Difference between revisions of "January 3rd, 2013 at 5:00pm PST, Agenda and Minutes"

From Web3D.org
Jump to: navigation, search
(fix typos , mstly mistyped wird was "skeleton")
 
Line 28: Line 28:
 
Otherwise, also please look at the paper from Yvonne 'hanim_w3d_08.ppt' which talks about the very important topic of how to multiplex or combine and otherwise control many predefined and live data sets that describe all or parts of an activity. I'm not sure where the technology and thinking is now, but it fits right in with current opportunities as in 'nwip_HAnim120912.pdf'.
 
Otherwise, also please look at the paper from Yvonne 'hanim_w3d_08.ppt' which talks about the very important topic of how to multiplex or combine and otherwise control many predefined and live data sets that describe all or parts of an activity. I'm not sure where the technology and thinking is now, but it fits right in with current opportunities as in 'nwip_HAnim120912.pdf'.
 
   
 
   
And of course other documents discuss h-anim 2.1 work on accomodating diverse models. THis effort is based on the idea that we can define data containers that hold author-defined offests between the initial skelton pose and the initial location of component parts of the character, such as segment and skin geometries. When this works, the skin from the center figure can match the bones.
+
And of course other documents discuss h-anim 2.1 work on accommodating diverse models. This effort is based on the idea that we can define data containers that hold author-defined offsets between the initial skeleton pose and the initial location of component parts of the character, such as segment and skin geometries. When this works, the skin from the center figure can match the bones.
  
 
------------------------------
 
------------------------------
Line 37: Line 37:
  
 
2) "define data containers that hold author-defined offsets..."
 
2) "define data containers that hold author-defined offsets..."
 
  
 
- On Dec. 9, 2012, Joe mentioned the following:
 
- On Dec. 9, 2012, Joe mentioned the following:
Line 46: Line 45:
 
   
 
   
 
The absolute location is:
 
The absolute location is:
http://web3d.org/membership/login/groups/X3D_h-anim/X3DExampleJoe20090205.zip
+
 
 +
http://web3d.org/membership/login/groups/X3D_h-anim/X3DExampleJoe20090205.zip
 
   
 
   
 
Please download and unzip.
 
Please download and unzip.
Line 58: Line 58:
 
The file 'JointCoordinateAxes.x3dv' shows the standard x3d default value for the h-anim coordinate axis. Prior to animation, in the h-anim default pose, all joints have this orientation.
 
The file 'JointCoordinateAxes.x3dv' shows the standard x3d default value for the h-anim coordinate axis. Prior to animation, in the h-anim default pose, all joints have this orientation.
 
   
 
   
Run the file named 'JoeSkin4MGSkinDBonesRun3EX.x3dv' which uses all files in the set. An attached sceen shot shows the complete scene. mouseover the center figure to activate the animation; reload to initialize. If this does not appear and run as shown please tell me.
+
Run the file named 'JoeSkin4MGSkinDBonesRun3EX.x3dv' which uses all files in the set. An attached screen shot shows the complete scene. mouseover the center figure to activate the animation; reload to initialize. If this does not appear and run as shown please tell me.
  
 
------------------------------
 
------------------------------
Line 65: Line 65:
  
 
"Example extracted from nist http://ovrt.nist.gov/projects/vrml/h-anim/mangloss35.wrl"
 
"Example extracted from nist http://ovrt.nist.gov/projects/vrml/h-anim/mangloss35.wrl"
 
  
 
'''2. NWIP's scope and responsibility'''
 
'''2. NWIP's scope and responsibility'''
Line 106: Line 105:
  
 
------------------------------
 
------------------------------
 
  
 
=Minutes=
 
=Minutes=
Line 114: Line 112:
 
''Attendees:'' '''Myeong Won Lee, William Glascoe, Joe Williams, and Dick Puk'''
 
''Attendees:'' '''Myeong Won Lee, William Glascoe, Joe Williams, and Dick Puk'''
  
