Create a .obj file from 3d array in python

Question

My goal is to get a .obj file from a nifty (.nii) format using python, with the purpose of open it on Unity. I know that the "scikit-image" package has a module called "measure" which has the Marching cube algorithm implemented. I apply the marching cube algorithm to my data and I obtain the results I expect:

verts, faces, normals, values = measure.marching_cubes_lewiner(nifty_data, 0)

I can then plot the data:

fig = plt.figure()
ax = fig.add_subplot(111, projection='3d')
ax.plot_trisurf(verts[:, 0], verts[:,1], faces, verts[:, 2],
                linewidth=0.2, antialiased=True)
plt.show()

I have looked for functions to save the data (verts,faces normals, values) as a .obj but I haven't found one. Thus, I decided to build it myself.

thefile = open('test.obj', 'w')
for item in verts:
  thefile.write("v {0} {1} {2}\n".format(item[0],item[1],item[2]))

for item in normals:
  thefile.write("vn {0} {1} {2}\n".format(item[0],item[1],item[2]))

for item in faces:
  thefile.write("f {0}//{0} {1}//{1} {2}//{2}\n".format(item[0],item[1],item[2]))  

thefile.close()

But when I import the data to unity I got the following result:

So my questions are the followings:

• What I'm doing wrong in the .obj making process?
• Is there a module or function that do this in a better way?
• Is it possible at all to do what I want?

Thank you.

More examples:

Python:

Unity:

Derk

What about your result is wrong compared to how you want it to be? Would you like it to be more smooth or? – Shell

Clearly the faces are wrong, I will try to debug it by creating the simplest figures, like a cube and then see what order they should be written, that is my plan. The .obj making process is somehow wrong, I still don't know why as it should be very simple. – Derk

Well I wasn't sure if it was the faces or an error in the mesh construction. Your plan sounds like a good idea with starting out with simpler meshes. Good way to debug. – Shell

Answer 1

Solution: After hours of debugging, the solution was very simple! Just add +1 to the faces data given by applying the marching cubes. The problem was that Python considered the verts as starting from 0 and unity consider them as starting from 1. That is why they didn't match! You welcome.

verts, faces, normals, values = measure.marching_cubes_lewiner(nifty_data, 0)

faces=faces +1

Success!

Derk