SeedBorn | Majora's Mask — 100% Speedrun with Arbitrary Code Execution @SeedBorn | Uploaded 3 years ago | Updated 3 hours ago
Stream premiere version with commentary and chat: youtube.com/watch?v=o7wGpmP6YH8
This is a tool-assisted demonstration of a Majora's Mask 100% speedrun performed with Arbitrary Code Execution (ACE). Briefly, ACE allows players to modify the game's code while playing it, which allows for an unprecedented level of control over the speedrun.
The run begins, as all Majora's Mask runs do, with first cycle. For those interested in skipping past the cutscene-heavy introduction, here are some timestamps:
22:42 Start of Meaningful Gameplay
24:05 Setup for SRM/ACE
24:49 All Hell Breaks Loose
At 37 minutes in there are some rapidly flashing colors for ~10 seconds. Viewers who are photosensitive may wish to skip this part.
Additionally, I've prepared a Q&A to cover the questions I anticipate will be most common:
Q: Is this a modified version of the game?
A: No! Everything that's happening here is occurring on an unmodified copy of Majora's Mask, and the video is of a continuous playthrough. It was recorded on an emulator (Bizhawk) which allows me to select the inputs I require each frame. In principle the entire run could be recreated on a real N64 console (though the inputs would be unrealistic for a human).
Q: What are you doing to this poor game?
A: During the ~12 seconds that Link plays the ocarina in the graveyard, controllers 2, 3 and 4 write a chunk of assembly code into RAM that essentially serves as the 'cheat codes' for the rest of the run. Those codes are activated by controllers 2-4 during the course of the run, and basically control everything that happens.
If you're interested in an in-depth technical explanation of how this is possible, see youtube.com/watch?v=RoEmGCNsbno.
Also, youtube.com/watch?v=AUVfgZnQLRk provides an introductory explanation of total control ACE in MM.
Q: Is this a real speedrun category?
A: Sort of. The ruleset I'm using here is that of the 100% speedrun (the full rules for which are available at speedrun.com/mm#100). That speedrun bans ACE, however (this video should be enough of a justification as to why).
Q: How does source requirement interact with ACE
A: ¯\_(ツ)_/¯
...but seriously. The 100% speedrun has a stipulation that requires items be obtained from their original source in the game (to ensure that some amount of gameplay is preserved in the presence of glitches that can place items directly into one's inventory). What exactly satisfies this requirement when we have full control over the game is up for interpretation and (since this glitch isn't legal for leaderboard submissions) has no conclusive answer. I've done my best to respect this requirement while creating this run, erring on the conservative side.
Q: Is this run as fast as possible?
A: Not even close! This relates a bit to the previous question, as a notion of 'fastest' or 'optimal' only exists when there is a rigorous set of rules governing things you can and can't do. However, even with the way I decided to interpret the rules for this run, the goal was never to create something that's optimal. While any speedrun that makes full use of ACE is naturally going to outpace non-ACE runs, in creating this video I opted for viewability over speed. At the end of the day this is meant to be an entertaining 'what if' sort of video. If I were cutthroat about optimizing it, I could easily see the post-ACE gameplay being 90+% load screens, with only a few frames spent in each scene. Not only would that have been unfun to watch, I wouldn't have enjoyed making it.
Q: Why make this video?
A: A few reasons. First, ACE is an incredibly cool glitch, but it's a little too broken for RTA runs. It also is infeasible for human runners to showcase its full potential due to the incredibly precise nature of the inputs required; it's much more at home in a tool-assisted environment. Even though ACE has existed for the better part of a year in the Zelda64 games, no one had used it for any sort of large-scale project. The idea of doing a longer speedrun with it -- something that wouldn't end immediately after performing ACE -- was lucrative.
Additionally, the existence of the source requirement rule is what made this project so interesting. That rule (at least in my interpretation of it) prevents using ACE to completely break the game apart, and relegates it to an insane speedup tool. It was a fun challenge to essentially have access to a RAM editor and have to come up with quick ways to obtain everything from its intended source.
Stream premiere version with commentary and chat: youtube.com/watch?v=o7wGpmP6YH8
This is a tool-assisted demonstration of a Majora's Mask 100% speedrun performed with Arbitrary Code Execution (ACE). Briefly, ACE allows players to modify the game's code while playing it, which allows for an unprecedented level of control over the speedrun.
The run begins, as all Majora's Mask runs do, with first cycle. For those interested in skipping past the cutscene-heavy introduction, here are some timestamps:
22:42 Start of Meaningful Gameplay
24:05 Setup for SRM/ACE
24:49 All Hell Breaks Loose
At 37 minutes in there are some rapidly flashing colors for ~10 seconds. Viewers who are photosensitive may wish to skip this part.
