[x3d-public] HAnim Features Points, 1-120 for your review
John Carlson
yottzumm at gmail.com
Tue Mar 19 14:39:59 PDT 2024
I have been using OpenCV and mediapipe with Python to generate mocap data.
It’s not HAnim, but I have been able to generate joint and bone data. I
have a task to convert to BVH or ROUTEs and interpolators. Right now, I
have skeletons (separate graphs) animated in Java over a localhost IPC
(sockets).
I call it my time series graph (or deafblind) project, but I’ve been unable
to capture a movie with the webcam on. But I just got OBS, so I will try
again, so I’ll be able to capture my whole screen plus me making “signs.”
My microphone is not the best. I recently recorded a 4-headed FreeWRL
mathematical visualization project.
I’ve posted my partially implemented protocol which is based on Katy’s
joint center format. I layer this on top of my “chat” protocol.
I’m not sure if you guys have been following my X3DEOM project (derived
from X3DUOM) that I started to make a scenegraph visitor pattern easier to
write for JSHumanoids and Blender import. Hopefully, someday, I can write
renderers and animations.
John
On Tue, Mar 19, 2024 at 4:04 PM Joe D Williams <joedwil at earthlink.net>
wrote:
> please see below
>
>
>
> -----Original Message-----
> From: Carol McDonald <cemd2 at comcast.net>
> Sent: Mar 16, 2024 8:58 AM
> To: John Carlson <yottzumm at gmail.com>, Joe D Williams <
> joedwil at earthlink.net>, Katy Schildmeyer KS APPAREL DESIGN <
> katy at ksappareldesign.com>
> Cc: X3D Graphics public mailing list <x3d-public at web3d.org>
> Subject: Re: HAnim Features Points, 1-120 for your review
>
>
> > Here is the attachment on the first 11 feature points.
>
> Thank You Getting the Part 1 Annex A up using the latest proposed data
> for this example of a "Standard" example with example hierarchies, spec
> names, and example locations,complete foran LOA4 is much needed.
>
> > The site for the translations do not have points numbered so I added in
> that column.
>
> Fine,
>
> > As you can see from this attachment - the feature points are independent
> of the joints as it does not matter which LOA that you choose.
>
> True, all five loa examples show all surface feature points. It might be
> to simplify the Annex A by only having one table for listing Site feature
> points.
>
> > But questions - during the animation - do the feature points move? How
> are they bound to the mesh and skin?
>
> The feature points are implemented as Site nodes that are children of a
> specific Segment. So, a Site follows Segment motions as controlled by the
> parent Joint and custom Site animations. A Site can be a sensor or geometry
> or anything x3d.
>
> However, for Level 2 'skinned' projects a shortcoming in this architecture
> may show. The animation for the skin vertex depends upon weighted rotation
> of bound Joint node(s) and may not exactly match motion of the Site
> controlled by a Joint node. Thus, in a default pose while it easy to get a
> visible Site feature associated with a specific skin vertex or area, some
> extreme, or even typical, animations could cause the Site to move relative
> to the target skin vertex. (Not a problem with Level 1 Segment geometry.)
>
> This is a hypothetical problem for Level 2 hanim because it may be that we
> have better control of position and orientation of a Site node than of a
> specific skin vertex or skin area. The best solution is to give the author
> tools to choose. This means upgrading the skin part of the spec to allow
> the author to assign Site functionality to one or a group of skin mesh
> points. I don't think this is giant technical jump and for x3d syntax or a
> browser to implement.
>
> > If the skin (mesh) breaks during a movement - what happens to the
> feature point?
>
> If the skin breaks, there is a different animation problem.
> From above, since a skin point position can be controlled independently of
> related Site position and orientation (with various interlocking
> dependencies) we can see it is possible for the site and skin point to be
> aligned to start but move away from each other during animation. If we
> can't have a skin point with features of a Site, then in some cases author
> may need another animation routine to keep the Site location better aligned
> with the skin point location. Finally, there may be existing authoring
> systems that allow the author to have sensors on sensors on skin points as
> well as skeleton points now.
>
> > Thanks
> > Carol
>
>
> John > I am thinking that humanoid_root and all other joint centers are
> relative to the LCS (local coordinate system) of the humanoid, which is 0,
> 0, 0 relative to any transforms
>
>
>
> Yes and 0 0 0 is at the floor, between the feet. the HAnim local
> coordinate system is same as the default x3d parent coordinate system,1_1
> in meters.
>
>
>
> John > This means that for every joint center and every translation, it
> would be good to have reason for why is it located where it is.
