uu8K8m8s

1
2 /$$$$$$$ /$$$$$$ /$$ /$$ /$$ /$$
3| $$__ $$ /$$$_ $$| $$ / $$| $$ / $$
4| $$ \ $$| $$$$\ $$| $$/ $$/| $$/ $$/
5| $$$$$$$ | $$ $$ $$ \ $$$$/ \ $$$$/
6| $$__ $$| $$\ $$$$ >$$ $$ >$$ $$
7| $$ \ $$| $$ \ $$$ /$$/\ $$ /$$/\ $$
8| $$$$$$$/| $$$$$$/| $$ \ $$| $$ \ $$
9|_______/ \______/ |__/ |__/|__/ |__/
10Designed by Aziz “Hax$” Al-Yami
11Produced & Engineered by Stephen “Streamlord” Kasmir
12Programmed & Engineered by Kyle “Simple Controllers” McDowell
13
14
15Written by Aziz “Hax$” Al-Yami
16 Last Updated: 04/16/2018 
17*****************************************************************
18Table of Contents
19(use CTRL + F)
20*****************************************************************
21[1] Introduction 1
22[2] Universal Rules 4
23 [2.1] Hardware 4
24 [2.1.1] Quantity of Buttons 4
25 [2.1.2] Button Locations 5
26 [2.2] Software 7
27 [2.2.1] Override 7
28 [2.2.2] Macros & Button Binds 7
29 [2.2.3] Order-Dependency 8
30[3] Gamecube Controller Overview 9
31 [3.1] Basics 9
32 [3.2] Zones 11
33 [3.2.1] X-Tilt/X-Smash & Y-Tilt/Y-Smash 11
34 [3.2.2] 50° Line 12
35 [3.2.3] Special Moves 13
36 [3.2.4] Roll & Spotdodge 15
37[4] Digital Controller Overview 16
38 [4.1] Modifiers 16
39 [4.2] Restrictions 17
40[5] Nerfs 18
41 [5.1] Travel Time 18
42 [5.1.1] Smash DI 18
43 [5.1.2] Pivot Tilts 21
44 [5.1.3] Notch Integrity 24
45 [5.1.4] Dash Back Out of Crouch 27
46 [5.2] Precision 32
47 [5.2.1] Y-Tilt + > 50° 32
48 [5.2.2] Neutral-B Integrity 33
49 [5.2.3] Shield Drop Down 35
50 [5.2.4] Ambiguous DI 37
51 [5.2.5] Ice Climbers Desyncs 38
52 [5.2.6] Airdodge 39
53 [5.2.7] Other 42
54 [5.3] Summary 44
55[6] Wavedash Mechanics 45
56 [6.1] The Ledge 45
57 [6.1.1] Intangibility Thresholds 45
58 [6.1.2] Ledge Elevation 47
59 [6.1.3] Airdodge Angle 48
60 [6.1.4] Jump Trajectory 49
61 [6.2] Skill System 52
62 [6.2.1] Difficulty 52
63 [6.2.2] Distance 55
64 [6.3] Traction Anomaly 57
65[7] Modifier Buttons 59
66 [7.1] Modifier 1 60
67 [7.2] Modifier 2 63
68 [7.3] Horizontal Modification Conditionals 65
69[8] Non-Dedicated Modifiers 67
70 [8.1] Restrictions 68
71 [8.1.1] Tilt/Smash Integrity 69
72 [8.1.2] 50° Line Integrity 71
73 [8.1.3] Down-B/Side-B Integrity 74
74 [8.2] Definitely Not Action-Direction Button Binds 76
75 [8.2.1] Firefox 76
76 [8.2.2] Slight DI 78
77 [8.3] Sort of Action-Direction Button Binds 79
78 [8.3.1] Shield Tilt (Automatic) 80
79 [8.3.2] Shield Tilt (Manual) 84
80 [8.3.3] Wavedash 88
81 [8.3.4] Home Row Upwards Airdodge 92
82[9] Other Interactions 97
83 [9.1] UF/DF-Smash 97
84 [9.2] D-Pad 98
85[10] B0XX Advantages 99
86 [10.1] Travel Time 99
87 [10.1.1] Quarter-Circle Smash DI 100
88 [10.1.2] Fastfall -> Side-B 101
89 [10.1.3] Run -> Crouch -> U/UF-Tilt 101
90 [10.1.4] Moonwalk 101
91 [10.1.5] Dash Back -> Dash Back 102
92 [10.1.6] Dash -> Jump With Backwards Trajectory 102
93 [10.1.7] Aerial Drift 103
94 [10.2] Precision 105
95 [10.2.1] No-Fastfall from Ledge 105
96[11] Gamecube Controller Advantages 108
97 [11.1] Hardware 108
98 [11.1.1] Most Intangible Ledgedash 108
99 [11.1.2] Actuation Time 112
100 [11.1.3] Wank Smash DI 116
101 [11.2] Analog 117
102 [11.2.1] Lightshield 117
103 [11.2.2] Shield Tilt 119
104 [11.2.3] Trajectory DI 119
105 [11.2.4] Automatic Smash DI 120
106 [11.2.5] Walk/Run 122
107 [11.2.6] Firefox 122
