RNG discussion

Talk about Advance Wars DS. Debate, laugh, cry, argue about everything to do with our favourite tag game here.
Post Reply
User avatar
-STL-

RNG discussion

Post by -STL- » Tue Jun 23, 2009 6:24 am

Duuno where this belongs, but here http://www.warsworldnews.com/index.php? ... ostats.php is Nell´s luck wrong
Nell´s D2D luck in AWDS is 20 not 15.

User avatar
Xenesis
Tri-Star CO
Tri-Star CO
Rank: Hydrocarbon Inspector
3DS Code: 2535-4646-7163
Location: 0x020232DD
Contact:

Re: Mistake on CO Guide

Post by Xenesis » Tue Jun 23, 2009 11:00 am

Eh? I copied that number straight out of the ROM.

Edit: It appears I copied it wrongly too. O_o
IST wrote:Even the worst individual needs to discover the joys of a chicken statue that is also a pregnant blonde housewife.

User avatar
-STL-

Luck Stuff

Post by -STL- » Tue Jun 23, 2009 8:09 pm

There´s another thing, a 100 Attack CO is not able to do 1HP with a 1% attack like from an Inf or Recon to a Megatank (apart from the Luck COs ofc) even if he gets max luck. While Kanbei is able to do so, maybe Hawke/Von Bolt aswell. And i remember OHKOing a Megatank with a Inf and Nell´s SCOP while some luck tests. So it seems that 110 Attack is needed to do really 1% Dmg with a weak unit against Megatank.
As u can look into the Rom, can u explain it?

User avatar
Xenesis
Tri-Star CO
Tri-Star CO
Rank: Hydrocarbon Inspector
3DS Code: 2535-4646-7163
Location: 0x020232DD
Contact:

Re: Mistake on CO Guide

Post by Xenesis » Tue Jun 23, 2009 8:21 pm

Well, you can provided there's no terrain defence, it's just a 1/10 shot.

When I've specified numbers of "+10%" or something, it really means that the RNG has 10 numbers, ranging from 0 to 9.

A 1% Attack can do between 1 and 10% damage to the enemy unit.

When there's terrain defence or the opponent has defence boosts, the result is multiplied by something less than one (A full HP unit on a city for example is 0.7, resulting in an attack range of 0 to 7% damage).
IST wrote:Even the worst individual needs to discover the joys of a chicken statue that is also a pregnant blonde housewife.

User avatar
-STL-

Re: Mistake on CO Guide

Post by -STL- » Tue Jun 23, 2009 8:27 pm

Xenesis wrote:Well, you can provided there's no terrain defence, it's just a 1/10 shot.

When I've specified numbers of "+10%" or something, it really means that the RNG has 10 numbers, ranging from 0 to 9.

A 1% Attack can do between 1 and 10% damage to the enemy unit.

When there's terrain defence or the opponent has defence boosts, the result is multiplied by something less than one (A full HP unit on a city for example is 0.7, resulting in an attack range of 0 to 7% damage).

Yes i know how luck works, but the Megatanks keeps out of this order for some reason. Andy is not able to do a full HP of dmg with an Inf against Andy´s Megatank on Road, even if he gets his 9% luck, while Kanbei can make 1HP dmg if he gets his 9% luck.

A Unit attacking a full HP Unit on City has a chance to get 0-6,3% extra dmg. The game only rounds down after adding luck and Base dmg and multipling it with Attacker´s HP.

User avatar
Xenesis
Tri-Star CO
Tri-Star CO
Rank: Hydrocarbon Inspector
3DS Code: 2535-4646-7163
Location: 0x020232DD
Contact:

Re: Mistake on CO Guide

Post by Xenesis » Tue Jun 23, 2009 11:52 pm

It appears to be a bug in the RNG code.

Especially as I've found where the game stores the final results of it's luck seeds and after it's done the firepower and defence calculations....it's a very weird finding. I'm doing this testing by attacking megatanks with infantry both on roads on the US Version of AWDS.

