[x3d-public] Fw: Re: [x3d] Minutes from Nov 12 X3D Blender suppport call

Thu Nov 14 06:47:43 PST 2024

The global matrix rotation to adjust coordinates  space in Blender is in
__init__.py.  I can’t comment on glTF.  Both export and import are treated
differently.  There’s have been recent discussions on the Blender website
which people are welcome to join:

https://projects.blender.org/extensions/io_scene_x3d/issues/24

https://projects.blender.org/extensions/io_scene_x3d/commit/6c5160405ed9f64610c78b76849cf547d493baa2

Note the RGB gizmo in my images versus Cedric’s.

On Thu, Nov 14, 2024 at 12:07 AM Michalis Kamburelis via x3d-public <
x3d-public at web3d.org> wrote:

> 1. I see there's some talk in this thread about Blender coordinate
> system. Let me try to clear things up :)
>
> I would recommend everyone to just fire up Blender and check what it
> actually does :) These things are trivial to see in 3D space in
> Blender, there's even a helpful gizmo showing axis, you can play
> around and verify what coordinate system it uses.
>
> To be clear, Blender doesn't use exactly Direct3D coordinate system
> (which I think Joe refers to when he writes "Dx").
>
> The way things are, up to my knowledge:
>
> - Blender, just like glTF, just like X3D, just like (by default)
> OpenGL, just like Castle Game Engine, uses right-handed coordinate system.
>
> - This is different than Direct3D that uses left-handed coordinate system.
>
> The "left/right-handedness" is an independent decision from "what is
> up by convention":
>
> - Blender, just like 3ds Max, chose Z as "up" vector by convention.
>
> - This is different than glTF, X3D, Unity, Godot, Castle Game Engine
> where we chosen "Y up". And that's why various exporters (from glTF,
> from X3D...) need an option to "change to +Y up convention".
>
> Disclaimer: talking about "coordinate system" for rendering libraries
> like OpenGL or Direct3D is only talking about conventions. They can be
> adjusted.
>
> See also graphic on
> https://www.reddit.com/r/gamedev/comments/b6ujvb/y_axis_up_or_z_axis_up/
> . Yeah, the world is complicated, unfortunately there are a few
> decisions and lots of things made different choices :)
>
> 2. Converting "up Z" to "up Y" (when both systems are right-handed) is
> a matter of rotating the model (around X axis by 90 degrees).
>
> You can explore how it is done by glTF exporter when "+Y Up" is
> selected in the GUI by checking out
> https://github.com/KhronosGroup/glTF-Blender-IO sources and grepping
> for "gltf_yup". You can also experiment by exporting models from
> Blender and looking at output (convert it e.g. using Castle Model
> Viewer to see X3D).
>
> From what I can see, it is *not* just adding a top-level rotation,
> it's actually rotating all the data inside.
>
> Regards,
> Michalis
>
> czw., 14 lis 2024 o 05:18 Joe D Williams via x3d-public
> <x3d-public at web3d.org> napisał(a):
>
> >
> >
> > -----Forwarded Message-----
> > From: Joe D Williams <joedwil at earthlink.net>
> > Sent: Nov 13, 2024 7:38 PM
> > To: Vincent Marchetti <vmarchetti at kshell.com>
> > Cc: X3D Ecosystem public discussion <x3d-ecosystem at web3d.org>
> > Subject: Re: [x3d-public] [x3d] Minutes from Nov 12 X3D Blender suppport
> call
> >
> > Hi Vince,
> > The basic idea is that x3d is cast in the GL coordinate system where +y
> is up, model faces +z is toward the viewer, and +x is to the model left,
> viewer right. Axis rotations follow right hand rule. + is ccw. .
> > So for the ideal HAnim source pose before animation all joints and
> points are 0 0 1 0.
> > The host world coord sys is same as hanim model coord sys.
> >
> > From what I seem to see I think the Blender world coordinate system is
> +z up and left hand rule for rotations. That is why the gltf needs the +Yup
> flag, because gltf is cast in GL coord sys.
> > I am not sure what that gltf export +Yup flag does to the data from
> blender. Does it just do that x-axis rotation, or more?
> >
> > So, this may be called the problem of using a Dx model in a GL system.
> Converting seems not so easy to export by just rotation of x-axis as shown
> in the example because joint and point coordinates may not be as expected
> GL and x3d 0 0 1 0 and animations may be a further problem.
> >
