Pathfinding / Beams through walls?

[Rookie One.pl]

Has anyone been able to make func_beams appear through walls? Alwaysdraw doesn't work, it seems a trace is being done at the moment of activating the beam from origin to endpoint, and it stops at any solid found on the way. I tried tracing forward from inside the solid (endpoint + forward vector - this is 1 unit deep inside the wall), but the game refuses to do so and traces from the surface (where the beam endpoint is), which is useless, and as a result I get an error (either all edicts have been used or a command overflow). Any ideas?

[jv_map]

I believe it is impossible

[tltrude]

Perhaps the beam will go through "player_clip". Then the walls could be nonsolid. You're not saying much about what it is for, but how about making a small hole in the wall?

[Rookie One.pl]

I'm afraid I can't reveal much yet. It's for the project I wanted to use pathfinding for. My algorithm worked partially, but I encountered a bit of a problem (would be much easier if sighttracing returned the point hit) and scrapped the idea.

The thing is that I need to display a bot's path before they follow it, and beams would be perfect for it, if only they could be made to work like that (or a valid pathfinding algorithm was used).

[neillomax]

I don't know much about it, but what do you use to get a plane to fly around a map ? Can't you do something similar with the function beam ? Make waypoints for it to follow ?

[lizardkid]

planes use a number of nodes that have no connection other than a little value saying what's next in the list. he wants a visible line from node to node.

[jv_map]

(would be much easier if sighttracing returned the point hit)

Use trace maybe?

[Rookie One.pl]

I think the picture below explains the pathfinding problem quite well.

While the sight trace fails because of the obstacle, the linear one succeeds. This is not a problem with larger obstacles (I can find a path e.g. around a corner), but quite an issue with smaller ones.

Why is this important? Here is an excerpt of my pathfinding script:

Code: Select all

		local.dest = $startnode.origin + (angles_toforward (local.ang + ( 0 local.i 0 ))) * local.dist * -1 // honestly I have no idea why this needs inverting :S
		if (sighttrace $startnode.origin local.dest 0 ( -16 -16 18 ) ( 16 16 76 )) {
			println "Found at offset " local.i
			local.passages[local.pindex] = local.dest
			local.pindex++
			waitthread draw_beam ($startnode.origin + ( 0 0 32 )) (local.dest + ( 0 0 32 ))
			break
		} else {
			local.dist = vector_length($startnode.origin - (trace $startnode.origin local.dest)) + 32
			println "Distance updated to " local.dist
		}

[Rookie One.pl]

Bump!

I guess I need your help again, jv. I'm still trying to get it somewhere.

Providing the alpha and a, how do I get the value of b?

[jv_map]

The length of b cannot be determined. One more side or angle needs to be given.

[Rookie One.pl]

Hmm. How about this?

I forgot to mark the beta angle, but it's there between the black and green lines next to alpha. And alpha contains beta within itself.
c = sin(beta) * a
b = c / sin(alpha)
Would this work?

[jv_map]

Nope,

c != sin(beta) * a

but

c = cos(beta) * a

Similarly,

c = cos(alpha) * b

Such that

b = c / cos(alpha)

Or

b = cos(beta) / cos(alpha) * a

[Rookie One.pl]

Oh. But hey, I was close, and I haven't had trigonometry yet. Anyway, thanks a lot, m8.

1 more thing. Can I use your cosinus from Skylimit? You'll be credited, of course.

[jv_map]

Can I use your cosinus from Skylimit? You'll be credited, of course.

Ofcourse, although I can't really take credit for a Taylor expansion

cos x ~ 1 - x^2 / 2! + x^4 / 4! - x^6 / 6! + x^8 / 8! - ...

Where x is in radians.

pi radians = 180 degrees.

The skylimit script takes x in radians, as well.

! denotes a factorial, e.g.

3! = 1x2x3 = 6

[Rookie One.pl]

Thanks. And I've already made the deg->rad conversion, don't worry. And yes, I do see the light!