Myeong discussed her group's content creation process (skelton and skin together) displace binding; deformers (not displacers)
+
Myeong discussed her group's content creation process (skeleton and skin together) displace binding; deformers (not displacers)
  
Joe described the H-Anim specification nodes: ''skin coord index and skin coord weight (a lot of vertices in the skin but all are bound to joint we use displacer for moving skin independent of joint motion. For example, to raise an eye brow, you send  commands to displacer node to move the eye brow representing motion of the skin while the skelton remains still.
+
Joe described the H-Anim specification nodes: ''skin coord index and skin coord weight (a lot of vertices in the skin but all are bound to joint we use displacer for moving skin independent of joint motion. For example, to raise an eye brow, you send  commands to displacer node to move the eye brow representing motion of the skin while the skeleton remains still.
  
Take a LOA3 skelton as an example. It has...joints. When shoulder joint rotates the entire arm moves. If I move the wrist, all the hand joint move too. It's done in a hierarchy like this to achieve realistic measurements.  
+
Take a LOA3 skeleton as an example. It has...joints. When shoulder joint rotates the entire arm moves. If I move the wrist, all the hand joint move too. It's done in a hierarchy like this to achieve realistic measurements.  
  
Joe explained the text file of the running man he sent to us a few weeks ago. The skin animation (the running human) the visible bones is the geometry part of the skelton. There are two things that move a skin-  
+
Joe explained the text file of the running man he sent to us a few weeks ago. The skin animation (the running human) the visible bones is the geometry part of the skeleton. There are two things that move a skin-  
  
 
'''1) joint translation
 
'''1) joint translation
 
2) joint rotation'''
 
2) joint rotation'''
  
Myeong and Joe discussed skin nodes. Joe summarized the skins he has seen in projects. Many were not built in skelton space. They were built in "skin space". Sandy Ressler's skin is 40 times larger than a normal adult human. His character has 30,000 vertices, which is too many to connect with conventional authoring tools. The only practical way to use ii is to bind to the joints. One must bind textured geometry to the segments. But we want a realistic moving skin. In the skin node, ...face set... Most of these are called "skull tip". That other skelton is approximate 1.77 meters tall. I think it's important to build a skin in skelton space to avoid a lot of transformations. If my skin points correspond to my feature points, then I have a realistic character. You see that everyone of these is a named feature point (see the H-Anim Spec Annex B). There is not enough detail to do a lot of things in Joe's example but he did add a few feature points for more realism.  
+
Myeong and Joe discussed skin nodes. Joe summarized the skins he has seen in projects. Many were not built in skeleton space. They were built in "skin space". Sandy Ressler's skin is 40 times larger than a normal adult human. His character has 30,000 vertices, which is too many to connect with conventional authoring tools. The only practical way to use ii is to bind to the joints. One must bind textured geometry to the segments. But we want a realistic moving skin. In the skin node, ...face set... Most of these are called "skull tip". That other skeleton is approximate 1.77 meters tall. I think it's important to build a skin in skeleton space to avoid a lot of transformations. If my skin points correspond to my feature points, then I have a realistic character. You see that everyone of these is a named feature point (see the H-Anim Spec Annex B). There is not enough detail to do a lot of things in Joe's example but he did add a few feature points for more realism.  
  
Joe explained how some of the Motion Capture (MoCap) systems and projects don't worry about the skin. They create the skelton and put any skin on it they want. He said a good example is what happened in the Lord of the Rings movie. The readouts had no skin. You can follow the points inferred from the skelton. They did capture skin motion initially but that's not the norm today.  
+
Joe explained how some of the Motion Capture (MoCap) systems and projects don't worry about the skin. They create the skeleton and put any skin on it they want. He said a good example is what happened in the Lord of the Rings movie. The readouts had no skin. You can follow the points inferred from the skeleton. They did capture skin motion initially but that's not the norm today.  
  
