[x3d-public] X3D, Blender, and which way is up.
John Carlson
yottzumm at gmail.com
Mon Nov 25 02:09:30 PST 2024
Probably this will be ignored. Joe makes a valid point on this one.
This comment is about X3D and Blender, not glTF and Blender. “Sorted out,”
but look at the coordinate axes with when you load a typical coordinate X3D
coordinate axes (with x,y,z labels) into Blender. Then look at what
Blender thinks in the coordinate axes gizmo (see the bulbous X,Y,Z
coordinates with red, green, blue coloring).
Nothing has changed but defaults, AFAIK, which I have been trying to
override.
What we really want is Blender’s “floor” to be in the X-Z plane. You can
make the “floor” as a “wall” be in the X-Z plane by choosing the correct
viewpoint, but try to move the gizmo slightly. What they’ve done is put
the “floor” grid in the X-Z plane temporarily (as a wall). Try moving the
gizmo and watch the “floor” flip out of the X-Z plane.
It’s a marketing ploy…the Emperor has no clothes.
We’d be perfectly happy if we could select which plane the Blender “floor”
was in, with no changes to the defaults. In python, please!
If you hit Num pad 7 in Blender, you will see the typical X3D viewpoint
with +Y up, +Z towards the viewer, and +X to the right, just like you
learned in high school, just like OpenGL. The only problem is the floor is
now a wall! Indeed, it looks a lot like the Cartesian Coordinate System we
all learned, with the grid in the a X-Y plane. But wait, the grid is
supposed to be the floor, I thought…
Don’t look at the model! Look at the gizmo! Look at them both!
I don’t know what Blender was thinking making the floor so hard to change.
If we could shift some concrete there, that would be great!
The great trick is to make the standard floor invisible and draw your own
floor, which I haven’t done yet because I’ve been struggling with
elasticman. If someone does that to you, try selecting the grid. You
can’t select the floor grid in normal Blender. Zorro unmasked!
Where’s this is really going to hurt everyone is when we try to introduce
animations, that’s where elasticman steps in. We’re going to have to apply
Blender’s ideas to our interpolators. That means negating z…what does that
mean for orientation interpolators? Just negate Z there? Really, if
someone has an answer for this, you earn a gold star in my book! Euler
angles with negated z???
Hopefully we can solve animations without forcing everything to default X3D
axes (which was what I was trying to do by setting up to be forward and
forward, up—please check my math—I’ll probably change my Blender importer
back to that). That’s in __init__.py and can be set on import separate
from export.
What happens when you swap -Z with Z in your animations? What do you also
do on export? Have we even considered export yet?
Hopefully, the underlying code is doing rotation and not scaling!
I think the safest bet is loading X3D models “laying down” and moving the
floor underneath their feet. This will give +Y above the models head, +Z
in front of the model, and +X to the models left.
I’m not seeing importing X3D animations into blender coming to X3D soon
until everyone gets their heads out of the sand.
I heard that HAnim was built on dead people’s measurements. If we get more
people’s eyes on the technical details, maybe Blender won’t kill X3D.
I really like that Blender uses a right hand coordinate system!
Oh, I believe you can set floor *constraints* for an object. If you can
set the floor constraints to the fake floor, maybe that would work?
I sure wish it was as easy as swapping -Z for Z. In my dreams!
Does glTF animation work in Blender?
I’m sure smarter people than me have been down this path, why HAnim is not
implemented in Blender, why animation import is disabled, etc. I’m just a
math dummy who thinks it’s possible, indeed, I wrote an HAnim exporter that
mostly worked with one use case. No one gave me any other use cases, so I
started my own HAnim importer.
Perhaps we should only focus on Blender exports? Anyone? This seems like
a higher priority. But then, do we expect HAnim names? How do we map
Transforms to HAnim? Been there, done that:
https://github.com/coderextreme/HAnimDecoratorAssembly let me know if
someone wants me to work on that, but I require HAnim names for
Transforms. My system is limited to 32GB. For larger skins (500000
polys), I’ll probably have to recode in something besides Perl. X3DJSAIL
is likely, if it’s ready to handle the workload.
Maybe moving Blender’s floor is a bit like moving the lid of a coffin?
https://youtu.be/6o2gISJYwQU
<https://youtu.be/6o2gISJYwQU?si=uuCoKE8W_-6NBBEp> Motel is glTF, Blender
is Tzietel, and Lazar Wolf is X3D? That scene was pretty terrifying as a 7
year old.
Maybe we should fork Blender?
Another thing I see with importing animations is PREF_FLAT equals True
(current). This currently wipes out intermediate transforms, and puts them
on shapes (there’s a hierarchy of shapes with associated matrices in
blender). So how the heck are we supposed to animate intermediate
Transforms if they aren’t there? This whole thing needs to be rethought.
We should start by looking at creating EMPTY objects as Transforms, as Joe
said. Then we need to think about what to do on export. Since I just
tripped upon a hierarchy of shapes (Thanks, Katy), this whole thing might
need to be thought out. Fortunately, Katy’s example exports well with
Blender’s exporter (but I’ve not tried recently, since the -Z/Z change.
