can't set the location of the object by program?

Question

in my blender file, I have created two vehicles objects which named as 'AudiA2' and 'JeepCherokee'

now I have created a simple program to show them on the scene with different positions created from the program.

 def choose_car_position_byname(name, pos):    

    bpy.data.objects[name].hide_render = False;    
    print("location     ", bpy.data.objects[name].location )
    bpy.data.objects[name].location = bpy.data.objects[name].location + mathutils.Vector((pos,1,0))
    print("location     ", bpy.data.objects[name].location ) 

    return bpy.data.objects[name]

 car1 = choose_car_position_byname('AudiA2',1)
 car2 = choose_car_position_byname('JeepCherokee',0)

However from the program executed, I can only see one car on the scene, which is the car2. as shown below

Ultimately I found out the second car will block the first car, because if I added the following code, two car will show up

bpy.data.objects["AudiA2"].rotation_euler.z = 270 / 180 * math.pi

as shown in here

---

Now the question is why I can't set the location of the obj. I have tried many ways, but none of them is not working, it is totally puzzling me. My environment is OSX 10.10.1, blender 2.76.

# first approach   
bpy.data.objects["AudiA2"].location.x += 3.0
bpy.data.objects["AudiA2"].location.y += 2.0
bpy.data.objects["AudiA2"].location.z += 0.0


# second approach
vec = mathutils.Vector((3.0, 2.0, 1.0))
inv = bpy.data.objects['AudiA2'].matrix_world.copy()
inv.invert()
vec_rot = vec * inv
bpy.data.objects['AudiA2'].location = bpy.data.objects['AudiA2'].location + vec_rot    

# third approach
bpy.data.objects["AudiA2"].location = (20, 20, 10)

# fourth approach
bpy.data.objects["AudiA2"].location[0] = 3
bpy.data.objects["AudiA2"].location[0] = bpy.data.objects["AudiA2"].location[0] + 3

#fifth approach
bpy.data.objects['AudiA2'].location = bpy.data.objects['AudiA2'].location + mathutils.Vector((3,2,0))

---

The code and blender files are all uploaded to dropbox. https://www.dropbox.com/s/fxar4owxvkxrxi7/scene2.blend?dl=0 https://www.dropbox.com/s/xvsg879k2bil2pn/batch3.py?dl=0

to work with the code, you have to pip install pypng, which could be used to output the png image.To run the application, the command is similar to this-

./blender_application_path -b blend_path -P python_path

for example

./Applications/Blender/blender.app/Contents/MacOS/blender -b scene2.blend -P batch3.py

Answer 1

When directly manipulating transform values fails (it shouldn't, but it does sometimes), I often just try to use the transform operators insated, and it often fixes the problem.

 def choose_car_position_byname(name, pos):    
    o = bpy.data.objects[name]
    o.hide_render = False;    
    print("location     ", o.location )

    bpy.ops.object.select_all( action = 'DESELECT' )
    o.select = True
    bpy.context.scene.objects.active = o

    bpy.ops.transform.translate( value = ( pos, 1, 0 ) )

    print("location     ", o.location ) 

    return o

 car1 = choose_car_position_byname('AudiA2',1)
 car2 = choose_car_position_byname('JeepCherokee',0)

By the way, if this still fails, check if you have keyframes on the car objects' locations. If you do, that could definitely explain why they aren't moving (the keyframes will keep them in place).

EDITED: based on blend file, my suspicion has indeed been confirmed that you have keyframes on your car models. You need to delete these keyframes to enable control over the objects' locations via script. Here's how to do this (mouse cursor is a bit off due to weird issues in recording on my laptop's screen, you need to select a car object, then right click on any of the location boxes in the properties panel):

Answer 2

I ran into a similar problem to this and it had nothing to do with keyframes. Just thought I'd share what I found. It seems that unless at least one of the layers an object is visible in is enabled at the time you change its location via the python API, the position of the object will not be updated when you subsequently enable one of those views. For example, start with the default single Cube visible in layer 1. Switch to viewing just layer 2 and run this script:

bpy.data.objects["Cube"].location.x += 3.0
bpy.data.objects["Cube"].location.y += 2.0
bpy.data.objects["Cube"].location.z += 1.0

When you return to the 3D view and click layer 1, Cube will still be sitting at 0,0,0 even though if you look at its listed location in the transform window it will show 3, 2, 1 like you'd expect.

Now do the same thing starting at the default Cube but run this instead:

layer_save = bpy.context.scene.layers[0]
bpy.context.scene.layers[0] = True
bpy.data.objects["Cube"].location.x += 3.0
bpy.data.objects["Cube"].location.y += 2.0
bpy.data.objects["Cube"].location.z += 1.0
bpy.context.scene.update()
bpy.context.scene.layers[0] = layer_save

What this does is enable layer 0 while the location is being modified and then return its state to how it started. Now when you return to the 3D view and click layer 0, the cube will have actually moved. To be safe, when moving not visible objects like this, it probably makes more sense to save the state of all 20 layers, enable them all, do the move then restore all 20 layers. I don't know how expense that is computationally if you are doing lots of object moving. I only have a couple in my application.

I don't know if this is a Blender bug or not, but the behavior of changing the object's location when its not in a visible layer seems rather counterintuitive to me. I tend to think its not intended behavior since if you repeat the first 3 line script at the top of this post, then return to layer 1 and click Object->Snap->Selection To Cursor, the cube will jump to the offset you specified at 3,2,1 instead of 0,0,0 where the cursor is.