Set Origin to center of mass has to be performed several times #65986

New Issue

Robert S · 2019-06-21T09:24:19+02:00

Robert S commented

2019-06-21 09:24:19 +02:00

System Information
Operating system: Windows 10
Graphics card: 1080

Blender Version
Broken: 2019-06-16 12da679fa0
Worked: (optional)

Short description of error
In the Object menu, the command "Origin to center of mass (volume)" needs to be activated several times in order to have the origin actually end up at the center of all objects.

Exact steps for others to reproduce the error

Open attached file
Select all
Perform "Origin to center of mass (volume)"

Result:
Note how in certain objects, the origin is still outside of the object (and note how it creeps closer on repeated activations; certain objects need several).

Expected result:
The result should perform the correct calculation at the first activation. Subsequent repeats should not be necessary.

1.FBX

**System Information** Operating system: Windows 10 Graphics card: 1080 **Blender Version** Broken: 2019-06-16 12da679fa094 Worked: (optional) **Short description of error** In the Object menu, the command "Origin to center of mass (volume)" needs to be activated several times in order to have the origin actually end up at the center of all objects. **Exact steps for others to reproduce the error** 1. Open attached file 2. Select all 3. Perform "Origin to center of mass (volume)" Result: Note how in certain objects, the origin is still outside of the object (and note how it creeps closer on repeated activations; certain objects need several). Expected result: The result should perform the correct calculation at the first activation. Subsequent repeats should not be necessary. [1.FBX](https://archive.blender.org/developer/F7427054/1.FBX)

Robert S commented

2019-06-21 09:24:19 +02:00

Added subscriber: @RobertS

Philipp Oeser commented

2019-06-21 09:51:34 +02:00

Added subscriber: @lichtwerk

Philipp Oeser commented

2019-06-21 09:51:34 +02:00

Confirmed at first sight, seems to be the case if origin is far off in relation to geometry bounds (possibly precission issue?), Origin to Geometry or Origin to center of mass (surface) dont seem to suffer the same issues, checking...

Confirmed at first sight, seems to be the case if origin is far off in relation to geometry bounds (possibly precission issue?), `Origin to Geometry` or `Origin to center of mass (surface)` dont seem to suffer the same issues, checking...

Philipp Oeser commented

2019-06-21 10:04:50 +02:00

simpler testcase:
T65986_simple.blend

simpler testcase: [T65986_simple.blend](https://archive.blender.org/developer/F7427642/T65986_simple.blend)

Philipp Oeser commented

2019-06-21 11:11:59 +02:00

We have two "competing" methods here:

Origin to center of mass (volume) was introduced in 37bc3850ce
it was soon found that there can be issues with this, so the Origin to center of mass (surface) was brought back 361c7cbbc5
so in the end this might not be a bug, Origin to center of mass (surface) will always work fine here... [this is the suggested workaround for now...]

Comments suggest the volume based method has following issues:

The new method while improved for solid objects doesn't work for non-manifold meshes, keep both
The resulting volume will only be correct if the mesh is manifold and has consistent face winding (non-contiguous face normals or holes in the mesh surface)

None of these apply to this report though (afaict), here is another testcase where a couple of different offsets are tested:
T65986_offset.blend
(still not sure if this is a precission issue alone? -- checking further...)

not a regression from 2.79 btw., same thing there...

We have two "competing" methods here: - `Origin to center of mass (volume) `was introduced in 37bc3850ce - it was soon found that there can be issues with this, so the `Origin to center of mass (surface)` was brought back 361c7cbbc5 - so in the end this might not be a bug, `Origin to center of mass (surface)` will always work fine here... [this is the suggested workaround for now...] Comments suggest the volume based method has following issues: > The new method while improved for solid objects doesn't work for non-manifold meshes, keep both > The resulting volume will only be correct if the mesh is manifold and has consistent face winding (non-contiguous face normals or holes in the mesh surface) None of these apply to this report though (afaict), here is another testcase where a couple of different offsets are tested: [T65986_offset.blend](https://archive.blender.org/developer/F7428259/T65986_offset.blend) (still not sure if this is a precission issue alone? -- checking further...) not a regression from 2.79 btw., same thing there...

Germano Cavalcante commented

2020-02-03 15:54:02 +01:00

Added subscriber: @mano-wii