Author Topic: 2009-05-24 - Updates But No Eye Candy  (Read 8572 times)

0 Members and 1 Guest are viewing this topic.

Re: 09-05-24 - Updates But No Eye Candy
Alright, dangle a carrot in front of a nose and the mouth gets greedy, any chance we can see a short clip of the A.I. in action.  Just a straight 1 on 1, 2 minute vid?  Hopefully not asking too much, if so disregard.

It was always something that I felt was lacking from the BtRL demo (naturally I'm sure it was being worked on), the A.I.'s not using glide and even side-thrusting.

I'm just curious how brutal the A.I. is going to be on Insane.

Quite another question, and hell, might be far from relevant anymore given the strides made in the A.I. but in the BtRL demo, there was a definite problem with the A.I.'s collision detection, was/is this something easily remedied?  And further can it handle something as complex as navigating the Big Potatoes interior?

 

Offline Sushi

  • Art Critic
  • 211
Re: 09-05-24 - Updates But No Eye Candy
Quite another question, and hell, might be far from relevant anymore given the strides made in the A.I. but in the BtRL demo, there was a definite problem with the A.I.'s collision detection, was/is this something easily remedied?  And further can it handle something as complex as navigating the Big Potatoes interior?

Well, part of the problem is that there are a lot of hardcoded values in the AI code that are tuned to retail Freespace distances and speeds. BTRL had ships moving at speeds of 200, and the AI just didn't handle that very well. Diaspora's scale is fortunately such that this is much less of a problem. As to navigating complex capships/structures/asteroids... I at least haven't done anything in that regard. The AI just isn't designed to navigate large, complex meshes. The "best" solution I've heard so far is to give the AI some true pathfinding capabilities (such as A*), and that's definitely on the wishlist.

 
Re: 09-05-24 - Updates But No Eye Candy
Quote from: Sushi
Well, part of the problem is that there are a lot of hardcoded values in the AI code that are tuned to retail Freespace distances and speeds. BTRL had ships moving at speeds of 200, and the AI just didn't handle that very well.

Ah right, I remember reading about those issues way back when over on the BtRL boards and it slipped my mind that there were collision detection issues as a result, thanks for reminding me :).

Thanks for the info regarding the A.I. navigating the complex structures, I was merely curious if there was any headway in that regard.

Cheers!




 

Offline Master_Drow

  • 27
  • Theseus CIC officer.
Re: 09-05-24 - Updates But No Eye Candy
That new flight deck is sounding awesome. That will bring a whole new aspect to the game.

Also the turret improvements will be something to look forward to.

Thanks for all the hard work.

 

Offline Black Wolf

  • Twisted Infinities
  • 212
  • Hey! You! Get off-a my cloud!
    • Visit the TI homepage!
Re: 09-05-24 - Updates But No Eye Candy
Quite another question, and hell, might be far from relevant anymore given the strides made in the A.I. but in the BtRL demo, there was a definite problem with the A.I.'s collision detection, was/is this something easily remedied?  And further can it handle something as complex as navigating the Big Potatoes interior?

Well, part of the problem is that there are a lot of hardcoded values in the AI code that are tuned to retail Freespace distances and speeds. BTRL had ships moving at speeds of 200, and the AI just didn't handle that very well. Diaspora's scale is fortunately such that this is much less of a problem. As to navigating complex capships/structures/asteroids... I at least haven't done anything in that regard. The AI just isn't designed to navigate large, complex meshes. The "best" solution I've heard so far is to give the AI some true pathfinding capabilities (such as A*), and that's definitely on the wishlist.


Could you hack in something along the lines of dockpaths? E.g. plot out a bunch of splines in PCS2 telling the AI how to get through obstacles like those asteroid tunnels? They'd have defined widths the sameway dockpaths do, and only kick in when the AI needs to get from one side to the other of certain obstacles (Imagine, as an example, a very simple path through the Arcadia's hole rather than forcing the AI to go round or attempt to ram the Arcadia out of the way). They'd not be able to fight there, sure, and it's a touch hackish I guess, but, given that they're already in engine it might be easier? I dunno though, not a coder.
TWISTED INFINITIES · SECTORGAME· FRONTLINES
Rarely Updated P3D.
Burn the heretic who killed F2S! Burn him, burn him!!- GalEmp

 

Offline Sushi

  • Art Critic
  • 211
Re: 09-05-24 - Updates But No Eye Candy
Could you hack in something along the lines of dockpaths? E.g. plot out a bunch of splines in PCS2 telling the AI how to get through obstacles like those asteroid tunnels? They'd have defined widths the sameway dockpaths do, and only kick in when the AI needs to get from one side to the other of certain obstacles (Imagine, as an example, a very simple path through the Arcadia's hole rather than forcing the AI to go round or attempt to ram the Arcadia out of the way). They'd not be able to fight there, sure, and it's a touch hackish I guess, but, given that they're already in engine it might be easier? I dunno though, not a coder.

No, it's a good idea, and one that I've thought about a bit. :) We have subsystem attack paths and dock paths, so why not transit paths for certain ways of getting around large objects? The only drawback is that depending on the object, you may need a heck of a lot of them for it to be convincing, but if we ever do it that way it will be Someone Else's (the model converter's) problem. :D

There would also need to be some way to encode how "tight" the path is, so that bombers don't try to squeeze through fighter-sized holes.

Anyhoo, yes, it's a good idea.