Randomisation in scripting

[Guest Lucilla Frost]

Hello

 

Question from a relative newcomer here with a bit of programming experience.

 

I'm trying to set up a script that has a random chance of going one way, and a random chance of going another way. I've based my code on the dog fight at the Copper Coronet which has a 50/50 of player win lose.

 

~ From my NPC.d file ~

 

IF ~~ THEN BEGIN OPPC4

SAY ~Lets try it and see what happens.~

IF ~~ THEN DO ~StartCutScene("OPTST")~

EXIT

END

 

 

~OPTST.baf~

 

IF

True()

THEN

RESPONSE #50

StartCutSceneMode()

StartCutScene("OPPCF1")

RESPONSE #50

StartCutSceneMode()

StartCutScene("OPPCM1")

END

 

 

The cut scenes OPPCF1 and OPPCM1 work fine individually, but my attempt at randomising them always results in the first placed being implemented. In this case OPPCF1. As far as I can see I haven't done anything wrong, but it doesn't work. Have I made a basic error through copying code I don't actually understand? If anyone can help I will be eternally in your debt.

 

xx

 

Lou

[devSin]

Cutscenes don't work the same as normal scripts.

 

Change your dialogue:

IF ~~ THEN BEGIN OPPC4
 SAY ~Let's try it and see what happens.~
 IF ~RandomNum(2,1)~ THEN DO ~StartCutSceneMode()
StartCutScene("OPPCF1")~ EXIT
 IF ~RandomNum(2,2)~ THEN DO ~StartCutSceneMode()
StartCutScene("OPPCM1")~ EXIT
END

You can then get rid of OPTST.

[berelinde]

They're supposed to have equal chances of coming up, but it never seems to work that way, does it? Maybe it would be more equal if you tried 40/60 or 30/70? The numbers don't have to total 100%, by the way. It's just convenient.

 

Edit: DevSin must have been posting at the same time. He's the expert.

[Guest Lucilla Frost]

Thank you so much. I love you forever! (In a purely friendly fashion).

[Mike1072]

They're supposed to have equal chances of coming up, but it never seems to work that way, does it? Maybe it would be more equal if you tried 40/60 or 30/70? The numbers don't have to total 100%, by the way. It's just convenient.

I think that each response is supposed to have a 1/num_of_responses chance of being chosen, and whatever number is associated with the response indicates the probability that when chosen, the response will actually happen. So, 2 RESPONSE #100s should make one of them happen, and give an equal chance of it being either one. Supposedly this doesn't work out in practise - the first RESPONSE always gets weighted higher than the rest - and the RandomNum trigger ends up being more reliable for producing random results.

[Ardanis]

Afaik the chance for RESPONSE #X is X/(X+Y+Z).

 

the first RESPONSE always gets weighted higher than the rest

Yep, have seen that mentioned somewhere before.

[Mike1072]

Afaik the chance for RESPONSE #X is X/(X+Y+Z).

Oh, maybe I was mistaken about that. Boo to scripting.

[Qwinn]

One thing I discovered in my PS:T modding is that the total weights allocated to the RESPONSEs -does- matter. If your total weight is under 400 or so, then it seems to work okay. But if you have more than four or five RESPONSE #100's, those after the first four or five will never come up. This actually kept a whole lot of lines from showing up as stringheads in the game (some characters were supposed to have as many as 15 different strings appear their heads). I fix this for PS:T... I wonder if anyone has tested or fixed this in the other IE games, assuming it happens?

 

Qwinn

[Taimon]

Pseudo-code for response picking:

r = rand(0, sum(R1 ... Rn) - 1)
sum_so_far = 0
for resp in responses do:
sum_so_far += resp.prop
if (r <= sum_so_far)
	return resp