So to all our artists we have something quite cool to announce. We have now fully dynamic LoD level computation.
As you know SuperTuxKart support level of detail to help speed up the game, however setting it up is tedious and takes time, since you need to create various version of the same mesh and set them and guess the drawing distance etc.
The new system is much easier for everyone on both SuperTuxKart and the Blender side. Right now only the engine side is fully ready.
- * Antarctica will evaluate how big is your mesh on screen. It takes into account the "size" of the object and the scale of it.
* Then based on the estimated size of your object it will compute the level at witch optimized geometry must be displayed.
Now imagine you have a desert and a jungle. In the desert you want your palmtree almost always visible because it's the only thing to see. While in the jungle it's better to cull a lot of objects to avoid a collapse of performances.
- * If the user set in the settings the geometry details to low or disabled the distance will be reduced
* If the track is very detailed, Antarctica will estimate its complexity and cull them more aggressively, to try to keep the performance balanced.
All of this is fully automated, you don't have to setup anything if your object use LoD it will automatically do it
. You can already try it in the current master.Regarding the blender side, it's not yet ready to be presented but it brings some very cool things. Stay tuned.
Inspired by:
https://github.com/Unity-Technologies/AutoLOD
https://docs.unrealengine.com/en-us/Eng ... Generation