With Adder, Max, Olaf, Colin, Eagle, Hawke, I was getting on luck rolls 1% + 1, 3, 5, 7, 9 to allow for 2, 4, 6, 8, 10% and no other numbers
Jake, Rachel, Andy, Sami, Hachi, Grit, Sasha, Kanbei, Sensei, Grimm, Drake, Jess, Javier, Lash, Koal, Kindle and Von Bolt seemed to be only able to get 1, 3, 5, 7 and 9%

Jugger seemingly worked fine and got all numbers on his positive RNG roll between 0 and +25% (That I tested so far)
Sonja got 0, 1, 2, 3, 4, 5, 7, 8, 9, but no 10 (less than 0 results = 0)

So, I confirmed on my cart with Japanese FWDS - Andy won't damage in this situation, Max will.

So what I said is technically correct, but the RNG for AWDS is screwy to all hell. It seems that for a fresh Megatank, only Adder, Max, Olaf, Colin, Eagle and Hawke will be able to damage it with a 1% attack off an infantry.
IST wrote:Even the worst individual needs to discover the joys of a chicken statue that is also a pregnant blonde housewife.

User avatar
-STL-

Re: Mistake on CO Guide

Post by -STL- » Wed Jun 24, 2009 12:04 am

But Colin can´t do do anyway cause he has 90% Firepower and i did 1HP dmg with Kanbei´s Inf against Andy´s Megatank on Road yesterday!

Edit: Just tried with Adder and he did 1HP in 9 out of 32 tries. Weird stuff.

More tests:
Andy´s Inf on Adder´s Mega
0 times from 16.

Andy´s Inf with his COP activated on Adder´s Mega
5 times from 16

User avatar
Xenesis
Tri-Star CO
Tri-Star CO
Rank: Hydrocarbon Inspector
3DS Code: 2535-4646-7163
Location: 0x020232DD
Contact:

Re: Mistake on CO Guide

Post by Xenesis » Wed Jun 24, 2009 12:39 am

i did 1HP dmg with Kanbei´s Inf against Andy´s Megatank on Road yesterday!
It had probably taken 0.1 to 0.9HP of counterattack damage in some fashion.

The minimum damage an attack can do is 1%. Every CO's infantry Vs Megatank matchup is 1%, so firepower modifiers don't come into it.

I just ran through with Colin:
9/30 inflicted 1HP damage to megatanks.

Kanbei:
0/50 inflicted 1HP damage.

It's possible that activating a COP/SCOP/TCOP will alter what numbers the RNG spits out, hence why Andy is getting evens D2D and odds under COP. I'll test later.

Edit: It seems based on the DS used, at least preliminarily. On my DS Rachel will hit the 10% against the megatank, but she won't on my emulator, while Hawke and Adder on my DS won't hit the 10%, but it will on an emulator. I'll grab my sister's DS tomorrow and do some more testing.
IST wrote:Even the worst individual needs to discover the joys of a chicken statue that is also a pregnant blonde housewife.

User avatar
Xenesis
Tri-Star CO
Tri-Star CO
Rank: Hydrocarbon Inspector
3DS Code: 2535-4646-7163
Location: 0x020232DD
Contact:

Re: Mistake on CO Guide

Post by Xenesis » Mon Jul 06, 2009 11:24 pm

Okay, I ran the tests on my sister's DS.

Here are the COs who could get maximum luck rolls (excluding Nell, Flak and Jugger) D2D.

Andy
Kanbei
Sonja
Drake
Adder
Kindle
Jake

It seems that the algorithm is flawed, and it is also based on the DS that you are using to play the game.
IST wrote:Even the worst individual needs to discover the joys of a chicken statue that is also a pregnant blonde housewife.

User avatar
-STL-

Re: Mistake on CO Guide

Post by -STL- » Sun Aug 30, 2009 8:50 pm

Stealth on Carrier Base dmg is 40%, there might be more Mistakes in the damage Chart.

