1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

rFactor Performance - Materials

Discussion in 'Bob's Track Builder' started by ehrlec, Jun 25, 2011.

  1. link
     
  2. Very interesting thoughts about rF! For RBR, I will always maintain that less materials is better.
    Forgive me non-technical language and my English of course too...
    I had various subjective observations in quite enough number of tracks and I would like to add that the track's performance also depends on number of objects that BTB is creating while export. He groups the same type of objects in some way that they cover area around 150x150 meters. The simple test would be (in BTB) to plant many trees of one kind on couple kilometers of the track, export to rF, import to 3dSimed - then you will see how they are groupped (and these groups are single .gmt's). The point is (I think) - to improve track's performance you could group them better (or even join other meshes which use the same texture to these groups). Because the number of polygons has less importance that the number of objects IMO. I think this works similar to rF.

    I think - on optimizing thread - there are many more aspects that someone could think of - sometime they are closer, sometime they are less connected with each other.. But I was always trying to spend much time on this - to make tracks more playable :)

    I appreciate your efforts to make tests like yours. Great work!
     
  3. The basic premise of the gMotor engine is having max 1000polys per GMT group, then it doesn't matter how many GMTs your track contains as long as your LODs are reasonable of course.

    I have seen major performance increases in certain tracks upon merging materials. IIRC I edited Fuji 2007 for the RDLMS, and I gained 20FPS (from 50-70) through material optimisation. This included some texture resizing also I believe, but nothing that would gain 20FPS. However, other tracks I have found no performance gains through material optimisation.
     
  4. True, it could also have been gained by reducing the stupidly sized crowd textures and a lot of excess mip-map negative bias. But the combining of identical textures definitely helped.