[x3d-public] 3D graphics, vertex attributes Tangent index and vector

Fri May 23 15:05:46 PDT 2025

Tangent planes seem more likely than tangent vectors, at least in 3D.

Just a naming convention.

John
On Fri, May 23, 2025 at 4:09 PM Michalis Kamburelis <
michalis.kambi at gmail.com> wrote:

> Don: I don't think that this decision (to never support Tangent node
> in X3D), is correct. I would actually say we should add Tangent to
> X3D.
>
> Let me present a clear argument "why". I copied some of the reasoning
> from mail below to
> https://github.com/michaliskambi/x3d-tests/wiki/Tangent-node-in-X3D ,
> to have a future reference.
>
> 1. Current state: As of now, both X_ITE and Castle Game Engine have
> extensions to define Tangent node.
>
>     So we have invented the "Tangent" node independently in 2
> implementations and, predictably, they are unfortunately
> similar-but-not-exactly the same. Links from this thread, where you
> can notice a difference:
>
>      - https://create3000.github.io/x_ite/components/rendering/tangent/ ,
>      -
> https://castle-engine.io/x3d_implementation_rendering_extensions.php#section_ext_tangent
>
>     Also glTF format has them. See
> https://registry.khronos.org/glTF/specs/2.0/glTF-2.0.html#meshes-overview
> , information about "TANGENT".
>
>     ( Holger, if you read this: I think I will change Castle Game
> Engine's Tangent node in the future, to match better both X_ITE and
> glTF, thus we will have matching definitions with X_ITE. Your approach
> is better, Tangent.vector should be 4D. )
>
> 2. Tangents *do not* just repeat the same information as normal
> vectors, DRY principle doesn't apply here. I.e. the tangent vectors
> are *not* a replacement for knowing normal vectors (nor is the other
> way around).
>
>      And just because the model provides normal vectors -- doesn't
> mean that tangent vectors are already precisely 100% determined.
>
>      There are multiple similar algorithms to determine tangent
> vectors. The normal vector alone doesn't yet define the tangent ->
> because in a 3D space, if you have a normal vector -> you still have
> infinite possible values for a vector "orthogonal to normal". So
> tangent vectors are auto-calculated (when not explicitly provided) to
> match the normal vectors, but *also* to match the texture coordinates
> used for normal maps. There are edge-cases when it is just not obvious
> what should be the "perfect tangent" so various implementations
> (MikkTSpace is just one example) can do different things.
>
> 3. Ultimately, the "perfect tangent value" is the one that was used to
> bake the normalmaps. If the authoring tool (like Blender) that you
> used to make normalmaps used a bit different algorithm to calculate
> tangents than your renderer -> then the rendering of normalmaps will
> be slightly "off".
>
>     There's a nice screenshot in
>
> https://gamedev.stackexchange.com/questions/146855/how-do-you-compute-the-tangent-space-vectors-with-normals-given-in-the-mesh
> , scroll to "There are multiple ways to generate tangent spaces for a
> mesh, and not all of them agree on the result.". The screenshot there
> shows subtle problems that result from having authoring tool and
> renderer calculate a *bit* different tangent vectors.
>
>     So the perfect thing to do, from a renderer, is to *not* calculate
> the tangent vectors, instead take the tangent vectors as provided by
> the authoring tool. That exactly why glTF has tangents, and why X_ITE
> and Castle Game Engine support them too -- we don't just discard
> tangent values from glTF (only to auto-calculate them), we prefer to
> take tangent values recorded in the 3D model, because this makes
> really good rendering.
>
> See https://github.com/michaliskambi/x3d-tests/wiki/Tangent-node-in-X3D
> , which I just created -- it mostly copies what I said above.
>
> I hope this helps. I'll be happy to explain it better on next Web3d
> Friday teleconference, if you're not tired of my talking :) I can show
> how the normalmaps are actually created.
>
> Regards,
> Michalis
>
>
> pt., 23 maj 2025 o 21:03 Brutzman, Donald (Don) (CIV) via x3d-public
> <x3d-public at web3d.org> napisał(a):
> >
> > The X3D group has looked closely at Tangent options in the past and
> decided not to support it.  Instead we use Normal nodes, as ever.