 
Myeong assigned members to the ten work items.
 
Myeong assigned members to the ten work items.
Line 140: Line 138:
 
#8 Joe
 
#8 Joe
 
#9 Don and William
 
#9 Don and William
#0 Myeong and Joe
+
#10 Myeong and Joe

Latest revision as of 04:50, 2 April 2013

Agenda

The H-Anim WG meeting will be held with agenda items as follows:

1. Review of Joe's Comments

- On Dec. 7, 2012, Joe mentioned the following:


Please everybody download this .zip file and run the X3D H-Anim example named:

JoeSkin4MGSkinDBonesRun3EX.x3dv

also available at:

http://www.hypermultimedia.com/x3d/JoeX3D/X3DSkeletonRunning/MainJoeSkin4BonesRun3EXC.x3dv

from:

http://www.hypermultimedia.com/x3d/JoeX3D/JoeX3D.htm

anyway, this is all free example I used at a meeting in Sunnyvale once, and of course it was too complicated to actually describe in a couple of minutes. The important thing, is that I think we as WG should collaborate to move all or parts of this example into the official X3D H-Anim examples. It includes an LOA3 skeleton with example deformable skin (connected using hints from the '2005-DHM-10.pdf' paper -- please look at) by made up mainly from vertices associated with Site locations, with animation derived from existing X3D H-Anim Examples Nancy and others (mouse-over to start/stop, reload to initialize). Please note that a topic we all must deal with is how the initial poses of the default h-anim skeleton definition is different from the initial skeleton geometry pose in the foreground and from the initial skin geometry pose of the figure in the background.

If anyone can't see that example running completely, then please tell me. Let's discuss how to share some work in getting all or parts of this set of stuff into the official Web3D example set.

Otherwise, also please look at the paper from Yvonne 'hanim_w3d_08.ppt' which talks about the very important topic of how to multiplex or combine and otherwise control many predefined and live data sets that describe all or parts of an activity. I'm not sure where the technology and thinking is now, but it fits right in with current opportunities as in 'nwip_HAnim120912.pdf'.

And of course other documents discuss h-anim 2.1 work on accommodating diverse models. This effort is based on the idea that we can define data containers that hold author-defined offsets between the initial skeleton pose and the initial location of component parts of the character, such as segment and skin geometries. When this works, the skin from the center figure can match the bones.


Discussion:

1) "how the initial poses of the default skeleton definition [are] different..."

2) "define data containers that hold author-defined offsets..."

- On Dec. 9, 2012, Joe mentioned the following:


For the Member area 'Post or Download Work Group Documents' in 'X3D_h-anim' a discussion I would like to have is some work with the file named 'X3DExampleJoe20090205' which is represented in the attached screen shot.

The absolute location is:

http://web3d.org/membership/login/groups/X3D_h-anim/X3DExampleJoe20090205.zip

Please download and unzip.

There are both .x3dv and .x3d files in this set. In general, the .x3d files are single or combined skeleton parts with 'skeletonJoe1.x3d' serving as the main scene for the standing skeleton in the foreground.

The file 'NISTSiteMarkerBodyColor5.x3dv' is a very detailed skin with various colorpervertex applied in order to recognize areas of geometry in the user code. Example extracted from nist http://ovrt.nist.gov/projects/vrml/h-anim/mangloss35.wrl

The file 'JointCoordinateAxes.x3dv' shows the standard x3d default value for the h-anim coordinate axis. Prior to animation, in the h-anim default pose, all joints have this orientation.

Run the file named 'JoeSkin4MGSkinDBonesRun3EX.x3dv' which uses all files in the set. An attached screen shot shows the complete scene. mouseover the center figure to activate the animation; reload to initialize. If this does not appear and run as shown please tell me.


Note: The following link does not work.

