(Link to video)

Game objectives

  • Emulator used: Bizhawk 2.3.0
  • Bios: SCPH1000
  • Hardest Difficulty
  • Tournament Mode
  • Aims for fastest time
  • Heavy Luck manipulation
  • Genre: Fighting

RNG

RNG calculation formula

RNG = ( PrevRNG * 0x41C64E6D + 0x00003039 ) & 0xffffffff RNG[0] = 0x00000000

Luck manipulation

When a small particle appears, RNG is increased twice per frame. When a red ball appears, RNG is increased 4 times per frame.

Voice RNG

There is 4 kinds of voice for a screen of YOU WIN, and there is a great difference in loading. It's important that I keep choosing shortest Voice#1 from those. Voice = (RNG >> 16) & 0x0003

Early input finish

Although there is a name entry on the last screen, I thought that naming is unnecessary. I think that it is more fun to finish the input earlier.

About details

When using a special attack, Chuji squats, so he can avoid the upper row attack.

OPTION MODE

I shifted upward to 1893-1894 1684-1685 twice by using the second controller and shortened one frame.

Stage 1

Chuji avoids high punches when charging. Special attack has damage three times the charge time. He deals 1.5 times more damage than normal by "counter hit". I got the fastest "counter hit". 57 x 3 x 1.5 = 256 I advanced RNG 10 times with kick particles for the next stage.

Stage 2

The attack he can't avoid when he behaves fastest, would be done, so I'm to consume a luck manipulation and 3 frames too much, and I'm making him choose defence. I advanced RNG 78 times by red punch for the next stage.

Stage 3

The posture he's charging can avoid a short kick. I got the fastest "counter hit". I advanced RNG 62 times by red punch for the next stage.

Stage 4

Luck manipulation is becoming difficult because the probability that he does ideal behavior is low. I got the fastest "counter hit". I advanced RNG 82 times by red punch for the next stage.

Stage 5

Since lightning occurs at this stage, RNG is always running. I got ideal behavior of FEI by luck manipulation and backward movement of 3 frames. I advanced RNG 6 times, and furthermore I advanced RNG one more time by waiting one frame on the screen of YOU WIN for the next stage and the future.

Stage 6

I got her more ideal behavior by luck manipulation and back dash. If you hit "counter hit" in the air, 1.2 times more damage will be given. 48 x 3 x 1.5 x 1.2 = 259 I advanced RNG 4 times.

Stage 7

The ideal RNG at this stage was only once in several hundred times. I got the fastest "counter hit". I advanced RNG 122 times by special attack.

Stage 8

Because he did not have perfect ideal behavior, I spend a little time for a "counter hit". I advanced RNG 78 times.

Stage 9

I waited one frame at the start of the fight and backed it three more frames. And with the luck manipulation I got the action of an "air counter hit." 48 x 3 x 1.5 x 1.2 = 259 I advanced RNG 40 times.

Stage 10

I got the fastest "counter hit". I advanced RNG 48 times.

Stage 11

Because I could not get the fastest "counter hit", I spent a bit of latency. However, RNG which can obtain this behavior was very rare. I advanced RNG 12 times. I advanced "InGameFrame" 4 by waiting 2 frames of "YOU WIN" skip, but this time the start time of the next stage is the same.

Stage 12

NORK is tough as it has resistance to special attacks. I got RNG which can obtain "Bravo Uppercut" many times. There is a small gap in the hit box of "Bravo Uppercut", so you can go through the attack by adjusting the dash. The movie is finished early by long jump low kick, but the adjustment seemed like hell.

Stage 13

Even with the same random number, UDAN behaves differently depending on the "InGameFrame" difference. My TAS tudio has always been frozen in the middle of this stage. So I had to do a lot of input patterns in the end by hand.

Others

If you place importance on "In Game Time" it will be less than 20 seconds. The movie that made the "time in game" fastest became less than 20 seconds. https://video.twimg.com/ext_tw_video/1078646264701415425/pu/vid/560x480/u_xDzeEe_8CrB9_9.mp4

Stage Times

STAGE Character Input Time Game TimeGame Time Total
STAGE 1 EPON 1.39 1.39 1.39
STAGE 2 CHUJI 1.46 1.46 2.85
STAGE 3 OLIEMS 1.40 1.40 (Same as 1.39) 4.25
STAGE 4 HOM 1.39 1.39 5.64
STAGE 5 FEI 1.46 1.46 7.10
STAGE 6 MARY 1.35 1.35 8.45
STAGE 7 ILL 1.39 1.39 9.84
STAGE 8 DOCTOR V 1.50 1.50 11.34
STAGE 9 CHACO 1.31 1.31 12.65
STAGE 10 GREN 1.39 1.39 14.04
STAGE 11 MUFU 1.44 1.44 15.48
STAGE 12 NORK 1.91 5.36 20.84
STAGE 13 UDAN 0.00 9.41 30.28

Lua Script and Aviutl Project

Special Thanks:

Spikestuff

ThunderAxe31: Judging.
ThunderAxe31: File replaced with a 207 frames improvement.
ThunderAxe31: This movie is nicely optimized, thanks to the RNG manipulation for each fight.
It's fun to see the last opponent committing suicide, and the reception was almost completely positive.
Accepting for Moons.
fsvgm777: Oh boy, more 480i. Processing.

Judge, Skilled player (1289)
Joined: 9/12/2016
Posts: 1645
Location: Italy
hopper wrote:
It seems like the decision was affected by a lack of feedback during the judging process?
With 10 votes (9 yes, 1 no) and 4 reaction posts (3 positive, 1 negative) at the moment of the verdict, I wouldn't say that there was a lack of feedback. What I meant in my previous post, is that I'm considering the chance that the specific users who gave the feedback, could happened to not have been a sample indicative of the opinion of the whole community, which could have misled me into doing a wrong decision. Edit: Published movie currently has 6 ratings, with an average entertaining rating of 3.2. For this reason, I'm now changing the Tier to Vault.
my personal page - my YouTube channel - my GitHub - my Discord: thunderaxe31 <Masterjun> if you look at the "NES" in a weird angle, it actually clearly says "GBA"
Site Admin, Skilled player (1236)
Joined: 4/17/2010
Posts: 11274
Location: RU
hopper wrote:
I happened to visit the front page, and I was intrigued when I saw a fighting speedrun in Moons. That's why I watched it, and that's why I commented. It seems like the decision was affected by a lack of feedback during the judging process?
Sometimes I use this strategically. I accept borderline movies to Moons if it's unclear, more people see them, more people rate, it becomes clearer.
Warning: When making decisions, I try to collect as much data as possible before actually deciding. I try to abstract away and see the principles behind real world events and people's opinions. I try to generalize them and turn into something clear and reusable. I hate depending on unpredictable and having to make lottery guesses. Any problem can be solved by systems thinking and acting.