I am developing a RTS game with a bird vision camera running while playing. One of the fundamental aspects of my game are the trees.
So, in the days before, I was playing with trees in Unity.
- First, I played with Unity Tree Creator, but later I realized that it's broken since 2018.3, and the developers said that it would not be corrected because it would be a package developed by the community in the future.
- Then I downloaded store resources that are said to be highly optimized trees, but they are mainly meant for "3rd person camera projects", which represents about 7 to 14,000 triangles.
After these experiments, I decided to export a prefabricated tree to fbx, then manually reduce the number of triangles to about 2.5k in 3ds max, and then re-import it into Unity. With this technique, I was able to produce quite good results, but I still am not very satisfied with it.
My point of view of the camera and my "dense forest" between 180 and 300 images per second … and only as 3 or 4 shadows
For the above image, I used a tree with 2 LOD and a display panel (BB).
After playing this, I realized that I could do better, especially by playing with LD numbers, optimizing materials, and optimizing default trees.
However, the problem is that I can not model pins that meet my needs in 3ds max. So I started to look at the SpeedTree documentation, but it seems a bit too professional for all my needs (all the tutorials also make trees with a poly / sort of 20k, for example).
So, concerning this story. I would like to ask the following question + a bonus on LOD:
- Can you model quality trees with SpeedTree for Unity that have triangles of about 1k-2k, or SpeedTree for high quality / high density trees, their algorithms will not work with a lower number of polygons, etc.? (I've heard that they have a simplify modifier, is it good? Like making a 7k triangle and then cutting it to 1k-2k)
- I think using 4-5 LOD and a BB for a smooth transition. In my photo, I use 2 LOD and a BB and sometimes, when you quickly move the camera, the transitions are not as fluid as I wish. So I would rather have meshes detailed below before moving to BB. So, does having a 4-5 LOD-s is a common practice or a better idea for this?