Question about node names for joints and collision geometries

Nodes

• As far as I've been able to tell, the node names don't matter. I mean, I have been able to import models with different rigs and bone structure and have them use the provided animations. In my case, I hadn't tried to use these as a player but in the player's datablock you can setup what animations the shape is linked to. So I don't see why it wouldn't work.
Collision

• The collision box can be named whatever you want. If you crack open the soldier_rigged.dae file and examine the model you'll find the collision box to simply be named 'Bounds'. If you were to go trying to name that collision box some kind of uniform name Torque would start throwing errors that the name is in use. Just name it anything and don't apply any material to it and it'll be invisible. Then you can just select it as your collision mesh in Torque in the Shape Editor.
Detail size

• I've seen suffixed detail size for LOD meshes, but not for collision. For collsion, usually a single shape will suffice (which you want to be as simple as possible). The LOD suffixes indicate how many pixels the object takes up on the screen. For example, if you have a model named 'object': If it was named object_LOD300 it would render that shape when the object takes up 300 pixels on the screen. If you moved farther away, obviously the model would take up less pixels. So you'd have another model named object_LOD200 that would appear when the model was farther away, only taking up 200 pixels.

Quote:
As far as I've been able to tell, the node names don't matter.

// Resolve spine
spineNode[0] = mShape->findNode("Bip01 Pelvis");
spineNode[1] = mShape->findNode("Bip01 Spine");
spineNode[2] = mShape->findNode("Bip01 Spine1");
spineNode[3] = mShape->findNode("Bip01 Spine2");
spineNode[4] = mShape->findNode("Bip01 Neck");
spineNode[5] = mShape->findNode("Bip01 Head");

void Player::updateAnimationTree(bool firstPerson)
{
   S32 mode = 0;
   if (firstPerson)
      if (mActionAnimation.firstPerson)
         mode = 0;
//            TSShapeInstance::MaskNodeRotation;
//            TSShapeInstance::MaskNodePosX |
//            TSShapeInstance::MaskNodePosY;
      else
         mode = TSShapeInstance::MaskNodeAllButBlend;

   for (U32 i = 0; i < PlayerData::NumSpineNodes; i++)
      if (mDataBlock->spineNode[i] != -1)
         mShapeInstance->setNodeAnimationState(mDataBlock->spineNode[i],mode);
}

Quote:
If you crack open the soldier_rigged.dae file and examine the model you'll find the collision box to simply be named 'Bounds'.

Quote:
DTS bounding box is simply an axis-aligned cuboid that fits around the shape. Torque 3D uses the centre of the bounding box as the object's origin when placed in the world editor, and it may also be used for simple collision detection.

It is important therefore for the shape bounding box to have the correct size and position. When exporting DTS models, this was done by adding a mesh called bounds to the root level of the scene. The DTS exporter would detect this special name and use the bounds mesh geometry to compute the shape bounding box.

This option is still available to models exported to COLLADA, and may be preferable if the model is animated and you want exact control over the size and position of the bounding box. For example, a walking character animation may move the feet or arms of the model outside the box containing the shape in its root pose, so you can use a custom bounding box to explicitly specify the bounding box extents.

If the DAE model does not contain a root level <node> called bounds with geometry attached to it, the Torque 3D COLLADA importer will automatically calculate a bounding box that encloses all of the geometry in the scene. For animated models, only the root (non-animated) pose is considered. For static objects this is usually adequate.

Every 3D shape includes an axis-aligned bounding box. This box appears around the shape when it is selected in the World Editor, and can be used for simple collision detection or mouse-hit picking. The bounding box is also used to determine which shape detail level to render (see LOD). The size of the bounding box is not fixed to the shape geometry. The modeler is free to define a custom bounding box extent for an object. This is normally done prior to DTS/DAE export by creating a cube mesh called "bounds" with the appropriate dimensions.

A mesh called "Bounds" can be used to define the object's origin (as opposed to your 3d content applications origin) and is used to define the speed that the object is intended to be moving at. Such as during a run sequence, if you had a character running forward at 1 meter a second in your 3d application's scene and had the "Bounds" object follow your character then when you bring it into Torque 3D and assigned to the player's run animation then the playback speed of the animation can be adjusted depending on the speed that the player is moving through the game world. What this means is that animations don't have any sliding issues caused by the character's in game speed not matching with the designed animation speed.

Animation sequences that move the character should include a ground transform. This tells the engine how fast the character would move along the ground when the animation is played back at normal speed. In the case of a Player object, this allows Torque to scale the animation playback speed to match the in-game speed that the Player is moving. For example, if the model was animated such that it would normally move at 3 units per second, but in-game was moving at 6 units per second, then the animation can be played back at double speed so the feet do not look like they are skating along the ground. Another use for ground transforms is to automatically switch between walking and running animations based on the in-game velocity of the Player.

The exact details of how to export ground transforms will depend on the modeling package. In general, the animation should be created so the character moves through space, rather than running or walking in-place. Animate the bounds node to move with the character so there is no translation relative to the bounds node. On export, the ground transform is determined by subtracting the movement of the bounds node from the walking or running animation so that it will play in-place in Torque 3D.

Quote:I've seen suffixed detail size for LOD meshes, but not for collision.

Quote:Oh, by the way, Torque uses both .dts and .dae still to this day.

Quote: Player spinal nodes are hardcoded for blending aiming/look/recoil animations.

TSThread* mHeadVThread;
TSThread* mHeadHThread;
TSThread* mRecoilThread;
TSThread* mImageStateThread;

Quote: I'm guessing the exact names hail from the days of the 3DS MAX exporter - the default node names are from the Biped default node names.

Quote:Yes, I know. But from an artists point of view the DTS can be considered as an internal binary format to the engine. That the normal and recommended current workflow should be Collada-based.

Quote:
The snippet in updateAnimationTree sets some nodes not to receive blended animations (i.e. the traditional way vertical look animations are implemented), which is very specific to the Tribes character/animation setup. I don't think you'll have any issues nuking that core entirely if you set up your own character and animations.