Extra rotation exporting to collada #65712

New Issue

Luis W. Sevilla · 2019-06-11T11:51:25+02:00

Luis W. Sevilla commented

2019-06-11 11:51:25 +02:00

System Information
Operating system: Ubuntu 18.04
Graphics card: Intel, Nvidia

Blender Version
Broken: 2.79b

Short description of error
When exporting to collada format it seems to add extra rotation to objects.

Exact steps for others to reproduce the error
I've created a single plane (file planos.bend) and exported it to file planos2.dae. Then imported and exported it 2 more times (attached results). End situation is added as planos_bug.blend.
I'm unable to export and object in the same position I have it on blender.
My target is google earth and it seems imports of .dae behave identically both in GE and in Blender

Based on the default startup or an attached .blend file (as simple as possible).

**System Information** Operating system: Ubuntu 18.04 Graphics card: Intel, Nvidia **Blender Version** Broken: 2.79b **Short description of error** When exporting to collada format it seems to add extra rotation to objects. **Exact steps for others to reproduce the error** I've created a single plane (file planos.bend) and exported it to file planos2.dae. Then imported and exported it 2 more times (attached results). End situation is added as planos_bug.blend. I'm unable to export and object in the same position I have it on blender. My target is google earth and it seems imports of .dae behave identically both in GE and in Blender Based on the default startup or an attached .blend file (as simple as possible). [planos.blend](https://archive.blender.org/developer/F7098411/planos.blend) [planos_bug.blend](https://archive.blender.org/developer/F7098412/planos_bug.blend) [planos2.dae](https://archive.blender.org/developer/F7098413/planos2.dae) [planos2-1.dae](https://archive.blender.org/developer/F7098414/planos2-1.dae) [planos2-2.dae](https://archive.blender.org/developer/F7098415/planos2-2.dae)

Luis W. Sevilla commented

2019-06-11 11:51:25 +02:00

Added subscriber: @LuisW

Philipp Oeser commented

2019-06-11 15:06:35 +02:00

Added subscribers: @GaiaClary, @lichtwerk

Gaia Clary was assigned by Philipp Oeser

2019-06-11 15:06:35 +02:00

Philipp Oeser commented

2019-06-11 15:06:35 +02:00

Can confirm.

@GaiaClary : mind checking?

Can confirm. @GaiaClary : mind checking?

Gaia Clary commented

2019-06-12 19:05:47 +02:00

Hi, @LuisW

Thanks for reporting this. This is an issue in the importer. However we can not fix this right now. But you can fix the data instead. here is the reasoning:

The problem is that the Object transform matrix contains a scale of 0 along Z. This causes 2 issues:

The normals become 0 along z which means the object is like a black hole (no light reflected)
The object transformation matrix is:

    Matrix(
        (  0.0, 0.0, 0.0, 0.0),
        (  0.0, 5.0, 0.0,  0.0),
        (-20.0, 0.0, 0.0, 20.0),
        (  0.0, 0.0, 0.0,  1.0)
    )

When decomposing this matrix back into scale, location and rotation, then Blender gets the rotation part wrong. I am not entirely sure why that happens, probably because the math gets odd when the scale in one dimension is 0 (note that scale and dimension are two different things here!)

After some internal discussion i am tempted to reject this as "won't be handled by the importer". As indicated above you can handle this in a much better and more clean way by keeping reasonable scales on your object, for example: scale = (20, 5, 1) would be a perfect fix for your particular case. It would even avoid the total black surface.

Although you most probably want to fix the data anyways, I am still trying to figure out if we can possibly handle this in the importer (i tested that other tools actually can read the colladafile correctly). So i keep this report open for now.

Hi, @LuisW Thanks for reporting this. This is an issue in the importer. However we can not fix this right now. But you can fix the data instead. here is the reasoning: The problem is that the Object transform matrix contains a scale of 0 along Z. This causes 2 issues: - The normals become 0 along z which means the object is like a black hole (no light reflected) - The object transformation matrix is: ``` Matrix( ( 0.0, 0.0, 0.0, 0.0), ( 0.0, 5.0, 0.0, 0.0), (-20.0, 0.0, 0.0, 20.0), ( 0.0, 0.0, 0.0, 1.0) ) ``` When decomposing this matrix back into scale, location and rotation, then Blender gets the rotation part wrong. I am not entirely sure why that happens, probably because the math gets odd when the scale in one dimension is 0 (note that scale and dimension are two different things here!) After some internal discussion i am tempted to reject this as "won't be handled by the importer". As indicated above you can handle this in a much better and more clean way by keeping reasonable scales on your object, for example: scale = (20, 5, 1) would be a perfect fix for your particular case. It would even avoid the total black surface. Although you most probably want to fix the data anyways, I am still trying to figure out if we can possibly handle this in the importer (i tested that other tools actually can read the colladafile correctly). So i keep this report open for now.

Gaia Clary was unassigned by Dalai Felinto

2019-12-23 16:34:06 +01:00

Germano Cavalcante commented

2020-02-06 18:37:07 +01:00

Added subscriber: @mano-wii

Germano Cavalcante commented

2020-02-06 18:37:07 +01:00

I can still reproduce the problem in blender 2.82
Given the time this report is open and the comment, I will mark it as a Known Issue.

I can still reproduce the problem in blender 2.82 Given the time this report is open and the comment, I will mark it as a `Known Issue`.

Philipp Oeser removed the

Download

What's New

Blender Studio

Manual

Developers Blog

Documentation

Benchmark

Blender Conference

Development Fund

One-time Donations

Extra rotation exporting to collada #65712