> > This is opportunity that needs some work.to understand. Right,we need a
> simple example that shows default x3d model coordinate system and a simple
> animation in that system,, import to Blender and see what happens, make
> some simple mods and animations then see what it exports and how it plays
> in x3d. Turns out that Vulcan is the only khronos that uses what think is
> the Dx as a basis
> >
> > John actually has some experience in this in his work with the kicker.
> >
> > Good Luck, it is Important, since may authoring and run times have
> chosen the Dx and others something different from either.
> >
> > All Best,
> > Joe
> >
> >
> > -----Original Message-----
> > From: Vincent Marchetti
> > Sent: Nov 13, 2024 1:24 PM
> > To: Joe D Williams
> > Cc: X3D Ecosystem public discussion
> > Subject: Re: [x3d-public] [x3d] Minutes from Nov 12 X3D Blender suppport
> call
> >
> > I have seen this issue brought up before. Is the problem particular to
> exporting HAnim type models, or is it generic to all Blender to X3D exports
> >
> > What would help understand and potentially create a solution would be if
> someone created a .blend file that had
> > -- a single shape, with low poly mesh but with a visually obvious front
> and up direction
> > -- that shape oriented in Blender space with the up direction along
> Blender +Z and the forward direction along Blender -Y, and this achieved
> without any rotation of the mesh defined in Blender.
> >
> > Then we can examine X3D exports with various options to see if there's a
> problem to fix in the code.
> >
> > Vince Marchetti
> >
> >
> > > On Nov 13, 2024, at 4:03 PM, Joe D Williams wrote:
> > >
> > >
> https://github.com/andreasplesch/x3dom/wiki/RiggedSimple-glTF-conversion
> > >
> > > X3D
> > > experimental xml
> > > ...
> > >
> > > Transform DEF='node_0' rotation='1 0 0 1.57
> > >
> > > This is not good. It shows the thing was drawn +zup.
> > > This should be fixed in blender either by fixing blender to do the GL
> default, not depending on the Transform of the model to put the thing into
> GL space.
> > > Thanks,
> > > Joe
> > > -----Original Message-----
> > > From: Extensible 3D (X3D) Graphics public discussion
> > > Sent: Nov 13, 2024 8:00 AM
> > > To: Vincent Marchetti , X3D Ecosystem public discussion , ,
> > > Cc: Joe D Williams
> > > Subject: Re: [x3d-public] [x3d] Minutes from Nov 12 X3D Blender
> suppport call
> > >
> > > So, to me for hanim it seems like we must fix Blender at the source.
> > > Just about the idea that Blender considers all Joints can affect all
> points, seems like that is a non-starter. Doesn't this liimit the
> skeleton-driven skin animation to less than 16 Joint nodes?
> > > If so, then Blender needs to be fixed to operate more like x3d hanim,
> and every other tool, to allow for more skeleton joints(?) Gltf has a
> suggested? limit of, four joints to affect any point.
> > > What am I missing?
> > > Thanks for all the great work on this import/export with Blender.
> > > Joe
> > >
> > >
> > > -----Original Message-----
> > > From: Vincent Marchetti
> > > Sent: Nov 13, 2024 6:36 AM
> > > To: X3D Ecosystem public discussion ,
> > > Subject: [x3d] Minutes from Nov 12 X3D Blender suppport call
> > >
> > > Minutes from the Nov 12 2024 meeting on Blender 4.2 support for X3D
> are available for view and comment
> > >
> > >
> https://docs.google.com/document/d/1wvjvDpJTMFDMpS53ImuoRHNOCt9zrCiq/edit
> > >
> > > Vince Marchetti
> > >
> > >
> > > _______________________________________________
> > > x3d mailing list
> > > x3d at web3d.org
> > > http://web3d.org/mailman/listinfo/x3d_web3d.org
> > >
> > >
> > >
> > > _______________________________________________
> > > x3d-public mailing list
> > > x3d-public at web3d.org
> > > http://web3d.org/mailman/listinfo/x3d-public_web3d.org
> > >
> > >
> >
> >
> >
> >
> >
> >
> > _______________________________________________
> > x3d-public mailing list
> > x3d-public at web3d.org
> > http://web3d.org/mailman/listinfo/x3d-public_web3d.org
>
> _______________________________________________
> x3d-public mailing list
> x3d-public at web3d.org
> http://web3d.org/mailman/listinfo/x3d-public_web3d.org
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://web3d.org/pipermail/x3d-public_web3d.org/attachments/20241114/9099d2fa/attachment-0001.html>