User avatar
HPD
Tri-Star CO
Tri-Star CO
Rank: Mentat
Location: The Mountain
Contact:

Re: Mistake on CO Guide

Post by HPD » Mon Aug 31, 2009 7:25 am

If the error is in the damage chart, you might want to e-mail GipFace about it. I believe he supplied that one.
"So when I say the fudge shaman flies he goddamn well flies and that's that." - Narts
"My motto is that there are far too many women in the world to waste time with men." - thefalman
"It's just that I'm not really aware of how a common conversation goes." - Imano Ob, talking on MSN about talking on MSN
"As for FE8, that was IS' variant of Man Spam - Dudes with Swords edition." - Xenesis

User avatar
ALAKTORN
Location: Italy

Re: Mistake on CO Guide

Post by ALAKTORN » Mon Sep 13, 2010 8:02 am

I ran some tests on emulator, and the thing about DSes affecting luck seems to be wrong, on my emulator I can get both sets of luck (odd numbers and even numbers) it just changes from one set to the other for some reason, maybe reloading has to do with it (also there are far too many strategies that require perfect luck that fidelcastro and STL made work on their DSes, so it can't be that on some DSes you can't have certain amounts of luck)

another test showed that with 1 star of TAG (Jake+Jess) all 15 possibilities of luck were possible
Last edited by ALAKTORN on Mon Sep 13, 2010 8:33 am, edited 1 time in total.

User avatar
Xenesis
Tri-Star CO
Tri-Star CO
Rank: Hydrocarbon Inspector
3DS Code: 2535-4646-7163
Location: 0x020232DD
Contact:

Re: Mistake on CO Guide

Post by Xenesis » Mon Sep 13, 2010 8:26 am

I'll check when I have the chance.

You have the European version of AWDS, right? I did the testing on a Japanese version and I know they fixed bugs between AWDS J -> U -> E. Perhaps they changed how the seed is initialised between releases.
IST wrote:Even the worst individual needs to discover the joys of a chicken statue that is also a pregnant blonde housewife.

User avatar
ALAKTORN
Location: Italy

Re: Mistake on CO Guide

Post by ALAKTORN » Mon Sep 13, 2010 8:35 am

Xenesis wrote:I'll check when I have the chance.

You have the European version of AWDS, right? I did the testing on a Japanese version and I know they fixed bugs between AWDS J -> U -> E. Perhaps they changed how the seed is initialised between releases.
that about Jugger I think it was actually a mistake on my part, because when trying to re-get it to make a screenshot, I couldn't get it anymore…

I have the US version (the one I use on emulator is a friend's rom+save)

User avatar
ALAKTORN
Location: Italy

Re: Mistake on CO Guide

Post by ALAKTORN » Tue Sep 14, 2010 2:39 am

sorry if it's redundant I just wanted to get all info in one post:

Sonja gets all 15 possibilities (-4%...+9%)
Jugger gets all 45 possibilities (-14%...+29%)
Flak gets all 35 possibilities (-9%...+24%)
Nell, like all other COs, gets either evens or odds
equipping Luck skill won't change the D2D luck from a set to all possibilities or viceversa (tested with Sami, Nell and Jugger)
1 TAG star makes the RNG create all possibilities (tested with Jack+Jess)
2 TAG stars will work the same as Luck skill (tested with Andy+Eagle)
3 TAG stars works the same as 1 star (tested with Sami+Eagle)

to change from a luck set to another, you can simply move around the cursor randomly while selecting an unit (you must move inside the unit's range) note: this doesn't always work, it might have something to do with how many times you move around, but I can't understand it that precisely

edit: breakthrough discovery: if either of your luck is odds (Sonja/Sami +15/-5 odds/odds; Jugger/Kindle +35/-15 odds/odds; Jugger +30/-15 evens/odds; Flak +25/-10 odds/evens) you get all numbers, Flak+Lash (+30/-10 evens/evens) you get stuck on odds luck without possibility to change it, tested with Recon vs Recon, Tank vs AA, Tank vs Tank and Tank vs BC always on Road full HP

edit3: AI's day has all possibilities and ours doesn't (it follows aforementioned rules)

edit4: a 100% method of changing luck set is to simply use up an RNG number by moving to a square which requires RNG to determine the path

I've also been testing how AI's luck set works, and I think it changes at the same time as ours, it can be either the same as ours or different, so I think at the start of our day a luck set is randomly decided for us and the AI
Last edited by ALAKTORN on Tue Nov 30, 2010 3:21 pm, edited 8 times in total.

