Override dupligroup's object's material (Python)

Question

How can I link a dupligroup of a group from File A in to File B and override the material of an object within the dupligroup with a material from File B?

The end goal would be to have multiple dupligroups of linked objects all linked from one file, but with different materials. Having the code in one script with the dupligroups and the materials to be applied would be ideal; like this:

I got the idea from this video on Python overrides in Blender: https://youtu.be/bOLMkw_wDGE

The video just shows how to override the attributes of materials and modifiers, but it seems like it should be possible to override a whole material of a dupligroup's child objects. For example:

• source.blend has a group calling 'thing' with an object called 'box' in it with a material applied called 'red'.
• dest.blend has a material called 'blue' and a dupligroup of 'thing' from source.blend.
• In dest.blend I want to override the 'red' material and put 'blue' on the 'box' that is within the linked dupligroup of 'thing.'
• The final goal is being able to import multiple dupligroups of 'thing' and have different materials (from dest.blend) on all of the different dupligroups' 'box' objects.

The closest I've come is this Python line within dest.blend in the python console:

bpy.data.objects['box'].material_slots['red'].material = bpy.data.materials['blue']

This does NOT give me an error message (and it does autocomplete to that point) but it just doesn't do anything.

This is the closest I've come because all my other attempts have given me error messages. It also looked like I was on the right track because just putting in:

bpy.data.objects['box'].material_slots['red'].material

returned:

bpy.data.materials['red']

which matches the syntax of:

bpy.data.materials['blue']

But as I said it didn't seem to do anything.

What am I doing wrong here? Is it even possible to do this with a dupligroup? Any ideas would be very much appreciated. Thank you.

A few notes: I do know about the Edit Linked Libraries add-on, but that doesn't work for what I want to use this method for. As far as I can tell the add-on just lets you more easily edit the linked object's source file thereby editing the linked object, whereas with this method I'd like to be able to have multiple instances of the linked object all linked from one file, but with different materials.

I also know about this thread: Multiple Linked Objects with unique Materials / Group Materials override but not only can I not get any of those examples to work, but also they're just dealing with linked objects and not dupligroups. Dupligroups are a lot more useful in my opinion.

I am new to Blender Python scripting so specific code examples with explanations would be very much appreciated. Thank you.

(This question was asked about Blender versions 2.7*)

This is a (vastly improved) repeat of a question I asked on BlenderArtists.org: http://blenderartists.org/forum/showthread.php?385851-Override-dupligroup-s-object-s-material&highlight= I hope that's okay. The thread died months ago without any solutions.

Answer 1

I came across this question because I have a similar problem. Since material_slots doesn't work, I tried the same method on the object data (in this case, the mesh). And voilá, that works! So instead of doing this:

bpy.data.objects['linked_object'].material_slots[0] = bpy.data.materials['new_material']

Try this:

bpy.data.objects['linked_object'].data.materials[0] = bpy.data.materials['new_material']

(You need to force a viewport update by clicking or so, but apart from that it seems to work fine)

Answer 2

Here is the basic code how to do it with Python: http://www.pasteall.org/66951

import bpy

#1 Link like you do now. Dupligroup object is selected
#-----------------------------------------------
#2 take copies (.copy()) of our group objects
#-----------------------------------------------
o = bpy.context.object
objs = [i.copy() for i in o.dupli_group.objects]

#3 link them to the scene
#-----------------------------------------------
bpy.ops.object.select_all(action='DESELECT')
for i in objs:
    bpy.context.scene.objects.link(i)
    i.select = True

#4 make single users (mat+tex)
#-----------------------------------------------
bpy.ops.object.make_single_user(object=False, obdata=False, material=True, texture=True, animation=False)

#5 change matslot 0 material 
#-----------------------------------------------
for i in objs:
    i.material_slots[0].material = bpy.data.materials[0]    

#6 Use original group name
#-----------------------------------------------
groupname = o.dupli_group.name

for i in objs:
    bpy.data.groups[groupname].objects.link(i)

#7 set dupligroup for empty
#-----------------------------------------------
o.dupli_group = bpy.data.groups[groupname]

#Now you have exact mesh clones with different materials and they are still linked on mesh level and you control them with empty object like before.

Answer 3

As for what you are asking, per the manual this is not possible (please see the first three paragraphs here), as it states the following:

Append and Link

These functions help you reuse materials, objects and other data-blocks loaded from an external source blend-file. You can build libraries of common content and share them across multiple referencing files.

Link creates a reference to the data in the source file such that changes made there will be reflected in the referencing file the next time it is reloaded.

Whereas Append makes a full copy of the data into your blend. You can make further edits to your local copy of the data, but changes in the external source file will not be reflected in the referencing file.

Meaning - Using Linked anything is a one-way street, to get the definition to reside in the source file (including active materials)

Even running the following code confirms that materials are not modifiable, consistent with this definition of the Manual.

So the remainder of this answer, is showing how to accomplish what you are intending to do, without keeping to the strict rule of linking, because the structure/design of the software does not allow this kind of usage.

---

import bpy

##I have three materials defined in my current file
###'a1', 'a2', 'a3'

##When I manually Append the Group in, I run this code:
newSet = bpy.context.selected_objects

count = 0
for i in newSet:
    i.active_material = bpy.data.materials['a' + str(count + 1)]
    if count == 2:
        count = 0
    else:
        count = count + 1
##Then I'm Done

##Now I can make instances of this group

##If I need another variation, you can do the Append again, and run the code again.
###When you do, you will just get another group name: 
####eg. the first one will be 'InstanceGroup', and the second will be 'InstanceGroup.001' with its own objects

##You can easily put some random logic or whatever to select the material in the desired method, but this works.

Here is a demonstration of the comments below.

As you will see, you can keep appending in the same group.

The reason for telling you that you should modify the code logic to fit your purposes, is because as demonstrated, you don't get a variation in color assignment every time you do these steps. So my suggestion is that you add some random assignment logic for this purpose.

Use this code instead if you want to see the material assignment randomized:

import bpy
import random

newSet = bpy.context.selected_objects

count = 0
for i in newSet:
    myint = str(random.choice([1, 2, 3]))
    i.active_material = bpy.data.materials['a' + myint]
    if count == 2:
        count = 0
    else:
        count = count + 1