Puzzle Bobble 4 is a game released by Taito in December 1997 for Arcades, and was later ported to the PlayStation, Dreamcast, Game Boy Color, and Microsoft Windows throughout 1998 to 2000. It is the fourth installment in the Puzzle Bobble series.
Cleon speedrun comparison:

Game objectives

  • Emulator used: Bizhawk 2.8
  • BIOS used: SCPH-5501.bin (U)
  • Heavy luck manipulation
  • Aims for fastest time
  • Aims for bubbles
  • Takes damage to save time
  • Uses highest difficulty
  • Uses a suboptimal character
  • Genre: Puzzle

Game Description

Bust-A-Move 4 is a tile-matching puzzle game in which players control a launcher at the bottom of the screen that fires colored bubbles toward a field of bubbles suspended from the top. The objective is to clear bubbles by forming groups of three or more of the same color. Once matched, bubbles pop and any unsupported bubbles beneath them fall, potentially creating chain reactions. The launcher can be aimed left and right, and bubbles may be bounced off the walls to reach difficult angles. The next bubble in the queue is visible, allowing for limited planning.
The game can end in two ways: either the player clears all bubbles on the screen, or the ceiling slowly descends until bubbles cross the bottom line, resulting in a loss.

Chain reaction

Chain reactions are kept quite simple in this Tas due to how long it takes before they finish.

Player vs Computer Story Mode

This Tas focuses on the story mode of Player vs Computer. We play this mode on expert difficulty and change the difficulty in the options from "normal" to "very hard" for additional difficulty.
12 Opponents need to be defeated.

Character choice for the tas and information

Bub/Bob:

Damage: Normal
Bub/Bob have the most basic damage layout that is in the game. It is mostly used as reference.

Alkanet:

Damage: Low
Alkanets damage output is quite low due to the fact that her bubbles are quite hard to counter. Rainbow bubbles become the bubbles color that are popped nearby, but sadly deal no damage to the opponent when it happens. They are useful in blocking the opponents grid.
The star bubble can be popped to eliminate all of that same bubbles color it hit on your side, which can come in clutch.

Marino:

Damage: Low
Rainbow bubbles again but mixed with green ones this time, but no star bubble. Still a good choice for tas.

Kurol:

Damage: Low
Kurols damage is low but his damage layout is the most useful in tas. On the bottom we can see that he gives the enemy only rainbow bubbles if he deals low yellow damage. If the damage becomes red, it will pull down their screen by 1 immediately after they have fired their shot. This is indicated on the character screen here by the yellow arrow on the top pointing downward, the bottom arrow pointing upward is normal damage. Kurol, Dreg and Packy are the only characters with 2 yellow arrows.

Tam Tam

Damage: Very low
Tam Tam just has 0 damage output when it comes to yellow junk. The only way to deal damage with him is to give red/blue damage to your opponent.

Cleon

Damage: Normal
Cleons damage output is quite chaotic. She can throw all sorts of bubble to the opponent. The new "bubble" you see in her layout is a wooden block, which is annoying to get rid of. The block can only be removed with a combo or chain reaction, its very annoying. If the wooden block reaches the bottom of the screen, it will simply fall off instead of making you lose.

G

Damage: High
G's bubble layout does not give his opponent orange bubbles, which means out of all 5 bubbles in this mode (red, blue, yellow, green, orange) you will only have to deal with 4 of them, meaning an easier time for the opponent. This also means that they gave G a slightly higher damage output.

Bramb

Damage: Normal
More woooden blocks on Brambs layout, which should normally mean that his damage output should be set to low, but we are missing orange bubbles again? Good choice for Tas.

Develon

Damage: Normal
Nothing to say about Develon, its the same as Bub/Bob.

Gigant

Damage: Very high
Extreme damage coming from Gigant. The fact that you only have to deal with 3 bubble colors from his layout makes it easily counterable. It could be a very good choice for tas if you manipulate the opponents grids and shooters beforehand.