Additionally, I've prepared a Q&A to cover the questions I anticipate will be most common:
Q: Is this a modified version of the game?
A: No! Everything that's happening here is occurring on an unmodified copy of Majora's Mask, and the video is of a continuous playthrough. It was recorded on an emulator (Bizhawk) which allows me to select the inputs I require each frame. In principle the entire run could be recreated on a real N64 console (though the inputs would be unrealistic for a human).
Q: What are you doing to this poor game?
A: During the ~12 seconds that Link plays the ocarina in the graveyard, controllers 2, 3 and 4 write a chunk of assembly code into RAM that essentially serves as the 'cheat codes' for the rest of the run. Those codes are activated by controllers 2-4 during the course of the run, and basically control everything that happens.
If you're interested in an in-depth technical explanation of how this is possible, see youtube.com/watch?v=RoEmGCNsbno.
Also, youtube.com/watch?v=AUVfgZnQLRk provides an introductory explanation of total control ACE in MM.
Q: Is this a real speedrun category?
A: Sort of. The ruleset I'm using here is that of the 100% speedrun (the full rules for which are available at speedrun.com/mm#100). That speedrun bans ACE, however (this video should be enough of a justification as to why).
Q: How does source requirement interact with ACE
A: ¯\_(ツ)_/¯
...but seriously. The 100% speedrun has a stipulation that requires items be obtained from their original source in the game (to ensure that some amount of gameplay is preserved in the presence of glitches that can place items directly into one's inventory). What exactly satisfies this requirement when we have full control over the game is up for interpretation and (since this glitch isn't legal for leaderboard submissions) has no conclusive answer. I've done my best to respect this requirement while creating this run, erring on the conservative side.
Q: Is this run as fast as possible?
A: Not even close! This relates a bit to the previous question, as a notion of 'fastest' or 'optimal' only exists when there is a rigorous set of rules governing things you can and can't do. However, even with the way I decided to interpret the rules for this run, the goal was never to create something that's optimal. While any speedrun that makes full use of ACE is naturally going to outpace non-ACE runs, in creating this video I opted for viewability over speed. At the end of the day this is meant to be an entertaining 'what if' sort of video. If I were cutthroat about optimizing it, I could easily see the post-ACE gameplay being 90+% load screens, with only a few frames spent in each scene. Not only would that have been unfun to watch, I wouldn't have enjoyed making it.
Q: Why make this video?
A: A few reasons. First, ACE is an incredibly cool glitch, but it's a little too broken for RTA runs. It also is infeasible for human runners to showcase its full potential due to the incredibly precise nature of the inputs required; it's much more at home in a tool-assisted environment. Even though ACE has existed for the better part of a year in the Zelda64 games, no one had used it for any sort of large-scale project. The idea of doing a longer speedrun with it -- something that wouldn't end immediately after performing ACE -- was lucrative.
Additionally, the existence of the source requirement rule is what made this project so interesting. That rule (at least in my interpretation of it) prevents using ACE to completely break the game apart, and relegates it to an insane speedup tool. It was a fun challenge to essentially have access to a RAM editor and have to come up with quick ways to obtain everything from its intended source.
![[JP] Bottle Content Dupe](https://i.ytimg.com/vi/h_Vcxl_QAxg/hqdefault.jpg "[JP] Bottle Content Dupe
In this video I demonstrate how its possible to dupe bottle contents in the spirit of Kaztaleks video (https://www.youtube.com/watch?v=fkf8za5uQLA), except on the Japanese version. Much like that video, I obtain NWBN while only ever catching one egg.
The principle is the same between Japanese and English: use temp B/stored B to copy an egg caught on Zoras B button multiple times. This is fairly straightforward on the English version, but requires a bit more setup on JP (which is why the video is so long). One large sticking point is that we do not have a quick, repeatable way of getting Zoras B button value written to stored B. The workaround for this is to first copy the egg bottle to Human B, and then make use of H&D3 to repeatedly get temp B on Zora and Goron (https://www.youtube.com/watch?v=GLevXEQAMPY). The first two cycles in this video are devoted to setting this trick up.
By using stored SoT to save while riding Epona, I save Heros Bow (temp B) on Human Link. This means that upon either game load or moon crash, the Human B button will be reverted to the stored B value. Using Glitchesandstuffs observation that the stored B value is preserved through moon crash, we can take Cremia ride as Zora to get the value for zora egg onto stored B. A quick moon crash via day skip then lets us begin a new cycle with a zora egg on Human B.
If you want to skip all of this setup, the 3rd and final cycle in which I actually dupe the eggs begins about 23 minutes in.")
