[x3d-public] Add quaternion utility methods to X3D SFRotation programming-language standards?
John Carlson
yottzumm at gmail.com
Tue Jul 23 19:51:06 PDT 2024
On Tue, Jul 23, 2024 at 7:32 PM Joe D Williams <joedwil at earthlink.net>
wrote:
> > Joe, I have exported HAnim as glTF JSON.
>
>
>
> I have not seen that. gltf has some 'standard' asset types and I was
> looking for that stuffs.
>
What were you looking for, I can search to see if the Blender glTF exporter
supports it?
>
> > I have also used the X3D blender importer and glTF exporter. We’ve gone
> over this.
>
>
>
> Sorry I missed the result.
>
> I have seen some simple examples of gltf assets that might be extended to
> handle hanim and sure they are logical but I still am wondering if it would
> be best to try and style some 'standard' gltf assets to look more like the
> special category of app that is of x3d hanim and skeleton-driven animations
> in general.
>
Do you remember the 146 joint animations that showed up in
castle-game-engine, and there was no easy way to select them all, we had to
click each one? That was glTF. We had a discussion with Michalis about
it. Hopefully he’s clued in for when he produces glTF from X3D. It blew
claude.ai’s brain asking to do parallel animations across multiple
keyframes in Blender Python. I’m just doing one ROUTE at a time myself.
Perhaps approach from a multiple interpolator framework would be better?
How do you convert multiple interpolators into keyframes, if not one by one
and ROUTE-based? Does the standard cover this? Imagine that I’m *not*
simulating an event system, I’m converting from one event system to
another. Understand that, AFAIK, Blender bone animation works the way I’m
doing it.
It’s not my responsibility to make the Blender glTF exporter export HAnim.
I don’t want to touch glTF. I don’t want to become a glTF expert. That’s
for the people interested in hexcode and databases. I want to create cool
Mathematical graphics, do *program* synthesis, and do multiuser systems. I
realize that equation visualization is not a strong point of X3D. Yes,
glTF was one thing I used for API by example (program synthesis). I’m also
interested in X3D API by example. I don’t even want to break up my
equations into triangles or quads, indeed, I’ve raytraced my surfaces with
Newton’s Bisection originally. Later investigation showed that breaking
things into triangles might give better results, and that has been my path
after poor results with Mathematica, years ago. If someone can figure out
how to find internal ray-surface intersections of self-intersecting
spherical equation surfaces without breaking up into triangles or quads and
do it in real-time, I’m all ears.
No this is not a “Hey, Mikey” moment yet again.
>
> LOA1 does not show much animation. Is loa4 too many clicks, or what? Trim
> down the loa3 joekick skeleton to lower loa and you lose a lot but you
> could still hook up the skin by moving the skincoordindex and weights to
> remaining parent Joint. I still think best to have loa4 working then drop
> back to lower loa in steps. Whatever the owner, I still say start with loa4
> and take the time to make all the measurements, then only animate the
> joints you need for your behavior choices.
>
LOA4 takes too long to import and load in Blender Python, not to mention
export. It makes programming a chore instead of enjoyable. One looks for
speedier alternatives. LOA4 is not the problem per se. We’re not going to
finish LOA4 if we continue this path. Simple examples, aka unit testing is
needed at this point. We’re not even to the point of QA, and you want to
do QA testing. We’re still in early development. Some small tests
succeed, but not with the same code. So we’re trying to suss out the right
code to install in the importer. Even going into the admin shell to copy
the addon to the Blender folder (and you have to remember to do it each
time you test) is painful. Editing in admin mode is a non-starter.
At this point, I’m nearly ready to hand the loader and importer over to the
stylesheet folks.
>
>
> > Examples without geometry and animation don’t help solving the
> animation issue, unless one is trying to solve Blender animation.
>
>
>
> The best would be to transport easily between x3d and blender. I think
> blender will do all the x3d stuff like easein/out and maybe even animation
> using spline curves directly.
>
The problem with Blender is that it uses local transforms for empties, and
HAnim uses centers relative to the origin. Then one throws animation on
top of that. I’ve gotten animation through blender, and simple animations
to work. If we could use Joint centers for Blender bone ends, we’d be lot
farther along. The problem is with attaching objects (empties, not
geometry, AFAIK) to bones and armature. Single skin is the easy case.
I’ve seen examples of multiple geometries connected to armature. This does
not show up as connections to bones in the Blender GUI. I’m not sure if
I’ve seen Blender animation of multiple geometry HAnim that actually
works. I think it’s possible to connect multiple meshes to an armature.