Why i pick Cleon for the Tas

Cleons ability to give opponents completely different bubbles each time plays a huge factor in making the tas. If you were to play characters like Bub/G/Develon/Gigant you will always get a predictable damage pattern which the CPU is more likely to solve. That does not mean that any of these characters are bad for tasing, theres just less ways to get good outcomes.
Cleons damage output is also not really completely random. For example, every 2 red stack damage a star bubble will appear on the opponents side which they can use to retaliate quite heavily by clearing the whole screen with it. The wooden blocks and rainbow bubbles appear every 1 red stack.

Ranking all characters for tas

9. Tam Tam
8. Bub/Bob and Develon
7. G
6. Gigant
5. Bramb
4. Marino
3. Alkanet
2. Cleon
1. Kurol

Strategy

The strategy should be obvious: Create a lot of chain reactions and combos and deal so much damage to the opponent that they will be overwhelmed and lose. Many people would believe that creating a very big chain reaction at the start of the round would be the most efficient and fastest way. There are a few flaws with this strategy though:
1. It takes too long to set up.
2. Creating a chain reaction big enough to make the opponent lose afterwards is not always possible, they may be able to counter.
3. The risk of getting damaged yourself by the opponent, ruining your chain reaction.
4. Opponent already cleared their screen before you start your big chain reaction.
5. Chain reaction is not big enough due to grid layout and shooter bubbles (might just be the 2nd point).
So what is the most optimal way to defeat the opponent?
Well we are still doing chain reactions, just not a big one. The main focus should not lie on our screen, we also need to look at what the opponent does. Each time the opponent fires a shot is a possibility for them to take damage and have their screen filled with bubbles until they lose. So what we have to do is have them take damage as soon as they fire, done by creating small towers and opening up a good grid we manipulated before.

Towering

This is a very good strategy for all of the bust a move games. Create a tower by firing different colored bubbles onto preferably 2 bubbles with the same color so that you can later pop those 2 bubbles to destroy the tower, dealing damage to the opponent. If there are different colored bubbles on the 1st layer with none of them having a matching bubble next to it, even better. No extended chain reaction will occur this way and we are getting very fast damage.
Remember that the grid we start with is the same grid the opponent starts with. Since the CPU does not know how to tower, they will shoot their bubbles onto matching bubbles. We do not get damaged like this.
If the grid spawns with a big cluster of matching bubbles underneath the 1st or 2nd layer it would be even better, because we can make a tower, collapse it, and then do another quick combo for additional damage. It keeps adding up.

Examples of a good starting grid for tas

Examples of a bad starting grid for tas

How to manipulate the starting grids and RNG

Manipulating the starting grids is easy. Before moving on to the next CPU on the universe map, you can delay the x input by 1 frame and already have a different grid for the battle. You can delay the x input in the meeting scene of you and the opponent aswell to change the grid.
One small issue.
There are only a couple of starting grids for each opponent. I've counted roughly 10+ possible starting grids for each individual opponent, limiting our ways to get fast defeats. Before picking your character and pressing x on the movie dialogue, we can change the grids too, but only viable for the 1st CPU. Out of the 10+ grids only 2-3 always seem good enough for tas.
For the Battle RNG, delaying your shots can give both you and the enemy different bubble damage, though miscellaneous bubbles like rainbow bubbles, wooden blocks and stars are always consistent. Delaying shots can spawn different colored bubbles that appear after firing 10 shots on top of the screen.

Unused Ram addresses

192EB8 - Current shooter bubble color
  • 1 - Blue
  • 2 - Red
  • 3 - Green
  • 4 - Yellow
  • 5 - Orange
188BE0 - Next shooter bubble color
192EBC - Shooter angle
192EB6 - Bubble state - 64 not shot - 128 fired - 0 Bubble locks into position
192EC2 - Bubble Position
167BBE - Round won animation
167A14 - Round won or lost
I could not find the address for the pre battle rng and battle rng, but i dont think it would have mattered anyway.

