Mankind knew that they cannot change society. So instead of reflecting on themselves, they charge cancel infinite'd the beasts.
Game objectives
- Emulator used: BizHawk 2.10
- Aims for fastest time
- Accidentally cuts a full minute and a half off of a 7 minute run while messing around for a different run
- Doesn't use Ky! What kind of Guilty Gear any% is this???
Don't mention Advance!!!
Guilty Gear is a long-running fighting game series created by Arc System Works, and it's a good thing that the Arc System works because this game doesn't. Guilty Gear: The Missing Link is the first game in the series, and true to its name it is in fact missing Link, much unlike Soulcalibur 2 for the GameCube. Other things it's missing include many iconic characters from later in the series, any sense of game balance, and the code that would have allowed Axl to charge moves.
I have a love-hate relationship with Guilty Gear, in that I love pretty much of the games in the series and hate Dust Strikers and Advance. GGML falls toward the loved side for me, it's clearly a rushed product with a lot of jank, but if you know anything about me that isn't "wow this chick loves to write" or "wow this chick isn't funny", it's "wow this chick loves jank".
By the numbers, this is a 5717 frame improvement to the
published run, though the actual improvement is closer to a nice 6900 frames due to the old run being done on PSXjin without the BIOS startup. The main improvement comes from better luck on stage 6, allowing for an easy instant kill, as well as finding a method to guarantee instant kills on the last three stages, skipping several rounds of fighting.
My love of jank is the reason this run exists. I've been interested in a Guilty Gear playaround for a while, now, and this is easily the best currently TASable entry in the series. For proof of this, look at the
Guilty Gear Game Group. It's half-Wonderswan, for heaven or hell's sake! Messing around with this game's jank to see if a playaround would be viable sort of accidentally turned into me trying to see if said jank could improve the published run, and this submission existing is proof that yes it can. I actually finished this run about a month and a half ago at time of submission. I originally didn't plan on submitting it at all, wanting to instead research further improvements and submit an even faster version, but with my spare time decreasing I found myself growing less and less interested in research and development, so I figured it was best to just submit as-is.
I'm not a competitive fighting game player, so pardon me if I misuse any of the lingo inherent to talking about how fighting games work. For those that don't understand what the hell I'm talking about, like me, I've helpfully provided links to the Fighting Game Glossary entries for whatever
I'm talking about.
Controls Glossary
Because it makes later notes slightly easier to understand.
- Square: Punch (P)
- X: Kick (K)
- Triangle: Slash (S)
- Circle: Heavy Slash (H)
- R1: Respect (R)
I will be using
numpad notation throughout these notes, as that's what's used on
Dustloop, which is the reason why there's any in-depth knowledge here whatsoever and not just
me trying desperately to make any sense of what's onscreen.
Testament can't.
Tech*nique* Notes
One of Guilty Gear's most iconic mechanics, IKs are in pretty much every game in the series except the latest one, Strive. In later games, the way they're performed and the effects they have change pretty drastically, but we're not playing later games. This is GGML. In GGML, every character's IK is performed the exact same way. You can either press P+K and land the move, or you can
instant block the opponent. In either case, the screen goes red, during which you
236+any attack to perform the IK. If your opponent doesn't react in time with a 214+any attack, you win the match. Like, entirely. You just win the whole thing right there. Both rounds.
If you've ever wondered why GGML doesn't have a competitive scene, keep what I just said in mind and read on.
Not Ky's wife, she's not in this game. The dizzy mechanic is weirdly fleshed out in GGML for absolutely no reason whatsoever. Every character has a hidden dizzy meter that rises when they're hit, with stronger moves adding more dizziness. When this internal number goes above that character's dizzy threshold (specifically above and not equal), they have a chance to be put into dizzy status every time they're hit. That chance, in theory, isn't 100% due to each character also having an internal "Luck" value that takes away from the chance of being dizzied... In practice, though, the chance is 100%.
Every character has a different dizzy threshold, and it occasionally makes sense! Tiny boy Kliff has the lowest threshold in the game at 16, fast boy Chipp is only marginally more tanky with 18, and rounding out the bottom three is... Sol. What, were you expecting May, a literal child, to be down there? Hell no, bucko! Hers is 38! She has the highest in the game!
While it sounds useless in a run where most fights need less than 60 frames of input, the dizzy mechanic is very important in that it's the only way I've found to be able to pull off IKs on the final three stages. CPUs almost always take the IK when they're in dizzy mode, at most only needing a couple frames of waiting before they won't dodge it.