Maybe one can attach an axe to a bone. I’m not really seeing connecting
empties to bones. Maybe worth a try.
>
>
>
> -----Original Message-----
> From: John Carlson <yottzumm at gmail.com>
> Sent: Jul 23, 2024 11:47 AM
> To: Joe D Williams <joedwil at earthlink.net>
> Cc: Andreas Plesch <andreasplesch at gmail.com>, Extensible 3D (X3D)
> Graphics public discussion <x3d-public at web3d.org>, GPU Group <
> gpugroup at gmail.com>, Johannes Behr <johannes.behr at threedy.io>, Subscribed
> <subscribed at noreply.github.com>, x3dom/x3dom <x3dom at noreply.github.com>
> Subject: Re: [x3d-public] Add quaternion utility methods to X3D SFRotation
> programming-language standards?
>
>
> Joe, I have exported HAnim as glTF JSON. That’s what the Gramps huff with
> alpha blend mode was all about. But the source wasn’t X3D, it was Katy.
> There’s probably many Blender exportable assets that can be rendered as
> glTF. Ask Doug.
>
> I have also used the X3D blender importer and glTF exporter. We’ve gone
> over this. Please try the Blender import/export for yourself before
> assigning tasks. There’s an annoying thing where the conversion of the
> animation creates separate animations in blender export when viewing the
> glTF. AFAIK, glTF is not a runtime format, grouping animations is up to
> the code in addition to glTF (but I think that Michalis would take issue
> with this); I believe I saw glTF animations with auto start. There might
> be a Blender export option. Perhaps work needs to be done to improve HAnim
> animation import. Back to the BVH thing and animation concatenation, and
> efforts on the Blender loader. It will take extra effort to identify which
> animations to convert when all are provided. I believe I have an example
> with just one animation. Joe Kick is a great example. I had previously
> worked on roundtripping animation. I’m thinking any effort on glTF will
> produce individual animations for each joint. Playing the animations
> together is the key. X3D to X3D is half decent, with stray geometry. The
> animation of Jin in Blender itself is awful (Joe Kick has worked), but the
> loader had shown we can create animations that don’t revolve around the
> origin. There’s potential to ignore any Blender animation and go straight
> to glTF. If we have a runtime that can play numerous glTF animations at
> the same time, there might be further progress. But expecting me to click
> 146 animations to turn all of them on is a non-starter. Especially
> multiple times. If someone has Blender python code for grouped animations
> (I’ll do a search), fess up.
>
> Examples without geometry and animation don’t help solving the animation
> issue, unless one is trying to solve Blender animation.
>
> Again, a call for an LOA1 animation example. We can try without
> geometry. LOA4 is a time sink at this point. The tendency for me append
> LOA4 animation to LOA1 skeleton, and then write Java for cleanup.
>
>
> If needed, i can provide compressed X3D JSON examples. I’m not convinced
> converting X3D to glTF is going to be smaller. Unreadable glTF formats,
> yes. Do we want unreadable formats?
>
> I believe there are others working on X3D to glTF. Perhaps Blender is a
> stumbling block.
>
> My tendency is to work on HAnim software any format can leverage.
>
> Do you want some beliefware? That’s what I woke up to this morning.
>
> John
>
> On Tue, Jul 23, 2024 at 10:49 AM Joe D Williams <joedwil at earthlink.net>
> wrote:
>
>> yes, this is a jsonized version of the xml and classic, though looking
>> more like classic by showing container type string, almost like the wrapper
>> tags, but still not having the mindset of the gltf asset composition.
>>
>> Seems like the 'gltf' approach would break it up with an asset for the
>> hierarchy and then maybe separately the geometry and another for the
>> animations.
>>
>>
>>
>> What I am looking for is, how to do a 'gltf' json encoding that follows
>> some existing 'standard' asset types of json. Has anyone exported HAnim to
>> json using the 'standard' json asset styles? I haven't but think those
>> asset files would look lots different than what we see here.
>>
>>
>>
>> The use case would be to compose and test using x3d user code, export to
>> json gltf asset types for delivery. Another would be to import a jsonized
>> hanim asset and convert to edit x3d hanim user code. Or just edit the
>> json(?) with access to the hanim nodes and hierarchy (Joint, Segment, Site)
>> to make it more relatable to HAnim.
>>
>>
>>
>> So, I appreciate this transcoding to show that the json can certainly
>> work to transport x3d hanim, but I don't see a direct use of it because it
>> is no more compact or easier to read or edit than xml or classic. Finally,
>> this does not really match the goals of the gltf, which is to deliver a
>> compact set of assets aimed at runtime, not authortime.
