What is better a scene or a heigthmap with a few models?
Depends what you are making and what your goals are, how many assets you plan to have etc.
Your question leaves few clues that would allow anyone to answer it.
First of all a height map with a few models could be considered a “scene”. But I think you mean the difference between generating a terrain from a height map and adding a few models or creating a scene in a 3D-editor, containing the terrain.
The main difference between the two is that with the height map, you terrain can only vary in the y axis ( height ), but it is VERY simple to add new terrain or modify the terrain. Loading a 3D model of the terrain obviously allows a lot more variability than a height map would.
So your question really comes down to another question. What do you want to make? Start out by defining in detail the requirements of your scene, THEN decide on the simplest method that would fulfil your requirements.
Ok, sorry I want to make a simple landscape with a little town (about 8 houses) in the middle and a few trees around the little town.
Well fi you dont need it to be changeable (digging ect), I woudl go with mesh since its a bit more efficient from a collision perspective. However it increases probably the plygon count of the terrain, wich could negate this. It depends on the size around the town.
I woudl go with mesh since its a bit more efficient from a collision perspective.
TerrainQuad uses heightfield collision and is significantly faster than regular mesh collision you would get from importing terrain from a different program.
Generally if you don’t need tunnels, caves, basements or anything else that would require holes in the terrain, then you can use a TerrainQuad. Else use a big model.
An exception might be big towns, because of the big flat area’s the polygons can be very big, which should make it slightly faster than a terrain too render, but physics might make the terrain a lot faster. What you sould shoose in a situation like this will probaly depend on how you handle collisions and the number of movable/physical objects.
Advantages of heightmap approach:
- Easier editing and texturing
- Built-in LOD
- Built-in scene partitioning
- Faster collision detection and physics
- Less flexibility - Can’t have overhangs
- Wasted polygons if terrain is mostly flat - but this is alleviated if you use the heuristic based LOD and not the regular one