My Final Pathfinding...
Community Forums/Showcase/My Final Pathfinding...
| ||
| I am satisfied with my final version of 3d Pathfinding. This has taken me months to get it in its present state. Once in a while it will jog off the path and if you trap the player sphere youll get a mav, but other than that I havent been able to confuse it. Press s to show the markers of the path... The Exe file... http://dodownload.filefront.com/9370766//1532576500828fdb31f71838a901ef86d5322449fe5b0b3eb4710af0fdc69be267933ba60ccc7278 The Zip file http://dodownload.filefront.com/9384626//1532576500828fdb31f71838a901ef86d5322449fe5b0b3eb4710af0fdc69be267933ba60ccc7278 |
| ||
| good stuff works great only get a mav when you said i would. nice work. |
| ||
| Well I think I found a bug :( Take a look to the following image:  Your algorithm decided to go following the small red dots but there is a shortest path this one with the bigger red dots. |
| ||
| If its based on A* then it doesn't necessarily return the shortest path. |
| ||
| Oops! sorry |
| ||
| Probably depends on the value used for diagonal moves?!?! |
| ||
| Its probably because nodes nearer to the finish are updated first, and farther away ones are ignored to speed up the algorithm. Just take the increased priority off of "being close to finish" and it will always find the shortest path. and it will run slower. |
| ||
| my download keeps FAILING... |
| ||
| I apologize Chwaga, I tested the download and it worked. I sent you an email with the zip file. Thanks for checking it out. As far as the shortest path, etc. The algo starts from the start cell and expands out from there based on G, H and F values per A* Pathfinding rules. Diagnol movement cost 14, horiz and vertical movement cost 10. If 2 cells tie eachother the algorithm takes the first cell on the open list and goes from there. It will not compare paths or find the shortest or quickest path, it will find a path and with pretty good speed that I can see so far. Thanks for checking it out. |
| ||
| In the illustration that Moraldi posted above, shouldn't the pathfinding algorithm be able to determine that the diagonal it has used ( between the second and third steps ) is invalid and not use it? Generally, it is considered bad form for pathfinding to allow you to cut corners, for fairly obvious visual reasons. |
| ||
| I did not include any checks for this kind of situation. The walls that get created were just a request from Ross to be able to create mazes and stuff to be able to test the algo more. For the project I'm working on, this situation wont happen due to the way I will be placing obstacles and stuff. Thanks for the input. Also Chwaga, the email returned saying the zip file I attached was an illegal attachment. Is there another email address I can send this to? |
| ||
| send it to chwaga@... |
| ||
| That is the email I sent it to. It got returned saying it had an illegal attachment. It was a zipped file. I also uploaded the zip file to filefront check the first post for the download link to that file. |
| ||
| You'll have to rar it. GMail does not accept exe's or zipped exe's. |
| ||
| /shrug |
| ||
| Ok, sent you the source files and exe in rar. Let me know what you think! Thanks! |
| ||
| i got it, thanks. |
   |