>>
>>
>>
>> some stuff from the metaverse (more x3d members should get involved with
>> the metaverse forum.
>>
>> Some past info from the forum dealing with gltf:
>>
>>
>>
>> 2023(ish) gltf development
>> physical materials
>> Tooling (usd <-> gltf)
>> Interactivity
>> Physics
>> Gltf Tools
>> Compression
>> iOS
>> Dampole Viewer
>> Asset Validator
>> Blender
>> gltf Extensions
>> specular anisotropy
>> WebP
>> GPU Instancing
>> meshopt compression
>>
>> Physics
>> collision shapes
>> Rigid Bodies
>> Blender Importer/exporter
>>
>>
>>
>> Overall, my impression is that gltf is still developing asset definitions
>> for complex hierarchies like needed for hanim and for containing different
>> types of animation data.
>>
>> And what is a Blendshape?
>>
>>
>>
>> To me, we just need to determine best practice for how x3d hanim user
>> code gets gltfized using 'standard' asset types or if some extended more
>> hanim looking assets need to be styled.
>>
>>
>>
>> So, John and All, Great and very significant work showing the versatility
>> of x3d.
>>
>>
>>
>> Thanks Again and don't forget to suspend your disbelief ...
>>
>> All Best,
>>
>> Joe
>>
>>
>>
>> -----Original Message-----
>> From: John Carlson <yottzumm at gmail.com>
>> Sent: Jul 22, 2024 1:57 PM
>> To: Joe D Williams <joedwil at earthlink.net>
>> Cc: Andreas Plesch <andreasplesch at gmail.com>, Extensible 3D (X3D)
>> Graphics public discussion <x3d-public at web3d.org>, Johannes Behr <
>> johannes.behr at threedy.io>, Subscribed <subscribed at noreply.github.com>,
>> x3dom/x3dom <x3dom at noreply.github.com>
>> Subject: Re: [x3d-public] Add quaternion utility methods to X3D
>> SFRotation programming-language standards?
>>
>>
>> There are plenty of HAnim JSON examples if glTF wants to glom on!
>>
>> See HumanoidAnimation in the X3D archives.
>>
>> E.g.:
>>
>> https://www.web3d.org/x3d/content/examples/HumanoidAnimation/Characters/JinLOA4.json
>>
>> If you want JSON from X3D, there’s JSON button in X3D-Edit.
>>
>> X3DJSAIL also offers conversion from XML to JSON.
>>
>> As does Holger’s x3d-tidy! (use .x3dj).
>>
>> Sunrize and X_ITE offer viewing of HAnim JSON.
>>
>> John
>>
>> On Mon, Jul 22, 2024 at 1:49 PM Joe D Williams <joedwil at earthlink.net>
>> wrote:
>>
>>> > My understanding is that there’s a desire to convert between
>>> axis-angle (SFRotation) and quaternions.
>>>
>>>
>>>
>>> Please, when you think of legal gltf quaternion data, you must think of
>>> _unit_ quaternions.
>>>
>>>
>>>
>>> This conversion is not a desire, simply that the gltf data and other
>>> transport forms use unit quaternions. Thus, if I understand, for example to
>>> export native x3d animation data to gltf we would need to convert, and to
>>> import we either need to convert to x3d native axis-angle or just use unit
>>> quaternions under the covers. But if we wanted to import for working with
>>> (editing/refining) the data, such as for orientation interpolators, would
>>> need to convert.
>>>
>>>
>>>
>>> So, maybe like the SAI has 3x3 and 4x4 matrix operations, we need a way
>>> to convert a single or sequence of axis-angle to quaternion for export and
>>> the other way for import. I don't think it is useful to allow unit
>>> quaternions in x3d user code for rotations or orientations (just hard to
>>> work with handed editing) but it seems like it is unavoidable if we want
>>> import/export to gltf. Incidentally, the required gltf forms for skeletal
>>> hierarchies and animations are still being worked on by a metaverse forum
>>> group for interoperable assets.
>>>
>>>
>>>
>>> Finally, looking at the gltf forms for skeleton transform hierarchies
>>> and animations, it seems like x3d should develop a set of hanim gltf assets
>>> specifically for hanim that use x3d hanim keywords and structures. From
>>> what I have seen, which is not much, the gltf useful for hanim content is
>>> very general and abstract to the point of not being directly relatable to
>>> hanim user code.
