For the sake of making it easier to visualize, I flattened the steering to 2D and disabled some of the gizmos, but the flocking and obstacle avoidance behaviors are working in conjunction with this new addition.
The flow field generator has a layer mask for what physics layers are impassable, then has a list of pairings of layer masks and ushorts. The list is for physics layers that should be considered a higher cost to go through.
The AI heads towards the goal cube, and when reached, the goal cube finds a new area to move to. The integration and flow fields are updated (but not the cost field as that hasn’t changed).
In an attempt to gamify things, I made it so you can (1) spawn more AI by pressing S, and (2) spawn a temporary ring of fire. To keep things optimized, only the cells in range of the fire preform another physics check to update their costs.
Using the past assignment’s level partitioning, I also made the fire able to easily find out how many of the AI are in the fire, and there’s a timer you have of the collective time the fishies have spent in the fire. I increased the speed of the AI’s steering, too, in an attempt to make it a lot more fickle to catch/camp them.
To explain the gizmos: Red cubes are impassable, the passable cells go from green to blue based on integration values, and the white lines are the flow values.
Fishie model by Crystal Wong. Rock texture by Matthew Conrad.