CPU Breakdown

The first 4 of the 12 CPU's have a very similar way of shooting their bubbles onto the grid, with the exception of Alkanet who likes to miss her shots. They will always avoid stacking towards themselves and lay the bubbles next to each other.
1. Bob/Bub:
Slow shooter but precise, takes so long to lose.
2. Alkanet:
Faster than Bob/Bub but terrible at shooting.
3. Marino:
On other difficulties, Marino can still miss his shots apparently.
4. Kurol:
Very fast shooter, can still miss his shots on other difficulties too?
The last 8 of the 12 CPU share the same way of shooting their bubbles onto the grid, with the exception of Tam Tam who is quite fast and shoots the bubbles next to each other like the first 4 CPU's. Cleon - Dreg always make a small tower like tas if none of the colors that are in their shooter match the ones on the grid.

Stage by stage comments

1. Bob - 19s

For Bob i chose a grid that let me set up a chain reaction extremely fast. Bob only gets 2 shots in before he starts taking damage. Bob can be improved by 1-2s if you find a grid that has atleast 9-10 bubbles of the same color that can be popped on the first shot, Bob will take damage after the first shot like that.

2. Alkanet - 14s

Basic. A normal grid that lets me get chain reactions fast. CPU's always prefer to aim at the star bubble should it be poppable. This lets you guide the CPU to stay somewhere near the middle for the most timesave, in exchange for a small timeloss of the star bubble being popped. Alkanet shoots to the right at the end with only some bubble damage remaining. If the CPU has too much bubble damage left then you will have to wait for all of the bubbles to finish travelling on their screen before they actually lose.

3. Marino - 14s

Same fate as the previous opponent, even the same time. Not sure how to improve it except for when Marino has to go towards the middle with the green bubble. But the damage was already set in stone.

4. Kurol - 13s

I had a lot of trouble optimizing Kurol because i just could not find a good starting grid for him. In the end i settled for one that had a lot of blue in it. Kurol never managed to get his blue bubble colors so it was a good risk to take.

5. Tam Tam - 10s

Now we are going a lot faster with the CPU changing their behaviour. Because the CPU likes to make small towers if there are no matching colors between the shooter and the grid, i need to block most of their layers with my bubble damage. This is easier said than done, as most grids never fulfill both criteria of having a good starting grid and unfavourable shooter bubble colors for the CPU. This is why some time is lost on Tam Tam when he shoots the red bubbles, otherwise a time of 8 second is possible. Tam Tam also follows the Cleon - Dreg behaviour for some reason, not the Bub/Bob - Kurol behaviour.

6. Cleon - 9s

The yellow bubble shot is delayed, otherwise the enemy Cleon will spawn a possible chain reaction on top of the screen and clear all of it.

7. G - 10s

I hate how slow G is, but all of his grids were just terrible to deal with. I chose this grid and even though i never play with a bad grid like this it somehow worked out. G does not completely clear his screen with all of the red bubbles and i manage to defeat him with a last single star bubble to finish.

8. Bramb - 7s

Oh how i wish all of the opponents went like this. We get 2 chain reactions and block all of brambs grid colors. Just amazing.

9. Develon - 8s

No optimal starting grids for Develon. I was stuck on 11s for a long time until i revisited several grids a few times and decided to try with this unfavorable one. Turns out that even ones that look bad initially can turn out to be quite fast.

10. Gigant - 9s

When i speedran this game my personal best was a 10s on Gigant, so i was under a little bit of pressure. Gigant didnt have many good grids and i decided to play this simple grid. Its not the fastest but i couldnt find a better outcome.

11. Madame Luna - 9s

Madame Luna has a lot of time waste when she goes to pop the orange and red bubbles. I hate repeating myself but the grids for her just werent any good. So 9s it has to be.

12. Dreg - 9s

Dreg has no meeting scene so you would think that there can only be 1 grid, but thats wrong. Right before Madame Luna loses, you can shoot a few more bubbles and they will change the grid for Dreg. Only 4-5 grids were available for him and i couldve gotten quite unlucky with a double digit time.

