Weird deformations on linked meshes modified by a curve #47068

New Issue

Karja Krähwald · 2015-12-27T11:05:57+01:00

Karja Krähwald commented

2015-12-27 11:05:57 +01:00

System Information
Windows 7 64, GTX 980

Blender Version
Broken: f337fea + ee2b583

Short description of error
I have a problem with linked groups that contain any mesh modified by a curve modifier.
If the linked group isnt placed in the scene origin, the topology goes crazy if I change the viewport shading (Solid to Textured Mode).

Exact steps for others to reproduce the error
LinkGroupFromHere.blend

Steps:

Create any Curve object (circle recommended)
Create any Mesh object
- Give this Mesh a Curve Modifier
  - Use Curve as target
(Make sure to edit curve and mesh to fit eachother and have some noticable deformation)
Assign both objects to a Group
Save file
Open a new file
Switch to Solid Shading Mode
Link the Group to this file
Move the Group object away from the scene origin
Bug: Switch to Textured Shading Mode = weird deformation

Note: The weird deformations are more extreme if you move the group object further away before switching to Textured Shading.
Note: No weird deformation if the group object stays in the scene origin. I guess there is a problem with curve-mesh-origin relation in linked data.

Not sure if this is related in any manner (he had problems with the shading as well):
#44154

**System Information** Windows 7 64, GTX 980 **Blender Version** Broken: f337fea + ee2b583 **Short description of error** I have a problem with linked groups that contain any mesh modified by a curve modifier. If the linked group isnt placed in the scene origin, the topology goes crazy if I change the viewport shading (Solid to Textured Mode). ![CrazyMesh 1.png](https://archive.blender.org/developer/F270526/CrazyMesh_1.png) ![CrazyMesh 2.png](https://archive.blender.org/developer/F270528/CrazyMesh_2.png) **Exact steps for others to reproduce the error** [LinkGroupFromHere.blend](https://archive.blender.org/developer/F270531/LinkGroupFromHere.blend) Steps: - Create any Curve object (circle recommended) - Create any Mesh object - Give this Mesh a Curve Modifier - Use Curve as target - (Make sure to edit curve and mesh to fit eachother and have some noticable deformation) - Assign both objects to a Group - Save file - Open a new file - Switch to Solid Shading Mode - Link the Group to this file - Move the Group object away from the scene origin - Bug: Switch to Textured Shading Mode = weird deformation Note: The weird deformations are more extreme if you move the group object further away before switching to Textured Shading. Note: No weird deformation if the group object stays in the scene origin. I guess there is a problem with curve-mesh-origin relation in linked data. Not sure if this is related in any manner (he had problems with the shading as well): #44154

Karja Krähwald commented

2015-12-27 11:05:57 +01:00

Changed status to: 'Open'

Karja Krähwald commented

2015-12-27 11:05:57 +01:00

Added subscriber: @karja

blender-admin commented

2015-12-27 11:05:57 +01:00

#47175 was marked as duplicate of this issue

Sergey Sharybin commented

2015-12-30 10:53:49 +01:00

Added subscribers: @mont29, @ideasman42, @Sergey

Sergey Sharybin commented

2015-12-30 10:53:49 +01:00

Seems toggling viewport shading caused group re-evaluation due to missing data needed for the textured view. And perhaps this will evlauate modifier with a new relative locations or so.

@ideasman42, @mont29, some extra brain here wouldn't hurt :)

Seems toggling viewport shading caused group re-evaluation due to missing data needed for the textured view. And perhaps this will evlauate modifier with a new relative locations or so. @ideasman42, @mont29, some extra brain here wouldn't hurt :)

Bastien Montagne commented

2015-12-31 17:06:12 +01:00

Yeah, there is something wrong with re-evaluation here… if, in Relations Extra panel of instancing empty, we enable 'Extra Data Update', and change frame, deformation goes back to normal expected result. Same thing with both depsgraph (old and new ones).

Bastien Montagne commented

2016-01-01 16:00:09 +01:00

Issue comes from Duplis (what a surprise…), duplilist_apply() generates the DM of the mesh only once - and in original object's space, not in dupli space. Placing copy_m4_m4(dob->ob->obmat, dob->mat); before DM generation fixes the issue in this case, but this will fail in case mesh is used several times by obspace-dependent modifiers, so think the only correct solution is to add a modifier flag saying 'I depend on obspaces', and when set, not caching a same DM for all duplis…

Issue comes from Duplis (what a surprise…), `duplilist_apply()` generates the DM of the mesh only once - and in original object's space, not in dupli space. Placing `copy_m4_m4(dob->ob->obmat, dob->mat);` **before** DM generation fixes the issue in this case, but this will fail in case mesh is used several times by obspace-dependent modifiers, so think the only correct solution is to add a modifier flag saying 'I depend on obspaces', and when set, not caching a same DM for all duplis…

Campbell Barton commented

2016-01-14 11:02:20 +01:00

Added subscriber: @Traverso

Campbell Barton commented

2016-01-14 11:12:41 +01:00

Note this is one of those limits in Blender's dupli system - its been a problem for a long time (just merged in a duplicate, but there would be others too).

Any time modifiers are calculated by the dupli system its not reliable.

Note this is one of those limits in Blender's dupli system - its been a problem for a long time (just merged in a duplicate, but there would be others too). Any time modifiers are calculated by the dupli system its not reliable.

Sergey Sharybin commented

2016-02-02 12:59:51 +01:00

Changed status from 'Open' to: 'Archived'

Sergey Sharybin closed this issue

2016-02-02 12:59:51 +01:00

Sergey Sharybin self-assigned this 2016-02-02 12:59:51 +01:00

Sergey Sharybin commented

2016-02-02 12:59:51 +01:00

Since duplicaiton system is one of the subjects for bigger re-consideration it makes sense not to spend time trying to solve existing system but combine forces focus on a new system now.

Thanks for the report, it'll definitely be a good test scenario for the new dupli system, but consider it all a TODO for now since it's just all limitation of existing system.

Since duplicaiton system is one of the subjects for bigger re-consideration it makes sense not to spend time trying to solve existing system but combine forces focus on a new system now. Thanks for the report, it'll definitely be a good test scenario for the new dupli system, but consider it all a TODO for now since it's just all limitation of existing system.

Sign in to join this conversation.

No Label

Download

What's New

Blender Studio

Manual

Developers Blog

Documentation

Benchmark

Blender Conference

Development Fund

One-time Donations

Weird deformations on linked meshes modified by a curve #47068