Charge
Different to standard charge mechanics in other fighting games so no entry link. 12 out of the 13 playable characters in this game can charge certain
specials of theirs to make them stronger. Sorry, Axl. Charging is done by performing a chargeable special as normal but pressing and holding R1 instead of the associated attack button. This might not sound useful for a speedrun, and by itself it isn't, but the way it's implemented allows for a fun mechanic that, among other things, completely destroys this game's competitive meta.
Charge Cancel
If you're familiar with later Guilty Gear games, you've probably heard of the
Roman Cancel, another one of its most iconic mechanics. If you haven't, well, the glossary page is a bit long so here's a quick summary: At the cost of some
meter, you can effectively cancel out of any animation and be able to act immediately. Different situations have different colors of Roman Cancel. The most relevant color here is red, which is the color you get after landing an attack. Roman Cancelling from there mainly means extending combos that would have otherwise ended from long
recoveries. Roman Cancelling is obviously very useful, and it's balanced by the fact that it costs meter. It'd be crazy if it was just completely free. That would just allow you to
infinitely chain
normals!
GGML doesn't have the Roman Cancel, but it does have the charge cancel, which is equivalent to a red Roman Cancel, and is in fact completely free and allows most characters to infinitely chain normals. GGML has around 10-15 frames of
hitstop after normals, with any charge input in that period of time optimally cancelling recovery frames.
Outside of attacking, charge cancelling can be used for something known as Respect Sliding: Combining a charge with a dash input in the right way will cause your character to slide across the ground as if they were dashing, but allowing for any action to be performed while maintaining forward velocity. This can apply to specials, including ones that would otherwise freeze you in place, allowing for you to be at greater distances from your opponent than intended while still allowing for a hit. Because Chipp's dash is so fast to begin with, respect slides aren't used in this run, but they'd be all over a playaround, should I ever have the creativity to make one...
This by itself defines what little GGML meta exists. Characters like Axl, who can't charge, and Potemkin, who can't dash, are bottom tier in the competitive meta since they're unable to keep up with the myriad infinites of pretty much the entire rest of the roster.
If you've ever wondered why GGML doesn't have a competitive scene and you kept that thing about IKs in mind from earlier, congratulations! Now you know.
Why Chipp?
Switching from Ky may seem like an odd decision since he has the fastest IK in almost every Guilty Gear game, and even odder because GGML is one of those games, but what Ky has in his faster IK, Chipp makes up for in a couple other unexpected areas, so both characters end up being a lot more even than expected.
IK Comparisons
As stated incredibly recently, Ky has the fastest IK in the game, optimally coming in at 151 frames timed from the initial P+K input to the moment DESTROYED starts to appear on screen. In comparison, Chipp's is 208 frames long from the same timing. This means that, over 10 IKs played out within the run's input time (since input ends as soon as the 11th is performed), Ky automatically has a 570 frame advantage over Chipp.
Most characters' IKs are around or over 200 frames. The only two other than Ky that are significantly under 200 are May's at 166 frames and Kliff's at 164 frames.
Prefight Animations
Fights don't begin until both characters finish their prefight animations. By my timing, starting from when the screen finishes zooming in right before the animation starts and ending when the UI appears, Chipp's is the second fastest in the game at 89 frames, while Ky's is the sixth slowest, being literally double the length at 178 frames. For further context, I've provided a helpful table below listing times for every prefight animation appearing in the run. This table is in fact very helpful, and in no way completely contradicts the entire preceding paragraph starting from the very first entry.
| Character | Prefight Frames | Against Chipp | Against Ky |
|---|
| Chipp (2) | 62 | lmao | n/a |
| Justice | 79 | see below | -99 |
| Chipp (1) | 89 | mirror | equal |
| Axl | 113 | +24 | -65 |
| Kliff | 119 | +30 | -59 |
| Zato | 122 | +33 | -56 |
| Sol (1) | 165 | +76 | see below |
| Ky (1) | 178 | equal | mirror |
| Potemkin | 189 | +100 | +11 |
| Millia | 206 | +117 | +28 |
| May | 224 | +135 | +46 |
| Baldhead | 233 | +144 | +55 |
| Testament | 235 | see below | +57 |
| Sol/Ky (2) | 245 | +156 | see below |
Just a quick reminder: This table was meant to prove that Chipp has the second fastest prefight animation and Ky has the sixth slowest. What it clearly shows is that Chipp has the fastest and third fastest and that Ky has the slowest and 7th slowest. Hope this helps!
Hello, welcome to the "below" that I told you to see several times earlier. Chipp's animation is a full 27 frames shorter when facing Testament and Justice. I'm not fully sure why, though I believe it's the lack of an attached voice sample. This discrepancy is irrelevant against Testament and their whole "second slowest animation in the run" thing, but against Justice and her "second fastest animation in the run" thing, it means Chipp actually loses no time in prefights to any character in this run.