> >
> > Of geometric note is that they are literally orthogonal and thus either
> representation can be used to the other.
> >
> > You will not find Tangent node in the X3D specification.
> >
> > X3D 4.1 draft Architecture, Annex Z (informative) Version content
> >
> https://www.web3d.org/specifications/X3Dv4Draft/ISO-IEC19775-1v4.1-CD//Part01/versionContent.html
> >
> >
> > Note that the X_ITE documentation specifically notes that support in
> that application is nonstandard.
> >
> >
> > SFNode [in, out] tangent NULL [Tangent] non-standard]
> >
> https://create3000.github.io/x_ite/components/geometry3d/indexedfaceset/#fields-tangent
> >
> >
> > There is no issue with codebases adding functionality that might help
> their capabilities - the X in X3D is extensible, after all - but anyone
> putting a Tangent definition into an X3D model file is creating
> nonstandard, invalid, noninteroperable content.
> >
> > Also please be aware that there is a serious design principle behind
> this decision: DRY.  Thus, a statement I don't expect to repeat: we do not
> intend to consider Tangent support in future X3D specifications.
> >
> > Wikipedia: Don't repeat yourself
> > https://en.wikipedia.org/wiki/Don't_repeat_yourself
> >
> >
> > And some "dry humor" on that page:
> >
> > The opposing view to DRY is called WET, a backronym commonly taken to
> stand for write everything twice (alternatively write every time, we enjoy
> typing or waste everyone's time).
> >
> >
> > Recommendations:
> >
> > if your model data is starting with tangent values (for whatever reason)
> then convert them to normal vectors, and
> > stick with standard X3D for best results.
> >
> >
> > Have fun with normal 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 Joe D
> Williams via x3d-public <x3d-public at web3d.org>
> > Sent: Sunday, May 18, 2025 5:03 PM
> > To: John Carlson <yottzumm at gmail.com>
> > Cc: Joe D Williams <joedwil at earthlink.net>; Extensible 3D (X3D)
> Graphics public discussion <x3d-public at web3d.org>
> > Subject: Re: [x3d-public] 3D graphics, vertex attributes Tangent index
> and vector
> >
> > ok, seems like from what I can study so far,
> >
> > the first use of Tangent vector is authortime production of
> >
> > normalized Normal vector for vertex or face.
> >
> > The  author is looking for relief effects (bumpmapping)
> >
> > that are produced by Tangent vector.
> >
> > This process (MikkTSpace algorithm) would fill Normal vector.
> >
> > This process of generating Normal vector space is called baking.
> >
> >
> >
> > For Level5 runtime,
> >
> > If Normal vector not null then used directly.
> >
> > If Normal vector null, then normals auto generated by creaseAngle
> >
> > If Tangent vector null, then nothing.
> >
> > If Level 5 and Tangent vector not null then possible relief effects
> >
> >  using length of Tangent vector.
> >
> >
> >
> > No "bake" in runtime.
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> > For runtime, the generated Normal vector is used and
> >
> > if Tangent vector is not null, for the author to produce relief effects
> >
> > using length of Tangent vector.
> >
> > Is that true?
> >
> >
> >
> >
> >
> >
> >
> > .
> >
> > For runtime
> >
> > -----Original Message-----
> > From: John Carlson <yottzumm at gmail.com>
> > Sent: May 17, 2025 1:15 PM
> > To: Joe D Williams <joedwil at earthlink.net>
> > Cc: Extensible 3D (X3D) Graphics public discussion <x3d-public at web3d.org
> >
> > Subject: Re: [x3d-public] 3D graphics, vertex attributes Tangent index
> and vector
> >
> >
> >
> > Please review Michalis’ response on this thread about tangentIndex and
> tangentPerVertex.  I believe the corresponding normal fields apply.
> >
> > Level 5? That depends on updated standard.
> >
> > On Sat, May 17, 2025 at 2:04 PM Joe D Williams <joedwil at earthlink.net>
> wrote:
> >
> > * must have tangentIndex (unless tangentIndex same as normalIndex).
> >
> > * If no tangentIndex, coordIndex used?
> >
> > * tangentPerVertex? Can tangent be applied to vertex or face (like
> normal and color)?
> >
> > * must have tangent vector (may be auto filled).
> >
> > * Level 5 required since level 4 does not have tangent.
> >
> > Thanks,
> >
> > Joe
> >
> >
> >
> > -----Original Message-----
> > From: John Carlson <yottzumm at gmail.com>
> > Sent: May 16, 2025 10:32 PM
> > To: Extensible 3D (X3D) Graphics public discussion <x3d-public at web3d.org
> >
> > Cc: Joe D Williams <joedwil at earthlink.net>
> > Subject: Re: [x3d-public] 3D graphics, vertex attributes Tangent index
> and vector
> >
> >
> >
> > Joe,
> >
> > X_ITE IFS:
> >
> > https://create3000.github.io/x_ite/components/geometry3d/indexedfaceset/
> >
> > Summary, only tangent field.
> > John
> >
> > On Fri, May 16, 2025 at 8:59 PM Joe D Williams via x3d-public <
> x3d-public at web3d.org> wrote:
> >
> > like  for normals, is there a tangentPerVertex true/false where tangent
> value
> >
> > is associated with vertex or face?
> >
> > If tangentIndex null, is coordIndex used?
> >
> > If Tangent vector null tangents are generated after Normal vector?
> >
> >
> >
> > Thanks,
> >
> > Joe
> >
> >
> >
> >
> >
> >
> >
> > -----Original Message-----
> > From: Extensible 3D (X3D) Graphics public discussion <
> x3d-public at web3d.org>
> > Sent: May 15, 2025 6:19 AM
> > To: X3D <x3d-public at web3d.org>
> > Cc: Holger Seelig <holger.seelig at yahoo.de>
> > Subject: Re: [x3d-public] 3D graphics, vertex attributes Tangent index
> and vector
> >
> >
> >
> > There are also glTF tests for tangents:
> >
> > Geometry with tangent node:
> >
> https://create3000.github.io/x_ite/laboratory/gltf-sample-viewer/?url=https://raw.githubusercontent.com/KhronosGroup/glTF-Sample-Models/master/2.0/NormalTangentMirrorTest/glTF/NormalTangentMirrorTest.gltf
> >
> > Geometry without tangent node (auto generated tangents using MikkTSpace
> algorithm):
> >
> https://create3000.github.io/x_ite/laboratory/gltf-sample-viewer/?url=https://raw.githubusercontent.com/KhronosGroup/glTF-Sample-Models/master/2.0/NormalTangentTest/glTF/NormalTangentTest.gltf
> >
> > The reflection in all half spheres should look the same.
> > Turn on IBL (Image Based Lighting).
> >
> > Best regards,
> > Holger
> >
> > --
> > Holger Seelig
> > Leipzig, Germany
> >
> > holger.seelig at yahoo.de
> > https://create3000.github.io/x_ite/
> > https://patreon.com/X_ITE
> >
> >
> > Am 15.05.2025 um 10:35 schrieb Holger Seelig via x3d-public <
> x3d-public at web3d.org>:
> > All nodes which have a normal field also have a tangent field in X_ITE,
> because tangents are very important for PhysicalMaterial to render
> correctly. Because tangents always come in conjunction with normals,
> normalPerVertex field is also used for tangents.
> >
> > If there is no Tangent node, the MikkTSpace algorithm is used to
> generate tangent vectors, as need by the glTF specification.
> >
> > If there is a Tangent node and the vector field is empty, results are
> undefined.
> >
> > Tangent node:
> > https://create3000.github.io/x_ite/components/rendering/tangent/
> >
> > IndexedFaceSet with tangent field:
> >
> https://create3000.github.io/x_ite/components/geometry3d/indexedfaceset/#fields-tangent
> >
> > Best regards,
> > Holger
> >
> > --
> > Holger Seelig
> > Leipzig, Germany
> >
> > holger.seelig at yahoo.de
> > https://create3000.github.io/x_ite/
> > https://patreon.com/X_ITE
> >
> >
> > Am 15.05.2025 um 10:24 schrieb Michalis Kamburelis <
> michalis.kambi at gmail.com>:
> > I don't know about Tangent in X_ITE...
> >
> > But in Castle Game Engine we have a similar extension -- Tangent node (
> https://castle-engine.io/x3d_implementation_rendering_extensions.php ).
