What is the definition of blobs?

[javierpagalo]

I understand that the function blobs is like a way to get porous media with a control in the particles or something like that.
But is there any theory behind these? Is there a paper, book or scientific article related to the formulation?

I mean these lines:

blobiness = np.array(blobiness)
shape = np.array(shape)
parallel = kwargs.pop('parallel', False)
divs = kwargs.pop('divs', 2)
cores = kwargs.pop('cores', None)
if np.size(shape) == 1:
    shape = np.full((3, ), int(shape))
sigma = np.mean(shape) / (40 * blobiness)
im = np.random.random(shape)

• Why do sigma is the mean of shape and it is divided by 40 the blobiness?
• What is the blobiness?

I suppose it is a type of particle size, or maybe a shape factor, but I would like to know the exact meaning according to theory.

Thank you.

[jgostick]

There is no science behind it. It's just meant to provide image for prototyping. The code is meant to provide a similar amount of blobs regardless of the image size, and that 40 is just a number that was chosen such that a blobiness of 1 gave reasonable sized blobs. Looking at the code, as you did, is probably the best way to understand how it works. Random numbers are generated between 0 and 1, a gaussian blur is performed using the blobiness to control the size and aspect ratio of the blurring kernel, and then the result is thresholded to give a boolean image with the correct porosity.

[josephinoo]

What range of values does blobiness allow? From 0 to infinity? Or does it have any limits?.I await your prompt reply, thank you in advance.

[jgostick]

I would stick with 0.5 to 5. It works for 2D images, so you can play with it as follows:

import porespy as ps
import matplotlib.pyplot as plt
X = [1, 1]  # Choosing different values for each direction creates an anisotropic image
im = ps.generators.blobs(shape=[500, 500], porosity=0.5, blobiness=X)
plt.imshow(im, interpolation='none', origin='lower')

[josephinoo]

Ok, I understand.
Thank you for the answers and recommendations.

[javierpagalo]

Good Afternoon.
I have another question.
I have been reviewed some Authors.
The method for the porous media generation in porespy.generators.blobs function seems to be based on the following algorithm presented by Jeffrey D. Hyman, C. Larrabee Winter, on their article "Stochastic generation of explicit pore structures by thresholding Gaussian random fields", page 3, https://doi.org/10.1016/j.jcp.2014.07.046.
I'm doing a research and I just wanted to relate the blob function with another previous studies.
Can you confirm if blob function is related on that algorithm?

Obviously is not equal, As you mention on the other answer, it is based on other parameters as blobiness and porosity.
But I mean, It could be a modification of that study.

Waiting for your response
Thank You.

[jgostick]

I did not pull the blobs function from anywhere in particular...maybe a from a stack overflow answer (but it was a long time ago). There is almost nothing of significance to the algorithm: generate random noise, apply Gaussian blur convolution, normalize and rescale the values, threshold, done. Something this simple does not need a citation imho, although you should probably try to explain the steps in your paper (by looking at the code).