Xen

Re: Mistake on CO Guide

Post by Xen » Tue Sep 14, 2010 10:08 am

What I found with the COs with dual RNGs (Positive + Negative luck COs) is that each of their RNGs results in all odds or all evens, but as they have two rngs, the chance of getting one with both all odds/evens is much lower so it appears they have more values. Most telling thing is whether they roll the maximum possible or minimum possible values (Eg, for Jugger to roll a -14 he needs to have an even +ve RNG and an odd -ve RNG).

I'll have to look into the Tag stars though, I never actually tested that effect.

User avatar
ALAKTORN
Location: Italy

Re: Mistake on CO Guide

Post by ALAKTORN » Tue Sep 14, 2010 10:50 pm

I've actually gotten every possible result with Jugger, Flak and Sonja, it wasn't an estimate, if that about dual RNGs odd/even were true, getting both minimum luck and maximum luck would be impossible, wouldn't it?

User avatar
Xenesis
Tri-Star CO
Tri-Star CO
Rank: Hydrocarbon Inspector
3DS Code: 2535-4646-7163
Location: 0x020232DD
Contact:

Re: Mistake on CO Guide

Post by Xenesis » Wed Sep 15, 2010 12:17 am

Possibly.

I think I may just have to disassemble the RNG code :|
IST wrote:Even the worst individual needs to discover the joys of a chicken statue that is also a pregnant blonde housewife.

User avatar
ALAKTORN
Location: Italy

Re: Mistake on CO Guide

Post by ALAKTORN » Mon Sep 27, 2010 6:02 am

hooray new discoveries: to change from a luck set to another, you can simply move around the cursor randomly while selecting an unit (you must move inside the unit's range) note: this doesn't always work, it might have something to do with how many times you move around, but I can't understand it that precisely <--- edit: this info is outdated
Last edited by ALAKTORN on Tue Nov 30, 2010 3:07 pm, edited 2 times in total.

fcastro

Re: Mistake on CO Guide

Post by fcastro » Mon Sep 27, 2010 6:55 am

Great! Also can u change the luck during enemy turn? And what about AI, does he have partial luck too?

User avatar
ALAKTORN
Location: Italy

Re: Mistake on CO Guide

Post by ALAKTORN » Mon Sep 27, 2010 7:35 am

fcastro wrote:Great! Also can u change the luck during enemy turn? And what about AI, does he have partial luck too?
AI does have either odds or evens, I think it works the same for AI and for yourself (so the CO stats posted previously are true for AI too)

about luck on enemy's turn, I doubt it's possible to change it from a set to another, but I can't prove it impossible so there is always a chance it could be possible

User avatar
Xenesis
Tri-Star CO
Tri-Star CO
Rank: Hydrocarbon Inspector
3DS Code: 2535-4646-7163
Location: 0x020232DD
Contact:

Re: Mistake on CO Guide

Post by Xenesis » Mon Sep 27, 2010 8:46 am

Makes it sound like it is more similar to Fire Emblem's RNG - you could do some crazy crap with it by simply jiggling the cursor to use up Random Numbers.
IST wrote:Even the worst individual needs to discover the joys of a chicken statue that is also a pregnant blonde housewife.