>>>
>>>
>>>
>>> Thanks for some deep thoughts on this,
>>>
>>> Joe
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>> -----Original Message-----
>>> From: Extensible 3D (X3D) Graphics public discussion <
>>> x3d-public at web3d.org>
>>> Sent: Jul 21, 2024 7:57 PM
>>> To: Extensible 3D (X3D) Graphics public discussion <x3d-public at web3d.org
>>> >
>>> Cc: John Carlson <yottzumm at gmail.com>, x3dom/x3dom <
>>> x3dom at noreply.github.com>, Subscribed <subscribed at noreply.github.com>,
>>> Johannes Behr <johannes.behr at threedy.io>, Andreas Plesch <
>>> andreasplesch at gmail.com>
>>> Subject: Re: [x3d-public] Add quaternion utility methods to X3D
>>> SFRotation programming-language standards?
>>>
>>>
>>> My understanding is that there’s a desire to convert between axis-angle
>>> (SFRotation) and quaternions. X3DOM has only supported quaternions in the
>>> past, looks like there’s progress! Cross-platform X3D is highly desirable.
>>>
>>> Quaternion conversion might be useful to use in Blender export with
>>> X3DPSAIL (x3dv.py), but Blender provides utility libraries already, AFAIK.
>>> Mostly, I am setting rotations in OrientationInterpolators for HAnim and
>>> other animation. Rotations also come into play for orienting coordinate
>>> axes. I don’t see quaternions useful there.
>>>
>>>
>>> For Blender import, working on importing animations for transform-like
>>> nodes should be paramount, especially in XML, and may require Blender
>>> modifications. If SFRotation methods can help with that, great. AFAIK,
>>> though, there’s no SFRotation class in the importer yet.
>>>
>>> I stop here talking about quaternions. Begin rant. I’ll peek at the
>>> SAI WD now.
>>>
>>> For X3DPSAIL, there’s more important considerations like automated
>>> DEF/USE (not for static scenegraphs, but for programs that generate many
>>> scenegraphs, with a wide variety of functions), X3D JSON export and
>>> integrating container field [ sic ] output in my mind. These should
>>> be priorities over quaternions. We both probably understand the pain of
>>> X3D JSON; the good news is, we have pretty reasonable stuff to move between
>>> that and XML or DOM. If someone tells me exactly how to read XML, JSON or
>>> VRML into an X3DPSAIL scenegraph, I will start looking at how to use
>>> X3DPSAIL for Blender import. One possibility is providing Python import to
>>> Blender! I totally understand that! Providing X3D to Blender Python
>>> conversions is critical for that. See:
>>> https://github.com/coderextreme/X3DJSONLD/blob/master/blend/localfinal.py for
>>> x3d blender python loader, which could be the start of a stylesheet.
>>>
>>>
>>>
>>> Addition of standard vector/matrix multiplication/inversion operations
>>> to X3DJSAIL should be a priority over quaternions. I don’t know if
>>> standard matrix operations are part of SAI or not. Since we support matrix
>>> and vector fields in X3DJSAIL, operations on these seem obvious. If they
>>> are in SAI, X3DJSAIL was lacking as a couple of months ago. I used
>>> vecmath.jar. I’m as embarrassed as you are! I realize that’s not the
>>> purpose of X3DJSAIL. A pointer to the documentation doesn’t help here.
>>>
>>> More priorities for X3DJSAIL: Comprehensive ways to select all instances
>>> of a given class in a scenegraph. Ways to filter on nodes with an
>>> attribute (DEF) present and value (ala HTML DOM and CSS), realizing
>>> libraries like JXPath exist (is there something better?). Convenience
>>> methods for removing nodes and statements in a set or two from the
>>> scenegraph would allow me to deprecate my code. Understood that this can
>>> probably be done with DOM. I may wish to load JavaScript into an X3DJSAIL
>>> scenegraph as subclasses of X3DJSAIL through class shims (see my
>>> X3Dautoclass.js files for graaljs).
>>>
>>> Primarily, I am working on HAnim, and making large models feasible in
>>> X3DJSAIL Java and CGE’s x3d Pascal library (X3dToPascal.xslt?) would be
>>> nice! Promoting a fluent interface in SAI (setter and adder functions and
>>> methods that return a pointer or reference to primary object being worked
>>> on) would help with the latter.
>>>
>>> Current Pascal generator:
>>>
>>>
>>> https://github.com/coderextreme/X3DJSONLD/blob/master/src/main/node/PascalSerializer.js
>>>
>>> These two HAnim .inc files have issues.
>>>
>>> https://github.com/coderextreme/pascalsai/tree/main/code
>>> If we had a pattern for breaking up a large Java code tree into separate
>>> methods , or lengthy Pascal code into “slices” or separate methods, cool.