"Example extracted from nist http://ovrt.nist.gov/projects/vrml/h-anim/mangloss35.wrl"

2. NWIP's scope and responsibility

What do you think of determining responsible persons for each scope? Please add responsible persons for each item. We recommend that the persons of responsibility work on draft and/or examples for conformance.


1) Definition of a humanoid character model capable of generating motion from captured motion data ==> Myeong/Joe

2) Definition of a humanoid character model capable of generating motion using 3D scanner data ==> Myeong/William

3) Definition of a humanoid character model capable of generating motion using a general motion definition such as keyframe, interpolation, kinematics, and dynamics for human figures ==> Myeong/Joe

4) Definition of motion parameters for transferring or exchanging motion between different human character models ==> Myeong/Dick

5) Definition of a motion data interface for including motion data ==> Myeong/Dick

6) Definition of a motion viewer’s functionality ==> Myeong/Joe

7) Definition of interoperable human behavior prototypes ==> Don/William

8) Definition of a 3D skinning interface ==> Joe

9) Definition of a recording interface for human motion such as gait ==> Don/William

10) Provision for easier exchange of characters from design systems to H-Anim systems. ==> Myeong/Joe


3. Scheduling next meetings

The Web3D Seoul Meeting 2013 will be held on January 29th (full day) - 30 (half day) at the Seoul Palace Hotel, Seoul, Korea.

Co-located at this meeting, our H-Anim WG meeting will be held on January 24 at 5pm PST (25th at 10am KST) with Web3D teleconference (offline and online).

This schedule may be subject to change (depending on the schedule for other WG meetings).


Minutes

The teleconference convened at 2000 and adjourned at 2230 hrs Eastern Time Zone

Attendees: Myeong Won Lee, William Glascoe, Joe Williams, and Dick Puk

Myeong discussed her group's content creation process (skeleton and skin together) displace binding; deformers (not displacers)

Joe described the H-Anim specification nodes: skin coord index and skin coord weight (a lot of vertices in the skin but all are bound to joint we use displacer for moving skin independent of joint motion. For example, to raise an eye brow, you send commands to displacer node to move the eye brow representing motion of the skin while the skeleton remains still.

Take a LOA3 skeleton as an example. It has...joints. When shoulder joint rotates the entire arm moves. If I move the wrist, all the hand joint move too. It's done in a hierarchy like this to achieve realistic measurements.

Joe explained the text file of the running man he sent to us a few weeks ago. The skin animation (the running human) the visible bones is the geometry part of the skeleton. There are two things that move a skin-

1) joint translation 2) joint rotation

Myeong and Joe discussed skin nodes. Joe summarized the skins he has seen in projects. Many were not built in skeleton space. They were built in "skin space". Sandy Ressler's skin is 40 times larger than a normal adult human. His character has 30,000 vertices, which is too many to connect with conventional authoring tools. The only practical way to use ii is to bind to the joints. One must bind textured geometry to the segments. But we want a realistic moving skin. In the skin node, ...face set... Most of these are called "skull tip". That other skeleton is approximate 1.77 meters tall. I think it's important to build a skin in skeleton space to avoid a lot of transformations. If my skin points correspond to my feature points, then I have a realistic character. You see that everyone of these is a named feature point (see the H-Anim Spec Annex B). There is not enough detail to do a lot of things in Joe's example but he did add a few feature points for more realism.

Joe explained how some of the Motion Capture (MoCap) systems and projects don't worry about the skin. They create the skeleton and put any skin on it they want. He said a good example is what happened in the Lord of the Rings movie. The readouts had no skin. You can follow the points inferred from the skeleton. They did capture skin motion initially but that's not the norm today.

Myeong assigned members to the ten work items.

  1. 1 Myeong and Joe
  2. 2 Myeong and William
  3. 3 Myeong and Joe
  4. 4 Myeong and Dick
  5. 5 Myeong and Dick
  6. 6 Myeong and Joe
  7. 7 Don and William
  8. 8 Joe
  9. 9 Don and William
  10. 10 Myeong and Joe