User avatar
ALAKTORN
Location: Italy

Re: Mistake on CO Guide

Post by ALAKTORN » Mon Sep 27, 2010 9:19 am

Xenesis wrote:Makes it sound like it is more similar to Fire Emblem's RNG - you could do some crazy crap with it by simply jiggling the cursor to use up Random Numbers.
do you know how to find AWDS's RNG on emulator? it would probably help a lot in deducing things but I don't know how to find it

User avatar
ALAKTORN
Location: Italy

Re: Mistake on CO Guide

Post by ALAKTORN » Thu Sep 30, 2010 6:10 am

Julian made the guess that the RNG gives all luck numbers when your luck ends with odds, bad and good are added together: Sonja +10/-5 = +5 (or +15?) etc, this theory works for all the tests made so far, Nell SCOP (100 luck) gives a set of luck

edit: nevermind, Sonja/Sami (+5 TAG) still have all numbers…

edit2: new theory: if either of your luck is odds (Sonja/Sami +15/-5 odds/odds; Jugger/Kindle +35/-15 odds/odds; Jugger +30/-15 evens/odds; Flak +25/-10 odds/evens) you get all numbers, Flak+Lash (+30/-10 evens/evens) you get stuck on odds luck without possibility to change it, tested with Recon vs Recon, Tank vs AA, Tank vs Tank and Tank vs BC always on Road full HP

User avatar
ALAKTORN
Location: Italy

Re: Mistake on CO Guide

Post by ALAKTORN » Sat Nov 13, 2010 4:37 am

news: AI always gets all numbers on its day, tested on random War Room map, Colin Inf vs Road Recon, made 10% 11% and 12%

AI has a set on our day, tested with 8HP Inf counter on Road 6HP Inf, damage dealt is 44, 46, 48, 49 and 51 only

we have all numbers in AI's day, tested with 10HP Mega counter on Inf, damage seen: 178% and 179%

in summary, AI's day has all possibilities and ours doesn't

Edit: Wrong info.
Last edited by ALAKTORN on Wed Jan 18, 2017 11:43 am, edited 1 time in total.

User avatar
ALAKTORN
Location: Italy

Re: Mistake on CO Guide

Post by ALAKTORN » Thu Nov 18, 2010 5:14 am

news: a 100% method of changing luck set is to simply use up an RNG number by moving to a square which requires RNG to determine the path

I've also been testing how AI's luck set works, and I think it changes at the same time as ours, it can be either the same as ours or different, so I think at the start of our day a luck set is randomly decided for us and the AI <-- Nitrodon (AW2 TASer) confirmed that AI's and our luck set change at the same time
Last edited by ALAKTORN on Tue Nov 30, 2010 2:51 pm, edited 2 times in total.

User avatar
ALAKTORN
Location: Italy

Re: Mistake on CO Guide

Post by ALAKTORN » Mon Nov 29, 2010 5:14 am

strange thing: Max's Artillery has all numbers of luck

edit: Nitrodon's (AW2 TASer) take on this is that all indirects get all luck numbers, and attacking them gives them all aswell, I've tested Jake's Arty and Rockets and it's true, haven't tested attacking AI's indirects though

User avatar
ALAKTORN
Location: Italy

Re: Mistake on CO Guide

Post by ALAKTORN » Wed Dec 01, 2010 4:09 am

RNG:

when RNG value is odds, it gives evens luck set and viceversa, direct units get a set, indirects get all numbers and RNG advances evens-odds-evens-odds etc

cheating the RNG value to 1 and attacking with indirect makes it go: 1 -> 10 -> 451 -> 815860 -> 708256005 directs: 1 -> 451 -> 708256005

Jugger with direct goes 1 -> 81560; 0 -> 451

RNG value in our attacks
0 = 1% luck
1 = 0% luck
2 = 7% luck
3 = 2% luck
4 = 5% luck
5 = 6% luck
6 = 5% luck, same as 4
7 = 2% luck, same as 3
8 = 7% luck, same as 2
9 = 0% luck, same as 1
10 = 1% luck, same as 0
893832000 = 1% luck, same as 0, only the last number counts; but this is never giving me 3, 4, 8, or 9% luck…

