Hi again,

Adjusting the diagonal cost helped a lot with the initial path - I feel really dumb for not thinking of that! The final path looks MUCH better now using the new heuristic and cutting out unnecessary nodes like Mark suggested. I''m also looking into splines to help smooth out the turns.

If anyone else is interested, both of the AI Game Programming Wisdom books have some excellent articles on pathfinding too.

Thanks for all the help and suggestions.

Ok, firstly, I am aware that there is possibly a bug in my own path optimisation code, it doesn''t always seem to work right (although I believe the principle is sound, which is why I posted the pic)

Secondly, I did originally allow diagonal movement in my A* algorithm, but I realised that it created a lot more work, and resulted in what was basically the same path after optimisation. So I removed the diagonals, hence the "zig-zag" of the blue path. My characters don''t actually use the blue path for walking around anyway.

What I did do, was left in the diagonal rule in the heuristic calculation, which causes it to zig-zag.

Mark

I did consider that, I think the line-of-sight can work as long as you''re going across the same sort of ground.

The A* output should provide the least expensive route, then you apply the line-of-sight afterwards. If you only use the line-of-sight across consistent ground, it shouldn''t make the route any more expensive ever.

Mark