Submission Text
ChuChu Rocket! is a simple 2D game where you have to place arrows on the floor to lead mice and cats to complete the goal.
This movie completes the Challenge mode or Stage Challenge as fast as possible. There are 25 stages which have to be completed under a time limit of 30 seconds.
They can have one of five different goals which are explained later:
• Get mice
• Run away
• Lunch time
• 100 mice
• Cat soccer

### Game objectives

• Emulator used: VBA-RR v24 svn422
• Completes Stage Challenge
• Aims for the fastest time
• Luck manipulation (in the last stage)

If a cat or a mouse runs into a wall it goes right. If then right is a wall they go left. If then left is a wall they go backwards.
In the Challenge mode you can place a maximum of three arrows and as soon as you want to place a third one, the one you placed first disappears.
Cats eat mice once they are touching.
A player can't overwrite an arrow of the other player

### Goals

#### Get mice

Get the mice into the rocket without losing a single one.

#### Run away

Basically the same as Get mice except that there are cats which you have to avoid.

#### Lunch time

Let the cat eat all the mice. And with all I mean all, because if a single mouse falls into a pit you failed the stage.

#### 100 mice

You have to lead 100 mice into the rocket to finish the stage

#### Cat soccer

You play against a (red) computer which can also place arrows. The goal is to lead 20 cats into the rocket of the opponent.

Nothing special

#### B1 (03.70)

The cats weren't even in the way

#### C1 (03.68)

Leading the mice into the cats as fast as possible

#### D1 (09.08)

The limitation of three arrows is a bad thing here

#### E1 (11.41)

The AI is confused if I place my arrow where he wants to... :)

THREE ARROWS AAH

#### B2 (09.11)

I'm basically trying to exchange the positions of cats and mice

#### C2 (03.85)

Nothing special, just fast eating

#### D2 (11.00)

I was surprised because letting the cat go through the mice parade a bit was actually faster (also some strange graphic glitch at the end :>)

#### E2 (16.80)

Cats move soo slowly...

#### A3 (05.00)

Suddenly everyone is going towards the rocket :D

#### B3 (07.81)

The cats were a bit of a problem here

#### C3 (08.20)

Hmm, yeah, nothing special

#### D3 (18.93)

Oh, this stage... I tested many possible strategies, but I ended up filling the first row with 100 mice (which is the maximum there can be on the screen), because it is the fastest way

#### E3 (30.00)

The rockets are really unfairly placed (where in E4, it is unfair for the computer). As you can see, the last two cats are almost there when the time ends ._.

#### A4 (08.68)

This was an easy level :>

Nothing special

#### C4 (07.20)

Turned out to be really fast

#### D4 (13.98)

Same strategy as in D3, filling up a loop and then leading them to the rocket

#### E4 (21.23)

Unfair placement of the rockets for the computer :>, easy win for me

#### A5 (16.70)

I had to test several lengths of the mice team to get the fastest, there are also some problems with the THREE ARROWS

#### B5 (06.30)

Fast and easy stage

#### C5 (11.31)

Leading mice and cat together as fast as possible

#### D5 (24.43)

Completing this stage in realtime is hard as hell because you have to hope that there is no cat spawning in your mice collection

#### End (08.08)

The Final Stage is a special stage, because the goal here is to have 100 more points than your opponent. There are 50-mice which give you 50 points. There are also ?-mice which can do different things, for example Mouse Mania, which spawns many mice, or Cat Mania, which spawns cats. It is random what is chosen, but it is time based so that is what was the best I could do.

###### TASVideoAgent
They/Them
Experienced Forum User, Moderator
This topic is for the purpose of discussing #3817: Masterjun's GBA ChuChu Rocket! "Challenge mode" in 07:01.06
###### Mr. Pwnage
Experienced Forum User
This run fails to beat the unassisted records, some by significant margins.
###### Masterjun
He/Him
Experienced Forum User, Site Developer, Skilled player (1821)
I see no proof. Beat my movie and then I'm going to cancel this.
###### Tangent
Experienced Forum User
###### Masterjun
He/Him
Experienced Forum User, Site Developer, Skilled player (1821)
Yep, that movie is faster in exactly one room. I'm going to cancel this, fix that, and then submit again.
###### TASVideosGrue
They/Them
Experienced Forum User
om, nom, nom... sweet!
###### Tangent
Experienced Forum User
A lot more than just one.
###### Masterjun
He/Him
Experienced Forum User, Site Developer, Skilled player (1821)
You can't count all the RNG based things, because this movie doesn't aim for the fastest times, but for the fastest credits!
###### Mr. Pwnage
Experienced Forum User
If a TAS on this game is to have any purpose whatsoever, it's pretty much required to manipulate luck to the fullest in the D and Final stages, as A/B/C/E stages are completely deterministic and even the unassisted records treat them as practically turn-based puzzles from which they're easy to analyze. I'd assume that by now, the records for every luck-based stage can only be set with the most favorable 0.1% of random seeds, if not less. The fact that the final stage here takes four times as long as the unassisted record, and in fact ranks behind all but two of the submissions on Cyberscore, is a fundamentally disqualifying flaw. It's not particularly fast, nor does it offset that with any kind of style points. Now, in the previous discussion topic (see the link to unassisted records in a prior post), it was found that the RNG manipulation available in this game is minimal, and consists exclusively of waiting and brute-force restarts. This seems to imply, if anything, that the game makes a poor candidate for TASing, at least for a single-segment stage challenge run (and the other modes aren't going to be any better). Either that, or the run needs to focus on cracking the RNG directly, as has been the focus of many RPG titles recently.
###### Masterjun
He/Him
Experienced Forum User, Site Developer, Skilled player (1821)
The good thing is, that my submission is not going to be rejected because of the "records on Cyberscore" because there is still a chance that times could be faked. There is no video proof of any of the times. To RNG manipulation: I found out that the address for the RNG is 030001F8 (4 byte) and the next RNG is calculated like this: newRNG=RNG*0x41C64E6D+0x00003039 So in the worst case you would have to wait 4294967295 RNG changes to get the best time on one stage. So if the RNG would change 6 times per frame, it would take about 138 days to get the best time...
###### Tangent
Experienced Forum User
That's... uh... not how math works.
###### Masterjun
He/Him
Experienced Forum User, Site Developer, Skilled player (1821)
Tell me where I'm wrong