Linking a Group of linked Groups don't take Dupli Visibility correctly #40164

New Issue

Gabriel Caraballo · 2014-05-12T17:54:34+02:00

Gabriel Caraballo commented

2014-05-12 17:54:34 +02:00

System Information
Ubuntu 12.04

Blender Version
Broken: 2.69, 2.70a
Worked: ?

If you link a GroupA containing 2 or more copies of a linked GroupB with Dupli Visibility set to layer 2, you will see only one GroupB inside the linked GroupA taking Dupli Visibility with layer 2, all other copies will set Dupli Visibility to layer 1.

If you have a group Tree with Dupli Groups sent to layer 2 only in the file Tree.blend linked into the file Forest.blend and grouped as Forest and link Forest group in the blend ForestPicture.blend you will see only one Tree.

Hard to explain, but easy to see the effect in the attached example. linkedBug.zip

Open linkBug.blend you will see 1 cube, and several empties.
Open linkedElement.blend and set layer 1 on at Dupli Visibility of group Group
All the cubes will be visible now on linkBug.blend

Regards,
Eibriel

**System Information** Ubuntu 12.04 **Blender Version** Broken: 2.69, 2.70a Worked: ? If you link a GroupA containing 2 or more copies of a linked GroupB with Dupli Visibility set to layer 2, you will see only one GroupB inside the linked GroupA taking Dupli Visibility with layer 2, all other copies will set Dupli Visibility to layer 1. If you have a group Tree with Dupli Groups sent to layer 2 only in the file Tree.blend linked into the file Forest.blend and grouped as Forest and link Forest group in the blend ForestPicture.blend you will see only one Tree. Hard to explain, but easy to see the effect in the attached example. [linkedBug.zip](https://archive.blender.org/developer/F88679/linkedBug.zip) - Open linkBug.blend you will see 1 cube, and several empties. - Open linkedElement.blend and set layer 1 on at Dupli Visibility of group Group - All the cubes will be visible now on linkBug.blend Regards, Eibriel

Gabriel Caraballo commented

2014-05-12 17:54:34 +02:00

Changed status to: 'Open'

Gabriel Caraballo commented

2014-05-12 17:54:34 +02:00

Added subscriber: @GabrielCaraballo

Lukas Tönne self-assigned this 2014-05-15 13:31:50 +02:00

Lukas Tönne commented

2014-05-15 15:10:21 +02:00

Added subscriber: @Sergey

Lukas Tönne commented

2014-05-15 15:10:21 +02:00

This is a problem with duplis modifying the base Object DNA data when constructing the duplilists. We had similar issues with modified obmats, these are fundamental design mistakes that are really hard to root out without breaking stuff ... Like you said, this problem exists at least in 2.69 already, it's not a recent regression.

Adding @Sergey as CC for further input.

Problem is essentially this behavior:
https://developer.blender.org/diffusion/B/browse/master/source/blender/blenkernel/intern/object_dupli.c$166
Equivalent line before the dupli restructuring in january (6940bf0):
https://developer.blender.org/diffusion/B/browse/master/source/blender/blenkernel/intern/anim.c;0c9d8970433c7653483431f96d33d72106b7c741$734

This code was originally added here:
21e7a5f142

It's not very clear what the issue was/is and whether we still need that, or if it can be implemented in a less hacky way that preserves the integrity of the Object DNA data better.

This is a problem with duplis modifying the base Object DNA data when constructing the duplilists. We had similar issues with modified obmats, these are fundamental design mistakes that are really hard to root out without breaking stuff ... Like you said, this problem exists at least in 2.69 already, it's not a recent regression. Adding @Sergey as CC for further input. Problem is essentially this behavior: https://developer.blender.org/diffusion/B/browse/master/source/blender/blenkernel/intern/object_dupli.c$166 Equivalent line before the dupli restructuring in january (6940bf0): https://developer.blender.org/diffusion/B/browse/master/source/blender/blenkernel/intern/anim.c;0c9d8970433c7653483431f96d33d72106b7c741$734 This code was originally added here: 21e7a5f142 It's not very clear what the issue was/is and whether we still need that, or if it can be implemented in a less hacky way that preserves the integrity of the Object DNA data better.

Bastien Montagne commented

2014-05-15 21:29:09 +02:00

Added subscriber: @mont29