Sorry, I guess I worded by problem a bit poorly there. I can easily limit what tiles are drawn by either decreasing the fog distance or hiding the mesh(Tiles are in their own separate dim array). One issue is that I need to repeatedly extend the array of all known tiles in order to do this which isn't so bad, but it is a pain to reset once the map is re-created since we don't want to have a dim array of (40,40) holding (25,25). My concern however, is whether or not the new map generation system(Or the current one for that matter) will accommodate this. The current system holds maps as static entities, which is why it pieces rectangle maps seamlessly with each other. Dynamic maps like the pocket dimension have to keep their pieces in check or they traverse into another room. Don't get me wrong, this can be dealt with, but nearly everything related to positioning in this game assumes it's above 0 on the y coordinate. I'd prefer to see what the new map generation system is like before coming to a conclusion on this, but you guys can decide on what you feel needs to be done.risingstar64 wrote:
If you want it to extend endlessly I could write you a script to add and remove tiles from the rendergroup as the player moves. I needed to achieve something similar to this when I was working with pygame due to python's poor calculation speed, and I ended up turning tiles on and off for collision checking every 10 frames. By breaking it up over a period of time over which even at max velocity the player could not leave the area of collidable objects, I was able to cut back on calculation time without using a separate thread. In the collision detection scripts, If the player was the object checking for a collision with a tile it would only check the player against tiles that were on for collision detection, but if the object was an npc it would check against all tiles to avoid npcs falling through the ground. Because in this scene there will only be 1 npc, I think this could be quite useful if we need to speed things up a bit. This worked better than you might think because the player's physics were far more computationally expensive since there were so many things the player could do, and states he could be in, where the enemies really just needed to stay level with the ground.
EDIT:
Just another thought, but wrapping would make the map more disorienting by allowing you to come back to the path after moving along the width of the map. Not sure if this is the kind of thing you want.
)