If people want to get serious about HAnim and animations in Blender, and
have mental capacity to understand the issues with PREF_FLAT, the effect of
animating X3D files in Blender with different coordinate systems and
import/export of animations, especially skin weights, maybe we can talk.
I’m yottzumm on Discord and I hang out in The CGE, MSF and Khronos servers
(friend me). I’m also on a Blender server, but there’s probably many.
Maybe we should just prioritize creating glTF and OpenUSD from X3D?
John
On Sun, Nov 24, 2024 at 8:11 PM Brutzman, Donald (Don) (CIV) via x3d-public
<x3d-public at web3d.org> wrote:
> Your definition sounds good to me. Glad to hear that this is consistently
> sorted out and implemented in recent Blender version.
>
> Here are X3D Tooltip hints for Viewpoint and orientation:
>
> - X3D Tooltips: Viewpoint
> - Viewpoint provides a specific location and direction where the user
> may view the scene. Viewpoints are the primary way for a user to navigate
> within a scene, and for an author to show critical aspects of a model.
> Unless modified by the orientation field, the default direction for a
> Viewpoint to look is along the -Z axis.
> - https://www.web3d.org/x3d/content/X3dTooltips.html#Viewpoint
>
> - X3D Tooltips: Viewpoint orientation
> -
> https://www.web3d.org/x3d/content/X3dTooltips.html#Viewpoint.orientation
> - *[orientation accessType inputOutput
> <https://www.web3d.org/x3d/content/X3dTooltips.html#accessType>, type
> SFRotation <https://www.web3d.org/x3d/content/X3dTooltips.html#SFRotation>
> CDATA <https://www.web3d.org/x3d/content/X3dTooltips.html#CDATA> "0 0 1 0"]*
> Rotation (axis, angle in radians) of Viewpoint, relative to default -Z
> axis direction in local coordinate system.
> *Hint:* this is orientation _change_ from default direction (0 0 -1).
> *Hint:* complex rotations can be accomplished axis-by-axis using
> parent Transforms.
> *Warning:* for VR/AR/MR/XR users wearing a head-mounted display (HMD),
> animating this field may induce motion sickness.
>
> Improvements always welcome. Have fun with X3D! 🙂
>
>
> all the best, Don
>
> --
>
> Don Brutzman Naval Postgraduate School, Code USW/Br
> brutzman at nps.edu
>
> Watkins 270, MOVES Institute, Monterey CA 93943-5000 USA
> +1.831.656.2149
>
> X3D graphics, virtual worlds, navy robotics
> https://faculty.nps.edu/brutzman
>
>
>
> ------------------------------
> *From:* x3d-public <x3d-public-bounces at web3d.org> on behalf of
> vmarchetti--- via x3d-public <x3d-public at web3d.org>
> *Sent:* Sunday, November 24, 2024 11:39 AM
> *To:* X3D-Public <x3d-public at web3d.org>
> *Cc:* vmarchetti at kshell.com <vmarchetti at kshell.com>
> *Subject:* [x3d-public] X3D, Blender, and which way is up.
>
> I've seen that the email thread starting with the minutes of the last
> Blender-X3D support call had a subthread about the orientation conventions
> that glTF, Blender, and X3D use.
>
> I recommend Michalis' answer at
> https://web3d.org/pipermail/x3d-public_web3d.org/2024-November/020882.html
> .
>
> One confusing aspect is that not only do Blender and glTF disagree about
> which axis is forward, they disagree about what the word "forward" means.
>
> For Blender, the word "forward" refers to the axis that the viewer is
> looking along when they are viewing the front of the model.
> The viewer is translated from the model in the -Y direction, and they are
> looking in the +Y direction. That, +Y, is Blender "forward"
>
> glTF defined "forward" by the model.Looking at the first sentence of the
> of section 3.4 of the glTF specification.
> https://registry.khronos.org/glTF/specs/2.0/glTF-2.0.html#coordinate-system-and-units
> ,
>
> glTF uses a right-handed coordinate system. glTF defines +Y as up, +Z as
> forward, and -X as right; the front of a glTF asset faces +Z.
>
>
>
> glTF says the "forward" refers to the direction the asset, a.k.a. model is
> facing and the spec shows this example picture . In Blender terminology
> this model is oriented "-Z forward"
>
>
>
> [image: boombox.png]
>
> The X3D standard has the same orientation convention as glTF, by virtue of
> the X3D default placement of the Viewpoint at a Z=+10 position and looking
> in the -Z direction. That is why in Blender terminology the X3D convention
> should be considered -Z forward. The code in the Blender extensions server
> was changed to this default for X3D import and export on Nov 4 2024 and
> this default is in release 2.4.0 of this Blender extension.
>
> Vince Marchetti
>
>
>
> _______________________________________________
> 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/20241125/7ce53328/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: boombox.png
Type: image/png
Size: 61294 bytes
Desc: not available
URL: <http://web3d.org/pipermail/x3d-public_web3d.org/attachments/20241125/7ce53328/attachment-0001.png>
More information about the x3d-public
mailing list