Kaizo Mario World is a well-known Super Mario World Hack, specifically for the reason that it was one of the hardest hacks of its time, and it even spawned a category of Super Mario World hacks known as Kaizo hacks. This run is of its sequel, Kaizo Mario World 2, and aims to complete the game as fast as possible.
An IPS of the version used in this TAS can be downloaded from: [dead link removed]
Game objectives
- Emulator used: Snes9x 1.43 v17
- Aims for fastest time
- Abuses programming errors
- Takes damage to save time
Stage by stage comments
Stage 1 Archway Mountains (ステージ１ アークレイさんち)
Done by Fethy75. After the midway point, the walljumping glitch is abused to pass over several sections that take too much time.
Stage 2 Mt. Silver (ステージ２ シロガネやま)
Done by Kaizoman666. This is a rather unexciting autoscrolling level, though I tried to keep it as entertaining as possible. At the end of the first room, walljumping and a glitch with screen scrolling is used to save time on the way to the pipe. A couple of glitches in the second room are also abused for entertainment, such as getting warped by the Hammer Bro platform.
Stage 3 Screw City (ステージ３ ねじまきシティー)
Done by Fethy75. By running into the wall to abuse the climb-anywhere glitch at the start of the level, everything was be easily skipped by simply jumping over it.
Stage 4 Queen Elizabeth (ステージ４ エリザベスごう)
Done by Fethy75 and Kaizoman666. Frames had to be lost in various places in order to manipulate the Boo Rings, the Fishing Ghost, and the Eeries. Screen scrolling is used in the first room to manipulate the Bowser Statue's fireballs.
Stage 5 Mount Everest (ステージ５ チョモランマ)
Done by Kaizoman666. The second room manipulated sprite spawning to reduce the amount of lag, at the cost of entertainment. In the third room, walljumps and damage abuse are used to skip waiting for the Digging Chucks' rocks.
Stage 6 Norfair (ステージ６ ノルフェア)
Done by Fethy75. Walljumps are abused in this level to save time and provide entertainment. Screen scrolling in the second room is used to manipulate the dolphins.
Stage 7 Sky City (ステージ７ とかいのそら)
Done by Kaizoman666. P-switch jumping is used to save time, as well as a glitch with Pirhanna Plants where ducking every other frame while big will avoid damage. Double grabbing is also used to save time in a couple places.
Stage 8 Pandemonium (ステージ８ パンデモニウム)
Done by Fethy75. Manipulation of various sprites, glitches, and damage abuse are all used to skip sections of the castle. Bowser is not very exciting, unfortunately, due to the addition of water.
Labyrinth (ラビリンス)
Done by Kaizoman666. Spinjumping through slopes, P-switch jumping, and shell jumping are used to skip most of the level.
Stage 9 Ultra Star Return (ステージ９ ウルトラのほしふたたび)
Done by Kaizoman666 and TheFinalBoss726. Walljumping is abused for both saving time and entertainment. Abuse of items near the end is used to skip a long section of the level.
Final Stage Castlevania (ファイナルステージ あくまじょう)
Done by TheFinalBoss726. Walljumping, screen scrolling, and damage abuse are used to manipulate the falling ceiling and skip sections.
Potential Improvements
- Avoiding the screen scroll for the lightning bolts in Stage 2. It is likely possible to get past these without having to screen scroll, which would save a ton of frames.
- Better manipulation of sprites in Stage 4 and Final Stage. We lost frames in order to manipulate things, but it is likely possible to get a similar result with less frames lost.
- Better manipulation of lag. We were paying attention to it in the first few stages, but gave up focusing on it in later ones.
Nach: Judging.
Toothache: Encode added
Nach: Runs of hacks are under much more scrutiny than runs for normal games. Not only does the run have to be good, so does the game hack, and exceptionally so.
The level design in this game is absolutely horrid. The levels themselves are not cohesive or aesthetically pleasing. Enemies which don't really go with the terrain are present, sometimes with graphical glitches. Enemies are seemingly floating in the air, or packed into absurd places, all in the name of difficulty. Nothing about the game itself bears any kind of professional polish.
Putting visuals and enemy placement absurdities aside, the levels aren't that great either. A bunch of levels are designed with very little room to maneuver. Notwithstanding a few exploitable situations, a run of these levels will generally look the same every time. "Impossible" segments were long, frequent, and placed together, instead of littered throughout a large game. The auto-scrolling segments for the most part were also boring.
In terms of how well this hack shows off the Super Mario World engine, I'd perhaps give it a 15% rating. There's so many great things SMW has, various items, levels with multiple exits, levels with varied routes, and so on. This game barely touches on what can be done. Many SMW hacks are also known for adding cool things onto the SMW engine, instead of using the minimum Nintendo gave us.
There are tons of great SMW hacks out there which have greatly altered the game, giving us a fresh look, and sometimes even varied mechanics from other Mario games, or something completely fresh. They have tons of amazing memorable levels which would make for a great TAS. Some of them have extremely hard segments throughout their levels, but not as concentrated as this one. I'd accept any of those despite being hacks.
The precedence till now for an adequate run with a mediocre hack has been to accept the best of a similar set of hacks. However, we also have a precedence with Super Metroid hacks that even a completely unrelated hack can obsolete one which has little to offer. If I were to accept this, I would have to allow it to be obsoleted by some other SMW hack, which I find rather arbitrary, as there are so many acceptable ones to choose from. The arbitrary nature of such a decision messes up obsoletion chains, which I'd rather not get into. I'd also have to consider if perhaps a publication of this run should be immediately obsoleted by one of our Super Demo World runs.
The run itself while looking amazing in reality doesn't benefit all that much from TASing, compared to how an average run of the game would look. This is because the vast majority of people playing this game will be using save states and other emulator features anyway. As above, too much of the game pigeonholes you into one possible route for a given segment.
So contrasting this run with our existing runs of SDW, or some WIPs I've seen of other SMW hacks, I don't see this run offering anything special that we can't get better elsewhere, or already have to an extent. Coupled with our strict guidelines for what makes a TAS special, and what basic requirements a hack needs to meet in order to be accepted, I have no choice but to reject this.
Nach: Being that the vault does not allow hacked games, I am rejecting this from vault as well.
DrD2k9: Unrejecting and re-evaluating due to rule changes.
DrD2k9: This run looks well played and meets our current guidelines for acceptance. While some optimization techniques for Super Mario World may have been found in the 13 intervening years that could make this run faster, it by no means makes this run look sloppy. Accepting.
SPECIAL NOTE: This acceptance is retroactive based on the emulator used being acceptable at the time the run was originally submitted. While there may be other old rejected runs (which used Snes9x) that may also be re-evaluated to an accepted state, Snes9x is still considered deprecated for new/future submissions.