Ky actually has the opposite: Instead of having a faster animation against specific characters, he has a much slower one when facing Sol. While Chipp gets Sol's normal 165 frame prefight animation, Ky instead gets the 80 frame slower special one, meaning he loses 80 frames as opposed to the mere 13 he would have lost otherwise.
Ky loses a significant amount of time on the 5 fights that have to wait for him: 99 frames on Justice, 65 on Axl, 59 on Kliff, 56 on Zato, and the aforementioned 80 on Sol, a grand total of 359 frames. This cuts into the original 570 frame IK advantage, leaving it at a mere 211.
Load Times
Yes, this matters too. I've provided another table, and thankfully this one doesn't immediately prove any information wrong!
| Versus | Chipp Load | Ky Load | Difference |
|---|
| Axl | 509 | 546 | 37 |
| Baldhead | 493 | 528 | 35 |
| Justice | 540 | 560 | 20 |
| Kliff | 501 | 524 | 23 |
| May | 511 | 517 | 6 |
| Millia | 472 | 497 | 27 |
| Potemkin | 559 | 559 | 0 |
| Sol | 514 | 519 | 5 |
| Testament | 513 | 531 | 18 |
| Zato | 488 | 508 | 20 |
| Chipp/Ky | 493 | 491 | -2 |
Chipp is faster on 9 stages, tied on Potemkin, and only 2 frames slower when facing Ky, presumably due to Ky's stage itself taking a tiny bit longer to load.
In total, Ky loses another 189 frames to longer loads, bringing his advantage down to a mere 22 frames.
Overall/Misc
And that's it in terms of pure animation speed. 22 frames of objective advantage. This does mean that Ky is still faster, of course, but it being less than half a second means it could honestly go either way depending on how actual gameplay goes.
All the Ky data came from me attempting an equivalent run with Ky, and as you can see, Chipp still won out. It really just depends on which character gets the better luck, and I was unable to get consistently good luck with Ky at all over the course of my testing. This seems like a constant, too: The published run goes through four full fights due to being unable to manipulate IKs on them, and Spikestuff's attempt from 2020 had a particularly nasty delay on the May fight that would've rendered the advantage obsolete.
On the note of IKs, the frame values I gave were strictly under optimal conditions, i.e you're as close as possible to the opponent when the actual IK goes off so you spend less time dashing toward or through them. Chipp is much better at closing the distance than Ky, he attacks fast and has less
startup on most of his moves, so in the multiple cases where extra attacks are needed for luck manipulating the IK, Chipp would not only perform those moves faster, but his movement speed would carry him closer to the opponent as a result.
In short, it's difficult to say who has the actual advantage. Purely in terms of animations, Ky wins out by a very small margin, but Chipp's gameplay could easily make up for it in a much harder to define way.
Compared to the mountain of explanations above, there isn't going to be much to these.
Zato
Dash, kick, charge cancel into IK.
Sol
Dash, slash, charge cancel into IK.
Potemkin
Dash into IK.
Kliff
Dash, kick, charge cancel, slight pause, IK.
May
Dash, kick, charge cancel, slash, charge cancel, slight pause, IK.
Millia
Dash, slight pause, IK.
Baldhead
Dash, slight pause, one frame left tap for positioning, double hit into IK.
Axl
Dash, slash, charge cancel into IK.
Axl really loves to start fights with Raiei Sageki, which catapults him up and offscreen for a while. Chipp's slash barely nicks Axl on the way up, keeping him grounded and allowing for the followup IK.
Ky
Pause to manipulate Ky's opening move, slight walk forward, dash into 5H, charge cancel, 5H, charge cancel, repeat 8 more times, IK.
Congratulations, this is the most dynamic fight in the run. Not pausing would cause Ky to charge and instant block every attack, so the quick pause was necessary to be able to land something. Since this is one of the last three fights, the CPU needs to be in dizzy status for the IK to connect. Chipp's fastest method is 5H over and over, with each hit raising the CPU's dizzy meter by 3. Ky has a dizzy threshold of 30, so 10 5Hs are needed to reach it, then the initial IK hit brings him over 30 to put him directly into dizzy status. Between hits, I only dash when I absolutely need to as each dash loses a frame.
