THANK YOU!! what an easter egg

Hiya,

Glad you all liked it.. Now about your 'little complains'..

Quote:

---

The only hiccup was when the player got stuck under the stairs.

---

Yeah I got that once as well. In your game, a layer above the Pathfinding should detect this(I set the is_bIsStuck flag of the Npc when he gets stuck by any reason) and attempt to get him unstuck and do to pathdfind again

Quote:

---

Also, if you continuously click the left mouse button, it will increase the function count by one for each click. So, if I press the button 50 times before the player gets to the spot, there will be 50 functions running on top of whatever runs all the time. If you click a null spot, it zeros out(except for those 6-7 always running functions), and the player stops moving.

---

Good catch.. I fixed it.

Quote:

---

Anyway, congratulations...this is perfect for a point & click game..adventure or otherwise.

---

Thanks

Quote:

---

http://i56.photobucket.com/albums/g182/obinkmalmsteen/Clip_2.jpg

When I click the ground bellow
Why he pick the straight line and jump, instead the stairs?

---

Because falling down makes a shorter path than going down the stairs. In a game like Quake3 would you jump directly off a platform or try to find the stairs?

Anyhow, I hear what u say.. Sometimes it might be preferable for the NPCs to take the longer but safer route.(might take damage if they fall down big distances). So I made an is_ai_MaximumFall constant that you can define the maximum that your NPCs are ready to fall down.

Graph re-made with is_ai_MaximumFall==100 pixels (instead of 500)



And here's a path



However, I like it better when the NPC takes the shortest route in this level, so is_ai_MaximumFall is originally set to 500, but feel free to change it to whatever, it's in the is_IPathfinding.wdl(don't forget to re-build the graph, when you do). More on this on the Documentation coming up..

Quote:

---

http://i56.photobucket.com/albums/g182/obinkmalmsteen/Clip_3.jpg

he's not takes the shorter way!

---

That's a case where String Pulling (responsible for making smoother paths) fails due to tricky ground(bridges and all), exactly like we talked with Damocles a few posts ago. You can usually overcome this by re-arranging the waypoints in the level.

I fixed that for the case you found. But maybe it can happen somewhere else as well. I'll give out some good tips on laying out your waypoints in the Documentation.




And a new version with the new fixes.. The original link a few posts ago is updated as well though.

Intense Pathfinding 1.01

Code:

---

 
Fixes from 1.0
------------------
Fixed function leakage when attempting to take a new path before the old one is finished.
Added functionality to choose between falling from big distances to take the shortest path, or take a longer but safer(no falls) path.
Rearranged the Waypoints of the level for smoother paths.
 

---

Thanks all for your comments. Keep ep coming!

Aris

Ari,

very generous contribution indeed! THANK YOU man.

P.S. Xathikame alla xairomai pou to project sou zei kai basileuei. Tous exeis kanei na paramiloun, twra oloi perimenoun th synexeia! Na'sai kala kai filia apo thn patrida, opou kai na'sai..

Here's a weird one:

http://www.geocities.com/hainesrs/game/ai1.jpg
http://www.geocities.com/hainesrs/game/ai2.jpg
http://www.geocities.com/hainesrs/game/ai3.jpg
http://www.geocities.com/hainesrs/game/ai4.jpg

Sometimes, if you click on the side of a block, the player will try to move there and sometimes it won't. Ideally, you're going to want it to never do it or do it everytime(unless certain conditions are met...see below).

I think it would be nice to be able to toggle this behavior. If it's turned on then when a wall(or any object) is clicked, the player will move to the closet "legal" spot to the point. Of course, it won't be able to get to the clicked spot, but it should be smart enough to stop short of it if it can reach anywhere near that area. If an area is truly off limits, then the AI should know this and then a null function could activate(the red "X" stuff).

That's very very generous!!!

Perhaps, only a small step for you, but surely a big step for the community!!!!

Nice contribution !!
This looks very promicing
You are comming closer and closer
to the perfect AI, keep it up !!

Hey, very much thanks for this nice constribution!

Regards,
#benny#

---------
GenuineMotors.de

Thanks alot Aris!

I've been learning about pathfinding lately, and this will hopefully help me understand it better. The code seems to be documented/commented pretty good so that's great.

PS. why did you choose the dijkstra algorithm? What are the advantages over say, A*? Is it faster? Easyer to implement? How is performance for long and short paths compared to other algorithms?

A* is the better sollution for RTS type of 2D Pathfinding, with thousands of nodes.
It would not be the optimal sollution for 3D "Shooter" Type of games,
as it is unnessicarily complicated too for this purpose.

Hello!

thank you for your comments everyone.

I'm glad I've published this already, cause I've yet to start on that stupid documentation. Hopefully, I'll get some free time tomorrow. You get some idea from the comments anyway, but I also want to give you some good tips that would save you a couple of hours looking around.. So, tomorrow then... sorry

Quote:

---

Sometimes, if you click on the side of a block, the player will try to move there and sometimes it won't. Ideally, you're going to want it to never do it or do it everytime(unless certain conditions are met...see below).

---

You're a hell of a beta tester.. I'll see what I can do about that, probably drop the target spot to the ground and attempt to pathfind there, if there is a path. Nice catch again..

Quote:

---

PS. why did you choose the dijkstra algorithm? What are the advantages over say, A*?

---

There aren't any. A* is better, it's as simple as that. The reason I haven't upgraded from Dijkstra to A* yet(it won't be more than 3-4 extra lines of code), is that Dijkstra is doing ok. You won't get any decrease in the performance from running that algorithm several times per frame, so I put my effort into optimizing other things instead.

Don't forget though, that the actual search algorithm(breadth-first, depth-first, Floyd, Dijkstra, A*) is only half the picture. The underlying data structure is also critical, if not the most important part. The first pathfinding I made for 3dgs was inspired by Philipp Walser's tutorial and did a run-time A* search on a dynamic square-grid search space around the NPC. Needless to say, all these useless nodes that were being placed in stupid positions and flooded the whole level made up for a terrible search space and it's no wonder that no real game ever used this kind of technique. Intense Pathfinding is at least 100 times faster than my earliest attempt because of it's improved Search Space, even though it uses a slower search algorithm(Dijkstra instead of A*).

Hope I made some sense. I never know for sure when I ramble like that.. Good luck in your research on pathfinding anyway. It's hell.. you're gonna love it


immersionfx, geia sou file mou (ksexasa to onoma sou, lol ), pos paei to paixnidi?? Eides vlakeia, tora eimoun 1 mhna ellada, kai de sinenoi8ikame na pame kamia volta. Prox8es gurisa pali skotia alla ama einai kanonizoume kati gia to kalokairi Ego esy kai o Hellcrypt


Cheers all,

Aris