108 [11.2.7] Airdodge 123
109[12] 1.0 Cardinal 124
110 [12.1] Overview 124
111 [12.2] Redesign 128
112 [12.3] Plan B 133
113[13] Conclusion 136
114[14] F.A.Q. 137
115[15] Patch Notes 139
116*****************************************************************
117[1] Introduction
118*****************************************************************
119Within the competitive Super Smash Bros. Melee scene, the
120ergonomic shortcomings of the Nintendo Gamecube controller have
121taken their toll on many players. Four years ago, I infamously
122joined this club. On May 4th, 2014 I incurred an injury while
123playing Melee that led to a series of surgeries on my left wrist.
124While this was ultimately resolved, it shed light on a condition
125that had even greater implications. My first MRI results revealed
126that my left thumb was showing signs of arthritis at the age of
12719. This inevitably led me to learn that the Gamecube controller
128was no longer an option.
129In December 2016, an opportunity arose. Hit Box, a company known
130for their moderately popular stickless controllers, had been
131looking to dip their toes in the world of Melee with their latest
132product: the Smash Box. Advertised as precise and ergonomic, the
133Smash Box was widely regarded as revolutionary. Confident that I
134could provide valuable insight, I decided to reach out.
135Fatefully, Hit Box shipped me a prototype Smash Box for
136playtesting purposes. The moment it arrived in the mail, I
137started evaluating its performance. But just as quickly as I
138began, I realized that finding flaws in the current model was the
139least of my concerns. The real question was how one would even go
140about rationalizing the Smash Box’s design. While it wasn't hard
141for me to point out improvements that could be made to the Smash
142Box’s button layout, nearly everything software-related was
143beyond my comprehension at the time. Like most Melee players, I
144had been competing for over 10 years without ever giving the game
145engine much thought. Suddenly, I was staring at a bunch of
146coordinates I couldn't recognize, and an instruction manual I
147could barely understand. Hit Box had opened the gateway to the
148future, and there was no going back.
149As I slowly familiarized myself with the Smash Box, it became
150clear just what I had gotten myself into. Hit Box's invention had
151seemingly endless potential, and while I wasn't sure where to
152start, I knew that I was determined to perfect it. As the weeks
1531 
154went by, these aspirations of mine struggled to align with a
155company that had deadlines to meet. In order to rebuild the Smash
156Box from the ground up, I needed complete creative control, which
157meant cutting ties with Hit Box on January 6th, 2017. While this
158decision wasn’t easy, I knew that it would be justified by the
159contribution I’d eventually make to the game. In the most
160interesting journey back to Melee I could've asked for, I sought
161out to give Hit Box's concept the iteration it deserved. I called
162this the B0XX.
163The B0XX would end up taking me over a year to complete. As I
164continued to make breakthroughs in my research and gain skill
165with the controller, my approach constantly shifted along the
166way. