LOD - any plans? how is feature devel. prioritized?

Hi,

many many things in Touch can be written by users (i.e. animation system, shaders, etc.). One thing that I don’t think can be developed by users is Level of Detail (i.e according to camera distance), which makes Touch very unwieldy as a multi-user, content-rich environment.

Are there plans to attack that issue anytime soon? I can’t think of any other missing feature more crucial and fundamental than this (among the features that can’t be developed by users).

Actually, in my book this is a prioritized list of features (maybe others can add/change/post their own?) in order of importance (as you go down the list feel free to point out things that may be developed by users).

  • Level of Detail
  • Redefined clip blending tools (I know that the old tools are not considered user ready).
  • Dynamics (i.e. hair simulation, collision detection)
  • Better particle system
  • more user-friendly CPU/GPU load analysis
  • better rigging tools (ie. interactive weight assignment/painting operator)
  • $T independent clock (is there such a thing? ie. if I wanted to do an editing system using Touch I would need some other way to play/pause/sync material - I know something like this was discussed internally at Derivative)

thanks!
dani

You mean rendering geometry with less vertices/polygons when it gets farther away from the camera right? This can be done by the users using Switch SOPs or maybe Blend SOPs (may need some improvement). The different LoDs should be pre-computed. Dynamically computing LoDs would be way too expensive and would end up slowing things down, not speeding them up.

In reality, its rare the number of vertices in your system will affect your speed as long as your models are reasonable. The two things that slow down synths 90% of the time.
1.The CPU
2.The pixel shader

A lower level of detail makes no difference to the amount of work a pixel shader needs to do.

Choosing the reduce the number of lights, or the complexity of the MAT for objs that are farther away from the camera will give a much better performance gain than rendering at a lower LoD.

To quote the God of real-time rendering, John Carmack, in a 2 year old article (when GPUs were even slower):