919450845 gives 8% luck… edit: with what did I test this? it's giving me 10% now with 20 luck, and 0% with 10

for counter damage:
1 1% luck
11 1% luck
21 1% luck
31 1% luck
41 1% luck
51 1% luck
61 1% luck
71 7% luck
81 3% luck
91 5% luck
101 7% luck
111 5% luck
(our attack damage was always 0%)

during Nell's SCOP (100 luck):
0 = 1% luck
1 = 10% luck
2 = 27% luck
3 = 52% luck
4 = 85% luck
5 = 26% luck
6 = 75% luck
7 = 32% luck
8 = 97% luck
9 = 70% luck
10 = 51% luck
11 = 40% luck
12 = 37% luck
13 = 42% luck
14 = 55% luck
15 = 76% luck
16 = 5% luck
17 = 42% luck
18 = 87% luck
19 = 40% luck
20 = 1% luck
21 = 70% luck
22 = 47% luck
23 = 32% luck
24 = 25% luck
25 = 26% luck

AI seems to have the opposite set as ours
Last edited by ALAKTORN on Thu Dec 09, 2010 6:00 pm, edited 1 time in total.

User avatar
ALAKTORN
Location: Italy

Re: Mistake on CO Guide

Post by ALAKTORN » Thu Dec 09, 2010 5:55 pm

16384 value makes the RNG overflow, the luck damage formula is ((x+1)(4x+1) mod 2^30) mod y where y = luck (100 with Nell SCOP, 10 normally, etc); it's 30 bits value stored in 4 bytes

thanks to Nitrodon and the MKPP guys who helped finding it

edit: I just noticed, the CO Guide says Flak +25% -10% ecc, shouldn't it be +25 -10 or +24% -9%?

more RNG advances

0 -> 1 -> 10 -> 451 -> 815860 -> 708256005 -> 884264318 -> 1321863 -> 308239976 -> 736035081 -> 706170482 -> 881537867 -> 496754780 -> 70247949 -> 318042342 -> 591982351 -> 924890832 -> 768947729 -> 469680858 -> 798381267 -> 594296772 -> 103118613 -> 479556174 -> 373389975 -> 751860024 -> 833244953 -> 384126274 -> 269325915 -> 108462380 -> 1006142493 -> 357712310 -> 7536159 -> 28644768 -> 551025697 -> 53730730 -> 112142307 -> 732927124 -> 482065701 -> 786517790 -> 925149607 -> 961477640 -> 311029033 -> 856515602

select unit: 0
positioning with possibility to attack: dir # of lucks for each possible combat, ind # of lucks for each possible combat * 2 alt: # of lucks * units in range * 2
click Fire: # of lucks for each possible combat * 2; for ind this also considers adj E units that can attack the ind unit, but not E ind units if you’re in range of any
what I mean with possible combat: Andy Tank is adjacent to Andy Tank and Arty = 3 (Tank vs Tank = 2, Tank vs Arty = 1)
attack: # of lucks (advances once per frame)

moving+Wait doesn't advance the RNG at all, unless it's needed to draw a path
and drawing path is 1 advance

not fully tested: loading a map (ex: TiS M1) is 12 advances
loading main menu is currently a WIP
map finish, before COs dialogue is 864119517 advances with RN 0, different depending on old RN
COs dialogue is 1 advance

AI DAY
simply ending day for AI is 1 advance for each RNG calculation, except the first time which is probably based on frames
Lake Fever AI D1 test: 0 -> 1049063516 -> 34059789 -> 853049574 -> 844480271 -> 635586256 -> 684000785 -> 622088922 -> 1021781203 -> 953929668 -> 606422805 -> 334824014 -> 420366999

difference from AI D1 to AI D2 advance is 638-13=625

D2 is the same, 1 frame based advance+11 normal

frame based advance is
x = Frame -1 + (oldRNG+1)*(4*oldRNG+1)% 2**30
newRNG = ((x+1)*(4*x+1))% 2**30
newRNG = newRNG + (newRNG//2)% 2

