Submission #6821: Xujhan's SNES Breath of Fire II "glitchless, good ending" in 4:44:37.64

An orphan boy, Ryu, sets out to catch a thief clear the name of his best friend. Along the way he finds new companions, encounters frightening demons, and stumbles across a plot to destroy the world.

Since the original TAS was recorded by Janus in 2010, a number of glitches were found that meaningful reduce the time it takes to complete the game. Consequently BoFII speedrunning has branched into glitchless and any% categories. Since the 2010 TAS predates those glitches, this submission can be considered a spiritual successor to that video. Notable improvements are:

The most significant departures of this TAS from ordinary gameplay are the almost total avoidance of random encounters, and the optimization of boss fights at much lower levels than usually possible. Both of these are achieved through heavy use of RNG manipulation. The RNG function in Breath of Fire II is updated every frame, which allows even very unlikely results to be achieved with only small amounts of waiting. The RNG function is however not affected by very many other things besides time, so we have little ability to improve results beyond simply waiting longer. As a consequence we sometimes take sub-optimal results, when waiting for an optimal result would lose more frames than the optimization would save.

Potential future improvements:

The code for Breath of Fire II is not very well-explored, so all the manipulation in this TAS was produced through trial and error. If we knew how to read particular values from RAM (enemy drops, encounter rate, etc) we would be able to find much more optimal paths through certain sections.

Suggested screenshot: frame 962140 No particular significance, but it's an iconic shot of the hardest boss in the game.

Edit: original file contained a bunch of blank input at the end, oops. The corrected movie is uploaded as a userfile: Userfiles/Info/64953270414370602

FractalFusion: Added encode.

Samsara: Hi! File replaced with a 34155 shorter file that removes the blank input. Also, judging! ...Feels nice to say that again.

Samsara: Setting submission to Delayed, pending a response and potentially an improved file from Xujhan, or a little more feedback from the community on the goal choice.

Samsara: Update - Keeping submission on Delayed, pending an incoming any% improvement, so that the differences in the categories can be more easily compared.

Samsara: Reverting status to judging underway with the submission of #6869: Xujhan's SNES Breath of Fire II "best ending" in 4:34:55.03.

Samsara: The optimization looks solid all throughout, especially given the nature of the game's RNG.

I watched both this run and the any% submission simultaneously, in order to judge whether or not this category is unique enough to warrant publication. The differences between the two categories amount to the glitches, naturally, so I'll go through them one by one:

This is by far the biggest source of time saves over the course of the any% run. The weapons created using this glitch are for specific characters, meaning they aren't actually used as soon as they're made, but once they're in use, some bosses are killed upwards of a full minute and a half faster.

This is used in order to bypass a small diversion in the Circus that takes about 3-4 minutes. It's set up about 45 minutes before it's used, in a case I'm just going to call Chekbook's Gun.

This is used at the exact same time as the payoff (lmao) to the unlimited money glitch, so it's hard to tell exactly how much time it saves in comparison to glitchless. Combined with the money glitch, it's about a 4 minute save.

A moving boulder is walked through, saving 1-2 seconds.

So how does the usage of these glitches in any% compare directly to this run?

The weapon glitch sounds like it should save a much more significant amount of time, given that the early bosses are killed so much faster, but the game itself presents a problem: The dragon form. The dragon form is an AoE attack that seems to do a set 512 damage to every enemy. This is used extensively in both runs as soon as it's acquired, meaning that the later bosses end up not being nearly as big of timesaves in any% as the earlier ones. Any% bosses are still faster, just not on the same level as being a full minute+ faster.

The unlimited money glitch skips a mini dungeon, but it's strictly a mini dungeon and is only about a 3-4 minute diversion as I said earlier.

Skipping over the text glitch, the intangible boulder glitch literally only saves a couple seconds of walking on a single room in a single dungeon.

The text glitch... The text glitch, or rather the explanation behind it, is one of the biggest reasons for my decision. It is apparently extremely powerful, leading to OoB and unrestricted access to the game world... But this isn't achieved in the run, and the reason for that is because the game is incredibly strict on flags. Nothing can be done out of order, and as a result it only makes one part of a run a bit faster. This is... Well, it's a bit of a shame, really. If OoB were viable, that would greatly change things, and it would be far, far more likely that glitchless would be an acceptable alternate category.

There's still one last, major problem with it, though, and that's a complete lack of feedback.

While this run has only received Yes votes, it's only received 5 of them after a month and a half on the workbench, and there's been close to no feedback in the thread, save for a single post (after I explicitly asked) saying that this isn't different enough of a category, and as of right now, I definitely agree with that. I'm hoping that this game and its glitches are studied further. The OoB is the most obvious choice here, even finding a single early trigger could be all that this category really needs to be able to stand on its own. If there's ever another major change to the any% run, I'd be glad to take another look at this submission, and it could be acceptable even with the known improvements (the dragon statue thing stuck out to me, but it's super early on in a 4 and a half hour TAS that's full of RNG manipulation so it's nothing I would reject over on its own).

But, until then, I'm rejecting this category for being far too similar to the any% run. It's definitely a nicely done run, but with such a small time difference relative to the run length, it's a hard sell to have both published.

Judgement on any% will be delayed by a couple weeks to allow more feedback to come in.