So I get that there is the resource cache of .mdl data which batch processes models.
The application here is a low poly city scene with several types of buildings (some scaled or with different materials). Nothing too crazy.
I’m trying to plan this out before I go down the wrong path. Not using the editor, just scripting. I’m making some simple models in blender. I figure I have the following options:
- Model each “piece” of my city (i.e. road, building 1, building 2, lamp, car, etc) and use a tile system to place the main static pieces in game, with some randomness perhaps. Maybe using nested tables to create a city grid and all that.
- Model groups of objects. For example I load premade complete city blocks or areas as .mdl. I piece those blocks together in a larger grid to make the city.
- Model a large blender scene of the whole city (or big parts of it) and import that as one huge mesh.
- Something else.
So what’s the smartest way to do this? What’s the performance impact between one large object (city block) vs. individual ones (singular objects), assuming the former is just the sum of the parts? Both cases would have multiple instances, but the premade blocks would be repeated less for sure… Or is there much a difference? I know Urho handles many objects very well, but does it do better with fewer large objects vs. much more smaller ones amounting to the same amount of polygons overall?