> Looks like we developed it independently from X_ITE :) It certainly makes
> sense to have a node that carry this information, to have perfect
> rendering. (Tangents can be auto-calculated, but there are various details
> in the auto-calculation algorithm that various renderers may do a bit
> differently. When the 3D author used "baking" to make a normalmap texture,
> it's better to use exact same tangents as known by the 3D authoring tool.)
> >
> > At least for Castle Game Engine, the answer is: if the Tangent node is
> not present, but we need tangents (e.g. for bump mapping), then the tangent
> vectors are calculated to match the normal vectors. So there's no
> additional tangentPerVertex or tangentIndex, they would not make sense. By
> knowing the normal vectors, by know all we need to generate reasonable
> tangent vectors.
> >
> > See https://castle-engine.io/x3d_implementation_rendering_extensions.php
> for more details.
> >
> > Regards,
> > Michalis
> >
> > czw., 15 maj 2025 o 07:43 John Carlson via x3d-public <
> x3d-public at web3d.org> napisał(a):
> >
> > Note, it’s Tangent vector=… not Tangent point=…
> >
> > John
> >
> > On Thu, May 15, 2025 at 12:38 AM John Carlson <yottzumm at gmail.com>
> wrote:
> >
> > I’d ask Holger, cc’d
> >
> > I’ve not used Tangent, myself.
> >
> > X_ITE Supports up to Level 4, so I guess you’re proposing a new level 5
> for the Geometry 3D component?
> >
> > Look at X_ITE IFS:
> >
> > <IMG_0516.png>
> >
> > Maybe check Holger’s create3000 Library for examples?
> >
> > To retrieve examples, last time I tried:
> >
> > git clone https://github.com/create3000/Library
> > Online at:
> >
> > https://github.com/create3000/Library
> > Let’s get v2 Joe Kick out the door, check v2 finger joints, and make
> sure all joints have converted to v2.
> >
> > John
> >
> > On Wed, May 14, 2025 at 10:45 PM Joe D Williams <joedwil at earthlink.net>
> wrote:
> >
> >
> >
> > Main questions, in IFS Level5:
> > Is a default tangent generated if Tangent point="" field empty?
> > tangentPerVertex? assigned to vertex or face?
> > tangentIndex? If no tangentIndex is coordIndex used?
> >
> >
> >
> > Thanks,
> > Joe
> >
> >
> >
> >
> >
> >
> >
> >
> >
> > -----Original Message-----
> > From: John Carlson <yottzumm at gmail.com>
> > Sent: May 14, 2025 3:00 PM
> > To: Extensible 3D (X3D) Graphics public discussion <x3d-public at web3d.org
> >
> > Cc: Joe D Williams <joedwil at earthlink.net>
> > Subject: Re: [x3d-public] 3D graphics, vertex attributes
> >
> >
> >
> > Joe,
> >
> > X_ITE has a tangent node:
> >
> > Tangent | X_ITE X3D Browser
> >
> > Enjoy!
> >
> > John
> >
> > On Wed, May 14, 2025 at 4:46 PM Joe D Williams via x3d-public <
> x3d-public at web3d.org> wrote:
> >
> > From this:
> >
> >
> https://docs.google.com/document/d/1hCetnGIkut_dyt4mwkDSTNiLTbrVg1jLo_EArE74DRg/edit?tab=t.0
> >
> >
> > In 3D graphics, vertex attributes define per-vertex data necessary
> > for rendering, animation, shading, and so on.
> > These attributes include geometric properties such as
> > position, normal, tangents, texture mapping information (UV coordinates),
> > shading data (vertex colors), and
> > deformation information (skinning weights and blend shapes).
> >
> >
> > In this discussion of vertex attributes, the only one I don't recall
> > seeing in X3D is tangent?
> >
> > Associated with this work:
> >
> https://docs.google.com/spreadsheets/d/18d1qy4jx-d_kLsam-JQAzig23w7dTau0Moe7M32lVvk/edit?gid=0#gid=0
> >
> >
> > Thanks,
> > Joe
> >
> >
> > _______________________________________________
> > 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
> >
> >
> >
> > _______________________________________________
> > 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/20250523/e072e297/attachment-0001.html>