Blender and Unity3D Game Engine – Rough Acquintance

I have rough legitimate suffering on learning Blender and Unity3D at the same time. Not just learnt on both workflow and tried to grasp what was on their very souls, but also to coordinate two different realms : technical and academical aspects. Hence the name applied : researching. On such matter, followed by curiosity and little bit of ambitious, I begin to recognize few things to be concerned for developing urban-scale digital simulation application with game engine. In this case I use Blender as 3D modeler with more than enough capability and still considerably faster than any proprietary software I’d ever known. Blender also has some exiting and effective system for texture mapping, the one most important thing for large scale architectural or urban visualization.
In the other hand, Unity3D has quite excellent interoperability with Blender native file (.blend). So it is convenience to switch and synchronize data between those two.
Here some of important issues I have been found so far :
1. Urban simulation is based on real existing urban fabric. So it is not necessary to develop 3D geometries by using procedural or generative approach as noted by some researchers. City generator would be important for creating ‘neverland‘ city to be used in certain type of game (not Grand Thief Auto’s type though…).
As such, 3D data would be developed either by manual (digitizing), semi-automatic or automatic approach (using DEM, Aerial Photo and LIDAR data). This data would then can be opened in Blender for further processing (Blender has wide range of imported file format).
2. Vegetation. How we can handle trees, shrubs, grass and other types of data?. Some of the vegetations might play important role as urban components, but some others just as greeneries. For certain case, using procedural approach (tree generator) would be better for gain realistic atmosphere (shadow and 3D meshes). But for others, billboard approach is necessary since it could reduce memory consumption.
3. Level of Detail (LOD). On game environment – in particular First Person Shooter Genre-, we ought to see the world from person point of view with certain view angle (60 degree ?). We see high detail of object nearby but (might be) blurry and low detail (object and texture) at the distance. Sometime (and at most common cases), computer rendering the texture at the same quality. Whether in game environment, this might cost memory and end with slow performance. The MIP mapping (Multum in Parvo – Lance.1983) therefore can be used for generating level of detail based on camera view. Again, Blender has built in such mapping parameter. Other is object culling. The objects which is obstructed by object facing to the camera will not be counted, and therefore will not be rendered.
It is common procedure for increasing performance in game engine.
4. Animated avatars. This is another story….I’ll add it later.


About this entry