What option is better?
A. Have 20-50 models that each has around 7-12 parts in average.
B. Have 10-25 mesh parts that each has in average 100-300 triangles ( they are mainly normal blocky buildings, simple).
What is more performant? Since my project is pretty big and planed to have subway system too and more things
Should I convert each of these blocks [ as in green] into meshes?
each of those has around 7-20 bricks , so i wonder if i should leave it as is or convert them to blender
Because if you look down here, you can see the city is bigger and needs more buildings in the future
I believe the only way to know for sure is with testing. Two methods I can think of is through the micro profiler and Render stats view.
Also to note these methods are also dependent on the positioning of the camera and probably a load of other assumptions. But they should be a good place to start as it provides quantifiable results and not just opinions.
Microprofiler method
One option I can think of is the micro profiler
Ctrl+Alt+F6
Then pause it and scroll through the yellow frame bars until you find scene which is the amount of time it takes to render the parts.
Ctrl+P
You will be able to monitor the time it takes to render the parts mine below is 2.81 ms.
And with a lot of parts Delta value is 24.7 ms with a lot of parts, also the Draw number is 108 instead of 26 before indicating well it has to render a lot of parts which takes time and lag.
Probably the best option for using parts is to turn off all their collisions and use a large invisible part for the collisions of each building/block depending on if the players will be able to get on top of the buildings.
