Calculating split normals using python?

Question

I'd like to use a python script to transfer normals to a new uv layer. Here is my first attempt:

import bpy
ob = bpy.context.object
for loop in ob.data.loops:
    normal = ob.data.vertices[loop.vertex_index].normal
    ob.data.uv_layers["XY"].data[loop.index].uv = (normal.x, normal.y)
    ob.data.uv_layers["ZW"].data[loop.index].uv = (normal.z, 0)

What I eventually want the script to be able to do is read custom split normals. I heard this was possible through the use of this script. The normal section underneath is supposed to calculate the normals using calc_normals_split or calc_tangents. However every time I try to incorporate this into my script I keep getting error messages or the result comes out as just one color. If anyone could help me to make it work would be much appreciated.

Answer 1

MeshLoop.normal

As commented by @scurest

After using ob.data.calc_normals_split() you access the normals with loop.normal. What error message are you getting? Make sure you are in object mode and not edit mode when you run the script.

Look at the loop normal, rather than the vertex normal to get the custom split normal values.

Test script, using foreach set to set the UV values rather than directly looping over them.

import bpy
import numpy as np
context = bpy.context
ob = context.object
me = ob.data
get or create uv layers
xy_layer = me.uv_layers.get("XY") or me.uv_layers.new(name="XY")
zw_layer = me.uv_layers.get("ZW") or me.uv_layers.new(name="ZW")
calc split nomrals
me.calc_normals_split()
norms = np.empty(3 * len(me.loops))
me.loops.foreach_get("normal", norms)
nx, ny, nz = norms.reshape((-1, 3)).T
nw = np.zeros(len(nz))
xy_layer.data.foreach_set("uv", np.array([nx, ny]).T.ravel())
zw_layer.data.foreach_set("uv", np.array([nz, nw]).T.ravel())