Since at least 2010 in bobmario511’s ABC speed TAS, Bowser in the Fire Sea has been known to need only one A press. Since then, the stage has remained a continual thorn in the ABC team’s side. A mandatory stage in any completion of the game (regardless of A button or not), it has seemed for a long while that the stage just may not have a solution.
Two notable strategies were found for the stage that ultimately proved fruitless. A tech called remote lava-boosting (RLB) created a method to lava boost from any height as long as a ceiling and floor were close together above lava (and a few other requirements). The stage’s geometry precluded this as a solution by just one unit. In mid-2018, tech for the Wii Virtual Console was discovered that could raise platforms in BitFS. This ultimately wasn’t a great solution, since it does not work on the original N64 due to Wii VC’s imperfect emulation and was not fast enough for a reasonable TAS (it would take about 3 days).
Another strategy was conceived by Tyler Kehne and bad_boot but seemingly died years ago. The idea was based around the idea of bully hyperspeed, a tech that works by putting speed into a bully over time and eventually transferring it back to Mario. This concept was dubbed “Bully Battery” based around the idea of the Bully being a sort of battery to store the speed. The strat had a simple initial issue, though- when the speed is given back to Mario, he is put into a bonk action. This will kill all of the speed given to Mario on the next frame, so canceling the bonk action is a requirement to use the speed.
Two options to cancel the action were found: entering water (which is not viable in BitFS) and a technique called a “squish cancel”. A squish cancel happens when a floor and a ceiling are close together, where one belongs to an object, and one is very steep. It was found that this situation exists in BitFS under the tilting pyramids near the start of the course. This is where the blueprint died- it appeared that in order to squish cancel, Mario would be snapped up to the pyramid’s floor and the speed gained would be killed. Despite much investigation, there was no way to revive this strategy, seemingly the only reasonable one left for BitFS.
Fast-forward a few months, players discovered that if the ceiling did not have a floor above it that the squish cancel could be done at any point in the ceiling (this exposed ceiling is sometimes called an “invisible wall,” which acts similar to a wall since Mario is unable to pass over a ceiling generally). This caused a reinvestigation, which ultimately was also not fruitful due to there not being any exposed ceilings on the pyramid.
However, as of April, the strat is no longer dead. Tyler determined that the disproof had missed the possibility to use an aspect of Mario’s movement called quarter-frames. Every frame breaks Mario’s movement into quarters and then applies that. Tyler’s idea overcomes this dilemma using these quarter frames.
A new issue had arisen, however. TASing this strat would be very precise and tedious to do with our current TASing support. To overcome this issue, a new project for TASing was started by bad_boot. This program, called “sim” (simulator) acts as an effective port of SM64 with more direct input editing. Due to the nature of how this would work, FPS would be incredibly high (in the tens of thousands) and changes in the TAS can be seen immediately down the road. Such a sim as this would allow easier brute-forcing, hexing, routing, and even just generally TASing. Tyler began working on the strat using this new simulator.
Tyler has confirmed portions of the strat, specifically the squish cancel and speed transfer from and to the bully. ds273 has confirmed aspects of the platform displacement, the method by which we turn horizontal speed into vertical distance. The TAS is currently being worked on and in the end will hopefully take less than 5 minutes total. With this progress, a theoretical “No A Button” run is in development, which would collect 70 stars in hopefully less than 2 hours.
In other news
It’s been a while since the last article! We will do our best to highlight some of the most important things that happened during our break.
- In 120 Star, Cheese set a new WR, the first 1:38 ever! Massive congratulations on that run!
- The 120 Star top 5 has also been shaken up generally. Only 2 weeks later, Puncayshun set the second-fastest run ever, what would have been WR if not for Cheese’s new WR. LiamKings bopped Batora for 3rd rank overall, the first time someone new has entered top 3 in the last 2 years (at least) and only the fourth person to ever get a 1:39. Paracusia bopped Simply for fifth place, also putting himself in another elite group. Congrats to the entire group, such progress in such a legendary category is massively impressive.
- Puncayshun set a new 70 Star WR, taking the WR from Taggo. Congrats to Puncayshun again!
- Akki set a new 16 Star WR, beating his former WR. Congrats to Akki as well!
- And lastly, KANNO took both the 0 Star and 1 Star WRs. The 0 Star WR is the fastest human completion of SM64, with a time of 6:36.09 . Congratulations to KANNO for taking such a legendary record.
- IsaacA and Superdavo0001 worked on multiple improvements for the 1 Key TAS, based around the idea of lag improvements using camera manipulation. This lowered the fastest theoretical completion of the game to a time of 4:20.77. Congratulations to both of these TAS authors.
- ds273 and pannenkoek2012 TASed an A Button Challenge improvement, 0.5x Get a Hand re-entry. This TAS required numerous float perfect positions and a ton of brute-forcing.
- sm64expert revealed a cannonless TAS for WMotR. This is the first of the cannonless/coinless/capless challenges ever completed for WMotR, and now allows the entire game to be completed without the use of a cannon. This is also a WR for that star.
- Two people have completed the legendary carpetless strat, something once thought impossible. Congratulations to Xiah and Mikko!
- Arthurtilly released his own take on a SM64 Randomizer, something that would get a fair share of popularity. Development is still on-going.