Final words and thoughts

To be honest i never thought i would make a full Tas of Bust a Move 4 because of how daunting the RNG Manipulation is. When i finally commited to it it turned out to not be that difficult. This Tas is optimized to the best of my ability.
Can you improve the Tas?
Yes, if you change the selection time of your character you may get different and better grids for the later half of the game, but its risky.
Thanks for watching my Tas and i hope you like it.
If the Tas gets accepted, i would like to have this userfile for the encode as publication, because it shows the score screen with me writing something and getting the game over.
Now you belong to me.

eien86: Claiming for judging.

eien86: A cute little puzzle game that offers a mode that allows you to 1v1 fight against another player. This movie beats the player vs. computer story mode in expert difficulty. It reaches the end significantly faster than the current RTA WR.
In terms of categorization, this game has other modes (Puzzle - Arcade, Puzzle - Story) that are speedran , so the mode needs to be specified in the goal. Then the difficulty is also an important distinction, although expert ought to be the default in my opinion, as it offers the maximum entertainment potential. The character selected (Cleon) does not seem to make a significant difference in gameplay to justify this distinction, altough this can change if properly justified.
Accepting to Standard
fsvgm777: Replacing movie file with one that includes a cycle count.

inconsistent: Processing. r3gamerz is handling the encodes for this one.


