Simple Python for loop causes Blender to suddenly hog 75% of all ram

Question

If I create a plane, and want to subdivide it, I can issue the Python command:

>>> bpy.ops.mesh.subdivide()

But if I want to subdivide 20 times, using:

>>> for i in range(20):
...   bpy.ops.mesh.subdivide()

this causes Blender to get stuck in an infinite loop, eventually taking up 75% of all available RAM.

Why would this happen, and how can I execute said loop?

Answer 1

Well, one problem I can think of is that you're asking blender to subdivide a mesh by cutting everything in half 20 times in a row. That's 2**20 = around 1 million cuts on every edge, meaning a square face would be cut into 1 trillion faces.

If this is really what you want, you can do it in a single step using a single

bpy.ops.mesh.subdivide(number_cuts=1048576)

My suspicion is that this is not really what you want to do. Mostly because I can't think of any machines that can stash the vertices and face loops for several trillion faces.

Answer 2

Try subdividing a plane 20 times manually by successively hitting W ->Subdivide. You won't get past iteration 9 (or 10, 11.. maybe 12) before your machine starts to break out in a sweat. This isn't so much a Python problem as misunderstanding about what subdivide does.

The default single cut per subdivide turns each face into 4 faces. here a short snippet to print out the numbers involved.

prev = [1]

for i in range(20):
    last = prev[-1] * 4    
    print(i+1, prev[-1], '->', last)
    prev.append(last)

outputs (here -> means subdivide):

1 1 -> 4
2 4 -> 16
3 16 -> 64
4 64 -> 256
5 256 -> 1024
6 1024 -> 4096
7 4096 -> 16384
8 16384 -> 65536
9 65536 -> 262144
10 262144 -> 1048576
11 1048576 -> 4194304
12 4194304 -> 16777216
13 16777216 -> 67108864
14 67108864 -> 268435456
15 268435456 -> 1073741824
16 1073741824 -> 4294967296
17 4294967296 -> 17179869184
18 17179869184 -> 68719476736
19 68719476736 -> 274877906944
20 274877906944 -> 1099511627776

after 20 iterations you would have 1 099 511 627 776 faces.

Answer 3

As the other answers mention, you are subidiving a progressively subdivided mesh over and over, which makes polycount grow exponentially. If you want to subdivide 20 times, the proper code would be:

bpy.ops.mesh.subdivide(number_cuts=20)