[x3d-public] Specification review progress: PNG3, Tangent node, HAnim weekly progress

[John Carlson]

>From Don McCurdy @ Khronos Discord:

Computed with `mesh.calc_tangents`:
https://docs.blender.org/api/current/bpy.types.Mesh.html#bpy.types.Mesh.calc_tangents
and then accessed here i think:
https://github.com/KhronosGroup/glTF-Blender-IO/blob/ed5100ab6c40472b7c3254fddfe0dd0d76d60644/addons/io_scene_gltf2/blender/exp/primitive_extract.py#L1512-L1517

Looks like this is export.  I’m not sure about import.

Anyone is welcome to have at it.  I don’t have a Tangent 3D model yet.

John


On Wed, Jul 9, 2025 at 11:16 AM Michalis Kamburelis <
michalis.kambi at gmail.com> wrote:

> Indeed, taking a look at Blender -> glTF exporter should be helpful.
> It can already generate the 4D tangent vectors in the same format we
> now accept in X3D (with the W component indicating "handedness" in the
> same way for X3D and glTF). Note that you need to select a proper
> checkbox when exporting glTF to have tangents written to the file.
>
> From my side, Tangent development:
>
> - Announced right now on
>
> https://castle-engine.io/wp/2025/07/09/support-for-new-tangent-node-in-x3d-with-vectors-as-4d-better-aligned-with-gltf/
>
> - Castle Game Engine now follows the X3D Tangent spec, with MFVec4f
> for vectors just like for glTF. (hooray, another win for standards --
> glTF, X3D in X_ITE, X3D in Castle Game Engine are now all aligned.)
>
> - Our tools, like Castle Model Viewer, convert glTF to X3D, preserving
> tangent information from glTF into X3D Tangent node. (e.g. convert
> models with tangents generated by Blender or downloaded from Sketchfab
> or obtained from Khronos tests)
>
> - In effect we have some sample X3D files using Tangent node :
> https://github.com/castle-engine/demo-models/tree/master/gltf/tangents
> . These were obtained by converting some glTFs from Sketchfab
> (Creative Commons licenses) -> X3D using CGE, see README there for
> details.
>
> - I also tested our implementation on Khronos testcase
>
> https://github.com/KhronosGroup/glTF-Sample-Assets/tree/main/Models/NormalTangentMirrorTest
>
> Regards,
> Michalis
>
>
> niedz., 6 lip 2025 o 07:29 John Carlson <yottzumm at gmail.com> napisał(a):
> >
> > Now….
> >
> > How do we import/export a Tangent node in Blender?   Can we create a
> .blend file example for it so I can try exporting, then importing?  What
> are the Blender API data structures and methods?
> >
> > I will probably look at glTF examples and intuit.
> >
> > John
> >
> > On Sat, Jul 5, 2025 at 8:23 AM Michalis Kamburelis <
> michalis.kambi at gmail.com> wrote:
> >>
> >> John,
> >>
> >> Thank you, it's cool, every precise question helps me to explain it
> better :)
> >>
> >> To answer that:
> >>
> >> The tangent vectors (i.e., Tangent node contents) only change the way
> >> "final normal vector at given surface point" (designated by N in the
> >> lighting equations) is calculated. That's how they influence the
> >> calculation done by lighting equations, and thus the final rendered
> >> color. They do not interact with colors or geometry in any other way.
> >>
> >> And the way they change this N is that tangent vectors (together with
> >> bitangent and per-face/per-vertex normals) define the "tangent space"
> >> in which the bump mapping (which is the moment where we take
> >> XxxMaterial.normalTexture into account) is calculated.
> >>
> >> There are some resources on the web explaining this, like:
> >> - https://fabiensanglard.net/bumpMapping/
> >> - Wikipedia: https://en.wikipedia.org/wiki/Bump_mapping ,
> >> https://en.wikipedia.org/wiki/Normal_mapping
> >> - hm, ok, and much more, searching for terms like "bump mappping" or
> >> "normal mapping" or "bump mapping 3d opengl" yields a lot of results
> >> :)
> >>
> >> ( One needs some disclaimer when reading them though:
> >>
> >> - There are various ways to define/deal with "handedness" of the
> >> coordinate system. X3D spec and glTF spec match in how they handle it,
> >> so the world 3D coordinate space is right-handed, and the "tangent
> >> space" handedness is determined by the W component of 4D tangents.
> >> Various articles on the web don't deal with this detail, or they deal
> >> with it differently.
> >>
> >> - The terminology distinction "bump mapping" vs "normal mapping" is
> >> honored/ignored differently in articles. (Admittedly, in my own
> >> writing sometimes too :) ) To be precise, "bump mapping" is more
> >> general idea, and "normal mapping" is a special case of it (when you
> >> do it using normal maps), and "normal mapping in tangent space" is
> >> even more special case of it. But...various resources simplify these
> >> things, because in practice "normal mapping in tangent space" is the
> >> most common way to do bump mapping in real-time 3D graphics (now).
> >> )
> >>
> >> I hope any of this helps you and Don and anyone else to make sense of
> >> it :) Feel free to incorporate any of this in spec prose.
> >>
> >> Regards,
> >> Michalis
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://web3d.org/pipermail/x3d-public_web3d.org/attachments/20250709/189592b6/attachment-0001.html>