TASVideoAgent
They/Them
Moderator
Location: 127.0.0.1
Joined: 8/3/2004
Posts: 17364
Location: 127.0.0.1
GJTASer2018
He/Him
Location: Stafford, NY
Joined: 1/24/2018
Posts: 404
Location: Stafford, NY
While the gameplay itself was impressive, seeing the same loading screen over and over again drags down the entertainment value of the whole run, which is why I voted "Meh". I know this is the OG Playstation here, but you'd think Taito would've at least tried to add some variety to the characters on the loading screen considering how often it gets used in this game...
c-square wrote:
Yes, standard runs are needed and very appreciated here too
Dylon Stejakoski wrote:
Me and the boys starting over our games of choice for the infinityieth time in a row because of just-found optimizations
^ Why I don't have any submissions despite being on the forums for years now...
Experienced player (564)
Location: Pyra
Joined: 8/22/2022
Posts: 71
Location: Pyra
Hello GJTASer2018, you have already complained about loading screens on my Pop'n Tanks! submission. which i can sort of understand. Voting "Meh" on a tas somebody worked hard on just because of 6s of loading screens before battles is very rude to the taser. I have put timestamps in the description for each battle so people dont have to sit through these very long loading screens.
Skilled player (1396)
🇧🇷 Brazil
Joined: 2/5/2012
Posts: 2012
Location: 🇧🇷 Brazil
64 has a version of this game as well you should TAS that since it's less loading screens lol yes vote
I want all good TAS inside TASvideos, it's my motto. TAS i'm interested: Megaman series, specially the RPGs! Where is the mmbn1 all chips TAS we deserve? Where is the Command Mission TAS? i'm slowly moving away from TASing fighting games for speed, maybe it's time to start finding some entertainment value in TASing.
Former player
Location: Zurich, Switzerland
Joined: 6/30/2010
Posts: 1135
Location: Zurich, Switzerland
somyeol wrote:
Hello GJTASer2018, you have already complained about loading screens on my Pop'n Tanks! submission. which i can sort of understand. Voting "Meh" on a tas somebody worked hard on just because of 6s of loading screens before battles is very rude to the taser. I have put timestamps in the description for each battle so people dont have to sit through these very long loading screens.
The three voting choices are simply a measure of how entertaining one finds the TAS. And that includes things completely outside the TASer's control, it's just something you have to accept. No viewer judges just the player inputs alone, it's always about the bigger picture, everything you see and hear in the playthrough. He even explained his reasoning, so there's nothing disrespectful about this. As for me, I can see the problem with the loading, but I found it pretty harmless in this game. I saw a fast paced, colorful puzzle game that was also short enough to not become repetitive, so I'm voting "yes".
Current project: Gex 3 any% Paused: Gex 64 any% There are no N64 emulators. Just SM64 emulators with hacky support for all the other games.
Experienced player (564)
Location: Pyra
Joined: 8/22/2022
Posts: 71
Location: Pyra
Hello, i was looking at the Bob fight again and successfully saved ~100 frames with another starting grid. I have also noticed that the RNG for the other 11 CPU changes because of that. If the judge wants me to, i can work on the 11 other CPU's again and try to maintain the ~100 frame timesave, but its not really guaranteed.
DrD2k9
He/Him
Editor, Judge, Expert player (2447)
Location: US
Joined: 8/21/2016
Posts: 1181
Location: US
somyeol wrote:
Hello, i was looking at the Bob fight again and successfully saved ~100 frames with another starting grid. I have also noticed that the RNG for the other 11 CPU changes because of that. If the judge wants me to, i can work on the 11 other CPU's again and try to maintain the ~100 frame timesave, but its not really guaranteed.
Since RNG changes everything else, I'd say the potential improvement doesn't prevent this submission's acceptance. The run as-is appears appears pretty optimal to a general viewer. My main reasoning for this conclusion is that there's no guarantee that the 100 gained frames won't be subsequently lost due to RNG changes later on in gameplay. If you do choose to continue re-doing the rest of the run before a judge is able claims this one, then upload the improved version to userfiles and this submisison can be updated. But don't feel that this is necessary. Just from a single viewing of the encode provided, I'd say this run is probably acceptable as-is and any full-run improvements submitted in the future could simply obsolete this run.
Experienced player (564)
Location: Pyra
Joined: 8/22/2022
Posts: 71
Location: Pyra
DrD2k9 wrote:
somyeol wrote:
Hello, i was looking at the Bob fight again and successfully saved ~100 frames with another starting grid. I have also noticed that the RNG for the other 11 CPU changes because of that. If the judge wants me to, i can work on the 11 other CPU's again and try to maintain the ~100 frame timesave, but its not really guaranteed.
Since RNG changes everything else, I'd say the potential improvement doesn't prevent this submission's acceptance. The run as-is appears appears pretty optimal to a general viewer. My main reasoning for this conclusion is that there's no guarantee that the 100 gained frames won't be subsequently lost due to RNG changes later on in gameplay. If you do choose to continue re-doing the rest of the run before a judge is able claims this one, then upload the improved version to userfiles and this submisison can be updated. But don't feel that this is necessary. Just from a single viewing of the encode provided, I'd say this run is probably acceptable as-is and any full-run improvements submitted in the future could simply obsolete this run.
Redoing the rest of the run would be no problem for me if i wasnt doing a run with Gigant at the moment. The main reason for the potential sudden loss of the 100 frames gained on Bub could be a slower Bramb (7 second fight is absurd) and Madame Luna and Dreg, because you dont know how many grids there are for Dreg and you could easily get a 12-13s time.
Experienced player (564)
Location: Pyra
Joined: 8/22/2022
Posts: 71
Location: Pyra
Before a judge picks this up, im working on a Gigant (different character) Tas of this. It will take 1-2 weeks depending on how it goes. Maybe they can judge both of the submissions at the same time? This is not meant to be an improvement either, just a different character who is faster.
Post subject: Encoding...
r3gamerz
He/Him
Publisher, Player (7)
🇮🇩 Indonesia
Joined: 3/7/2025
Posts: 73
Location: 🇮🇩 Indonesia
Claiming to encode this submission.
Never stop innovating.
Post subject: Movie published
TASVideoAgent
They/Them
Moderator
Location: 127.0.0.1
Joined: 8/3/2004
Posts: 17364
Location: 127.0.0.1
This movie has been published. The posts before this message apply to the submission, and posts after this message apply to the published movie. ---- [6778] PSX Bust-A-Move 4 "Player vs Computer, Story, Expert, Cleon" by somyeol in 08:00.205