfc=ps.generators.spheres_from_coords(df,smooth=True,mode='contained')
packing_sfc[27:,:,:]=0 # remove second particle, because spheres_from_coords does not allow only one particle

print(f'np.sum bnd: {np.sum(packing_bnd)}')
print(f'np.sum no bnd: {np.sum(packing_nobnd)}')
print(f'np.sum sfc: {np.sum(packing_sfc)}')


print(f'Shape bnd: {packing_bnd.shape}')
print(f'Shape no bnd: {packing_nobnd.shape}')
print(f'Shape sfc: {packing_sfc.shape}')

fig,ax=plt.subplots(1,3, figsize=[12, 6])
ax[0].imshow(np.sum(packing_bnd,axis=1))
ax[0].set_title('packing bnd')
ax[1].imshow(np.sum(packing_nobnd,axis=1))
ax[1].set_title('packing no bnd')
ax[2].imshow(np.sum(packing_sfc,axis=1))
ax[2].set_title('packing spheres from coords')

[heinsimon]

• pseudo_gravity_packing: Not adding all possible spheres and input radius not correctly used #935
• spheres_from_coords does not allow to add one sphere and adding spheres in same plane (not at 0!) will result in 2D-image #936