After disassembling:
iRNG = RNGStep(iRNG) + GameTime; Edit: This is for the bottom screen only, top screen has an added “- 1” to this line for whatever reason (probably timing at which the GameTime is read), lol.
iRNG = RNGStep(iRNG);
if(iRNG & 2)
iRNG++;
return iRNG;
aka:
read, write, write, read, write, write
(read)
1. Step
2. + GameTime
(read)
3. Step
4. +1 if second to last bit is set

TiS M1 from main menu to good RN 427876124 advances

Menu advance formula in C++
unsigned __int32 RNGMenuAdv(unsigned __int32 iRNG)
{
for(int i = 0; i < 80; i++)
{
iRNG = RNGStep(iRNG);
if(iRNG & 2)
iRNG++;
}
return iRNG;
}
Last edited by ALAKTORN on Wed Apr 20, 2016 12:58 am, edited 38 times in total.

User avatar
ALAKTORN
Location: Italy

Re: Mistake on CO Guide

Post by ALAKTORN » Thu Dec 16, 2010 1:25 am

weird thing:

Image

(counterattack 2 is counter b4 def calculation) AI estimates 0-88 but attacks anyway with Inf, what I can gather from tests and those numbers is that the Inf estimates the battle from Road (lowest def terrain?), but then attacks from City instead for some reason

edit: the NE Inf could attack from Road, but never does, he always estimates from Plains… I don't get it

User avatar
ALAKTORN
Location: Italy

Re: Mistake on CO Guide

Post by ALAKTORN » Mon Mar 14, 2011 1:12 am

I can't believe I'm going back and seeing we had the RNG advances totally wrong for positioning and Fire

more stuff

Grit 1 unit Rockets/Art
2+2+1
Grit/Jake 2 units Art/Rockets
4+4+1
Jake 2 units 8HP Art
4+6+1
Jake 4 units Rockets
(7+1)+8+1
Jugger 1 unit Art/Rockets
4+4+2
4+6+2 9HP Art
Jugger 2 units Art
8+8+2
Jugger 3 units Art
12+12+2
###12+14+2 probably not full HP
Jugger 4 units Rockets
(15+1)+16+2
###16+18+2 Rockets/Art probably not full HP
Jugger 5 units Rockets
(14+6)+20+2
Jugger 5 units 1HP Art
(18+2)+24+2

edit:

Andy 9 units 4HP Rockets
(8+10)
then I used SCOP, 90%HP
(9+9)

does it take into consideration our HPs then…? /// I’ve been cheating my way around, I kept getting 9+9 even with having 2 AI units at 8HP, all 9 full HP, me at 90%HP and me at 40%HP… it doesn’t seem to be the HPs either… I think I attacked from Road in all these /// removing the SCOP gave me 8+10 again, with 40%HP and all 9 units full HP -> this just changed into 9+9 without me changing anything in the setup O____O and it’s not going back to 8+10 no matter how many times I try… -> I got 8+10 again after using SCOP and removing it through cheats, hmm… but again right after, so it’s not about the removal of SCOP through cheats >_< -> just had it go 7+11 after another SCOP removal cheat O_o and keeping the power usage address to 0 always makes it go 7+11… oh god… -> I just had it go 9+9 then 7+11 while keeping power address to 0 and HP to 90%, changing nothing else… this is weird
Last edited by ALAKTORN on Tue Feb 28, 2012 12:23 pm, edited 11 times in total.

User avatar
ALAKTORN
Location: Italy

Re: RNG Discussion

Post by ALAKTORN » Thu Aug 11, 2011 10:12 am

we should change the topic’s title since I hijacked it

so we found a little something more, when AI asks the RNG on what unit to build, if the choice is between Inf and Mech, the current RN will be advanced once and the last 2 digits of the advanced RN will be considered, 0-19 is Mech, 20-99 is Inf (ex. RN 0 before build, advances once to 1, 1 is between 0-19 so Mech built)

btw congratulations to STL who figured it out by himself

User avatar
-STL-

Re: RNG discussion

Post by -STL- » Sat Aug 13, 2011 4:50 am

Why did the topic title change too on first post? XD

Anyway so capping Nmost City is 0-49 and Smost City is 50-99 of the last 2 number of advanced RN.