>
>
>
> The Part 1 Annex A dimensions are just examples derived from historical
> data base of some sample humans. The names for Joint, Segment,and Site
> nodes are standard medical names. These are generally 'male' and it would
> be nice to have dimensions for corresponding female type.
>
>
>
> Thanks Again,
> Joe
>
>
> On 03/16/2024 7:08 AM PDT Carol McDonald <cemd2 at comcast.net> wrote:
>
>
> I have been looking into this. And I am agreeing with John's assessment.
> I will somehow need to get everything into Rhino for visualization and
> coding.
>
> From John on March 14th.
> I am thinking that humanoid_root and all other joint centers are relative
> to the LCS (local coordinate system) of the humanoid, which is 0, 0, 0
> relative to any transforms outside the humanoid.
> That's my best guess right now.
>
> What is not noted on the General is the exact file or model for which
> these translations were built for. This means that the list of the
> translations on the website only valid for the humanoid model that they
> were obtained for, the distance of that model from the origin and are not
> appropriate for Gramps or any other model.
>
> This means that for every joint center and every translation, it would be
> good to have reason for why is it located where it is.
>
>
> https://www.web3d.org/documents/specifications/19774-1/V2.0/Architecture/BodyDimensionsAndLOAs.html#General
>
>
> https://www.web3d.org/documents/specifications/19774-1/V2.0/Architecture/BodyDimensionsAndLOAs.html#LOA2DefaultSiteTranslations
>
> https://www.web3d.org/documents/specifications/19774-1/V2.0/Architecture/BodyDimensionsAndLOAs.html#LOA3DefaultSiteTranslations
>
> https://www.web3d.org/documents/specifications/19774-1/V2.0/Architecture/BodyDimensionsAndLOAs.html#LOA4DefaultSiteTranslations
>
> Is this the model for which the translations are built for? Can someone
> send me the proper link? Is this it? I will see if I can import it into
> Rhino 8.
> HumanoidAnimation/Skin/Joe*.x3d
>
> Back to updating every 120 points currently listed as to recommendations
> of how to locate it and adding the additional feature points that I have
> requested.
>
> https://www.web3d.org/documents/specifications/19774-1/V2.0/Architecture/FeaturePoints.html#FeaturePointsOnTheHumanBody
>
> Carol
>
> On 03/16/2024 4:23 AM PDT John Carlson <yottzumm at gmail.com> wrote:
>
>
>
>
> On Sat, Mar 16, 2024 at 2:38 AM Joe D Williams <joedwil at earthlink.net>
> wrote:
>
> > You can make the skin opaque to see that I'm really grabbing this from a
> HumanoidAnimation/Skin/Joe*.x3d example, just my own incantation of scripts
> to draw the humanoid geometry. I'll be checking those in soon
>
>
>
> Great John,
>
> I will check this out soon.
>
> Any ideas for the 'virtual' feature points?
>
> All Fine,
>
> Joe
>
>
> Joe, do you mean user drawn, not in current standard? User defined
> points? I have scripts for sites to place coordinates, but I don't have a
> way to get them into the hierarchy yet, so...
>
> John
>
>
>
> -----Original Message-----
> From: John Carlson <yottzumm at gmail.com>
> Sent: Mar 15, 2024 12:11 AM
> To: X3D Graphics public mailing list <x3d-public at web3d.org>, Carol
> McDonald <cemd2 at comcast.net>, Joe D Williams <joedwil at earthlink.net>
> Subject: Re: HAnim Features Points, 1-120 for your review
>
>
> Carol, I'm discovering it makes a difference where you put the Transform
> in the HAnimSegment as to whether to reuse the center. I don't have
> something that doesn't put out warnings yet, or takes the Sites away from
> the Joints. Hmm!
>
> My best Humanoid so far is Humanoid4J.x3d.
>
> I'm going to get some sleep.
>
> On Fri, Mar 15, 2024 at 12:53 AM John Carlson <yottzumm at gmail.com> wrote:
>
> Good use of Billboard for annotated feature points
>
> Please consider this a contribution to the Web3DConsortium examples. I
> release my copyright and licensing privileges, and give them copyright and
> licensing under the normal X3D resources example.
>
> Thanks to Joe and Don for good examples to draw from!
>
> No, I don't have any metadata yet. Let's set up a time and figure out who
> all contributed.
>
> You can make the skin opaque to see that I'm really grabbing this from a
> HumanoidAnimation/Skin/Joe*.x3d example, just my own incantation of scripts
> to draw the humanoid geometry. I'll be checking those in soon.
>
> John
>
>
>
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://web3d.org/pipermail/x3d-public_web3d.org/attachments/20240319/71fda7fd/attachment-0001.html>
More information about the x3d-public
mailing list