Advanced models, customising, animating and texturing

There are threads on this BB which relate to the scenario in decorating people or buildings with items, textures and animations.

Some of these are typically

  • 1. How do i make my character look like he has a beard ?.

  • 2. How can i attach a sword

  • 3. How can i have a high FPS with 50 people in one place all with different clothes and animating individually

1. How do I make my character look like he has a beard ?.
theprism said:

  • 2. How can i attach a sword

I once suggested a technique similar to the quake1-3 approach. That would be having morph target animations (very performant as opposed to skinned skeletons) and some kind of anchor points to attach meshes to each other. Quake did that by using three vertices as the anchor/reference point, my approach uses a separately animated Node (animated using a SpatialTransformer which then has to be synchronized to the mesh's KeyframeController)
It works ok for still objects, like swords. However, the synching of the two controllers turns out to be unreliable on slow machines. Of course, using it exactly the same way as I describe it would require you to use blender to at least some degree.
This technique is, as said above, probably faster than using skeletal animation (unless you are able to use shaders for skinning), but definitely has it's downsides as well. Identifying them is left as an exercise for the reader :D

I had tried to attach a window to a building in testing customisation. The model had a wall named wall 2. The issue i found at the time was that when a model is loaded, all of it children are relative to the model space, and not their respective children.

Its an interesting concept Hevee

? has anyone had sucess rendering alot of standard models that are customised ??

Any advantage in putting them in imposter nodes if the animation isnt quick ??

Imposter nodes give the advantage of not rendering a model every frame. They are especially useful for complex far away items such as houses or trees, etc. where drawing several polygons may result in just a few or even one pixel being drawn on the screen (iow, a real waste of gpu cycles)

Just wondering wether it would be effective to use this fps throttling mechanism with imposters in conjunction with LOD.

The LOD may not need to be checked every frame.

Also, if something changes and say the user switches to run mode, the update for the imposter would need to be more rapid, but only for close objects.

Have put some feelers out on the web.

one thing that has come back is Procedural animation

Anyone with any thoughts / experience with this