User avatar
ALAKTORN
Location: Italy

Re: RNG discussion

Post by ALAKTORN » Sat Aug 13, 2011 5:23 am

-STL- wrote:Why did the topic title change too on first post? XD
because I’m powerful B)

edit: I don’t know if it’s written elsewhere or whatever, but I’m just gonna say: “sets” are just the RNG advancing an evens amount (2, 4…), 10 luck CO Tank vs. 10 luck CO Tank advances the RNG twice, once for our luck, then for our opponent’s, and if the RNG advances twice it will always keep the same set; but if you change that test to Jugger on one side, the RNG advances thrice and it will alternate from one set to the other set

Slazzy

Re: RNG discussion

Post by Slazzy » Wed Nov 27, 2013 2:03 am

why is this topic called rng discussion if the original post is about nell's luck? he made that title, (stl)
..........
--[Olaf]--

Description: What do you get if Santa Claus dyes his beard and changes
his clothes? OK, enough jokes,

User avatar
ALAKTORN
Location: Italy

Re: RNG discussion

Post by ALAKTORN » Wed Aug 27, 2014 4:51 am

Dumping.

Counting after frame-based advance:
162f 1 wave 5 steps 1-2 advances in quick succcession, refuse.
165f 2 waves 10 steps 1-2 advances in quick succession, refuse.
There’s a big pause after the first 2 waves for some reason.
203f 3 waves 15 steps 1 advance, pause, 1 advance during or after attack.
There is no 4th wave.
6 advances total. All were on a custom map with only 1 attack possible, Colin Recon on Kanbei Mountain Inf.

Frame based advance suicide attack:
109688965 = 15 Neo 15 Mech esti 15 36 3% 24% dmg
352142917 = 14 Neo 17 Mech esti 1 39 17% 21% dmg
780779473 = 15 Neo 16 Mech esti 8 38 10% 22% dmg

4 steps take 4 numbers: atk pos luck, atk neg luck, def pos luck, def neg luck. Don’t know why we thought they were shared or something.

STL

Re: RNG discussion

Post by STL » Thu Aug 28, 2014 5:04 am

Alright trying to figure out what the AI does on its day.

Playing m1 and just ending day gives the following advances:

Playing CO without bad luck: 7,50,53
Playing CO with bad luck: 9,66,70

Not sure what the first advance is.

The second advance is the estimation phase.
it's 43 = 14 * 3 + 1 and 57 = 14 * 4 + 1 steps.
Not sure why 14 estimations, I have 16 units that are all attackable and then there's the extra step.

The third advance is the actual attack, at least that's clearly identified and we can calculate the actual outcomes.


More results:

Bad luck CO m1:
Day 1: 9,66,70 (14*4 + 1)
Day 2: 9,61,65 (13*4)
Day 3: 9,58,62 (12*4 + 1)
Day 4: 9,53,57 (11*4)
Day 5: 9,50,54 (10*4 + 1)
Day 6: 9,33,37 (6*4)
Day 7: 9,33,37 (6*4)
Day 8: 9,37,41 (7*4)
Day 9: 7,34,38 (6*4 + 3?)
Day 10: 9,25,29 (4*4)
Day 11: 9,17,21 (2*4)
Day 12: 9,16,20 (1*4 + 3?)
Day 13: 6,17,21 (2*4+3?)
Day 14: 6,9,13 (0*4+3?)
Day 15: 9,13,17 (1*4)
Day 16: 5,9

Post Reply

Who is online

Users browsing this forum: Kim Jong-un [Bot] and 2 guests