Tom/Paul {l Wrote}:As in title, so far for the walking algorithm the smallest atomic area the creature occupy or walks through is one gamemap Tile. How about make it a little denser ? Splitting into 4 or 9 or 16 positions, also the number could be a parameter at compile time.
Akien {l Wrote}:Why? Could you explain what would be the benefits, for which feature you would like to have this possibility, what could be the impact on performance for the pathfinding algorithms and creature behaviours?
Danimal {l Wrote}:i think he wants to add more than 1 tile big creatures that fill several tiles
Maybe yes, maybe not? Sub-dividing the grid has no benefits I can foresee so far, and won't help with handling big creatures in the end IMO.
Why? Because, you must then think about why you first want to have creature using several tiles at once. To me, it's usually when you want to make creatures block tiles to other creatures, usually with the hope to make them auto-decide of a road next to one another. But in practice, it doesn't suffice to work that way. Because the creatures that are crossing paths need to resync their paths to take one another in account, (and so not decide to both use the left area to cross their path, for instance.)
Sub-dividing the grid will also make the pathfinding slower, and that could be a huge drawback, as there is a lot of pathfinding going on while playing.
At the time, we (and mostly hwoarangmy) revised the pathfinding algorithm, I proposed to make creatures non-blocking and use a random position on a tile as a first pass at it, because it's cheap and in the end, not that bad looking.
The next thing in term of path-finding, IMHO, is rather to add a check when setting a random position on a tile, to make sure the creature doesn't overlap with walls when it's possible.
I'm and always be open for suggestions in this area, of course, but I'm pretty much sure we're close from what should be the end game path algorithm. Thus, I'm against sub-dividing the grid.
Best regards,