Re: [Glob2-ideas] non-flat terrain

[Kai Antweiler]

> The idea is nice. One technical (but solvable) issue is that units should take
> terrain altitudes into account while pathfinding and thus the pathfinding
> memory should be bumped to 16 bits per tile (to add precision to cope with
> various types of slopes).

Bradley once had an idea to cluster some fields for gradient computation.
With some fields clustered we might not have to add additional 8 bits per
field but still could address the fields efficiently.
On the other hand 16 bit would give us some capacity for the future ...


> This is not a big deal but it would slow down
> pathfinding (but to which extend I cannot estimate, maybe it is negligible,
> maybe not).

I think this will really become an issue, unless the gradient update proposal
works well. Gradient computation doesn't use expensive math - it's load is
all about accessing memory.

By the way:
We already have the same "problem" with different speeds on different terrains.
Slow swimming globs take the short route through the water even when they
would be faster on land.  We argued that they nevertheless optimize distance
and that food-levels depends totally on traveled distance and not on time.
The same argument also fits now.
-- 
Kai Antweiler