>>> Maybe some online research would reveal something. Maybe we have an
>>> existing stylesheet, like the ROUTE visualizer? Hopefully, at some point,
>>> I will get some inspiration to solve this problem. It just seems like a
>>> command-line parameter that I haven’t looked up.
>>>
>>> John
>>>
>>> On Sun, Jul 21, 2024 at 5:01 PM Brutzman, Donald (Don) (CIV) via
>>> x3d-public <x3d-public at web3d.org> wrote:
>>>
>>>
>>>> Interesting to see in the online issue discussion below that X3DOM
>>>> includes quaternion methods.
>>>>
>>>> The SFRotation representation has some similarity to quaternions, but
>>>> they are of course mathematically distinct. Quaternions have long been a
>>>> suggested alternative for X3D, with SFRotation remaining the primary
>>>> representation since it is much more reviewable and editable (among other
>>>> reasons). Nevertheless, for programmers, additional flexibility can be
>>>> considered without conflict. This may encourage X3D API usage with other
>>>> 3D programming libraries.
>>>>
>>>> Please consider and advise whether quaternion-based accessor (get and
>>>> set) utility methods ought to be added to the X3D specifications for
>>>> SFRotation data type. Specifically they might become required in X3D Scene
>>>> Access Interface (SAI), along with all of the future corresponding version
>>>> 4.0 revisions to X3D programming-language bindings (ECMAScript/JavaScript,
>>>> Java, C/C++/C#, Python).
>>>>
>>>>
>>>> - X3D Scene Access Interface (SAI), version 4.0 draft
>>>> - https://web3d.org/specifications/X3Dv4Draft/ISO-IEC19775-2v4.0-WD1
>>>>
>>>>
>>>> If so, please list what utility methods seem most broadly valuable. Based
>>>> on responses, Dick and I will work on crafting a Mantis issue and draft
>>>> specification prose for further review.
>>>>
>>>>
>>>> At that point, once reasonably well defined, am also willing to add
>>>> convenience methods in Java X3DJSAIL and Python X3DPSAIL (x3d.py) libraries
>>>> for testing. Contributions of exemplar source code for those libraries
>>>> will also be welcome, if anyone wants.
>>>>
>>>> Further interesting is the possibility that we are not only supporting
>>>> future X3D programmers, but potentially also supporting Metaverse
>>>> requirements for 3D Web Interoperability.
>>>>
>>>> For those wanting additional background, more information about X3D
>>>> data types and SFRotation (plus further links) can be found at
>>>>
>>>>
>>>> - X3D Tooltips 4.0: field types
>>>> - https://www.web3d.org/x3d/tooltips/X3dTooltips.html#type
>>>> -
>>>> https://www.web3d.org/x3d/tooltips/X3dTooltips.html#FieldTypesTable
>>>> - https://www.web3d.org/x3d/tooltips/X3dTooltips.html#SFRotation
>>>>
>>>>
>>>> Thanks in advance for considering the possibilities and all feedback.
>>>>
>>>>
>>>>
>>>> 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:* Andreas Plesch <notifications at github.com>
>>>> *Sent:* Wednesday, June 26, 2024 11:03 AM
>>>> *To:* x3dom/x3dom <x3dom at noreply.github.com>
>>>> *Cc:* Subscribed <subscribed at noreply.github.com>
>>>> *Subject:* Re: [x3dom/x3dom] SFMatrix4f() documentation for
>>>> setRotation and translation (Issue #1322)
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> SFRotation field methods are also available but less well tested.
>>>>
>>>> SAI code may look like this:
>>>>
>>>> SFRot90 = x3dom.fields.SFRotation(1, 0, 0, 1.57);
>>>> locR = SFRot90.multiVec(hitP);
>>>> trans = new x3dom.fields.SFVec3f(440000, 3670000, 0);
>>>> locT = locR.add(trans);
>>>>
>>>> —
>>>> Reply to this email directly, view it on GitHub
>>>> <https://github.com/x3dom/x3dom/issues/1322#issuecomment-2192318975>,
>>>> or unsubscribe
>>>> <https://github.com/notifications/unsubscribe-auth/AB23BYDQOLK25VKQG3TBTJ3ZJL65PAVCNFSM6AAAAABJRIL4AOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCOJSGMYTQOJXGU>
>>>> .
>>>> You are receiving this because you are subscribed to this thread.Message
>>>> ID: <x3dom/x3dom/issues/1322/2192318975 at github.com>
>>>>
>>> _______________________________________________
>>>
>>>
>>>> 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/20240723/f28d1053/attachment-0001.html>
More information about the x3d-public
mailing list