This might sound a bit inefficient at first. Since dizzy status is caused when the threshold is *surpassed* and not *reached*, reaching that threshold on one attack then surpassing it on the next makes it seem like I'm wasting a move. In a way, this is correct, but not in that particular way, and not for Chipp at all. The maximum "damage" I can do to an opponent's dizzy meter while maintaining the ability to charge cancel is 3 per hit. The IK starter itself deals 2 damage, which means the opponent can be one below their threshold when hit by it and still get knocked dizzy. In Ky's case, his threshold of 30 should mean I only have to damage him to 29. This would still be 9 5Hs in a row, but the 10th would be replaced with a faster, 2-damage hit. For pretty much any other character, this would save a couple frames of startup on the last pre-IK hit.
From my testing, however, Chipp doesn't even have a 2-damage hit outside of his IK starter. All of his normals, save for two, deal 1 dizzy damage. 5H and 2H both deal 3 damage. 5H actually has a slower startup than 2H, but 2H is a
launcher and IKs don't work on launched opponents even if the starter lands. With Ky's 30 threshold evenly divisible by 3, this means that Chipp is forced to reach that threshold exactly before surpassing it.
Earlier, I mentioned Ky having one last objective advantage left, and this is the fight in which it would take place. Ky's stage 9 is against Sol, who only has a dizzy threshold of 24, meaning Ky doesn't have to infinite as long in order to put Sol into dizzy status. This would save around 50 frames: ~45 coming from two less loops, and another small handful since Ky actually has attacks that deal 2 dizzy damage, meaning he can use a move with a faster startup (2S comes to mind) to get Sol to 23 so the IK starter can push him over.
Secret bonus Chipp tech, though: Since Chipp's IK requires him to dash through the opponent first, he can actually start his IK 4 frames faster than normal if the opponent is against the wall. This only saves time on these last three fights, however, and it would be negated if a way was found to IK them all instantly.
Testament
Dash into 5H, infinite chain 9 more, IK.
Testament also has a threshold of 30, so outside of the more optimal opening, it's the exact same strategy as Ky.
Justice
Dash into 5H, infinite chain 10 more, kick, IK.
Justice has a threshold of 35, which means two more hits are needed in the infinite to put her in dizzy status compared to the previous fights. Not being divisible by 3, however, means I can replace the twelfth and final 5H with a kick. This pushes Justice to 34, then the IK starter pushes to 36, causing dizzy status.
Possible Improvements
- The biggest improvements are all in RNG manipulation.
- Instantly IKing the last three fights would save 10-15 seconds, though throughout all my testing I was unable to get this to happen.
- I feel I got very lucky all throughout the run with the first 8 IKs being quickly and easily manipulable within the game's limits, but if better luck manipulation is possible then all of those fights should be able to be done instantly.
- This would also allow for taking damage on every fight to avoid PERFECTs, as seen in the Baldhead fight.
- It could be possible to throw one or two 2Hs early on into the infinites to save a few startup frames, as long as the opponent can land by the end.
Incredibly Cursed Possible Improvement
Going off of the main improvement point of RNG manipulation, there is actually a definitive fastest character that isn't Chipp or Ky. I made a point of comparing IK and prefight animations when comparing Ky and Chipp, as they have the fastest of their respective categories, and one character just so happens to have both a very fast IK and a very fast prefight animation.
The problem is, that character happens to be Kliff.
Kliff has a 164 frame IK, the second fastest in the game, only being 13 frames slower than Ky's and a solid 44 frames faster than Chipp's. Kliff's prefight animation is 119 frames, 30 frames slower than Chipp's but still 59 frames faster than Ky's. If my math is right, those two factors come out to be a 364 frame objective advantage over Chipp and a 153 frame advantage on Ky (since Ky isn't losing time on opponents with slower prefights). I haven't checked Kliff's loading times for reasons that will become apparent in the next paragraph, but I seriously doubt they'd overtake either one of those advantages.
Clearly, there has to be a downside to Kliff given that I've dropped this reveal and haven't just submitted a faster Kliff% run. There is, and it is a downside that completely kills the possibility of an improvement with current knowledge: He can't dash. Dashing is incredibly vital to this run both for closing distance and maintaining the infinites used in the final three stages, so it's guaranteed that Kliff would lose his entire advantage as long as those fights remain how they are in this run. The gameplay difference between Chipp and Ky is nebulous and hard to really quantify either way, but it's very clear with Kliff. The lack of dash just kills him.
Were a proper method of midfight luck manipulation to be found, along with a way of guaranteeing instantaneous IKs on the final three fights, then Kliff is the fastest character on paper.
Suggested Screenshots
Frame 9735:
Frame 14466:
Frame 18594:
Darkman425: Nice analysis on the pros and cons between Chipp and Ky for their advantages. Seems like getting the opponents to behave right still remains a challenge but getting an IK in all fights is a pretty solid improvement.
Nice work!
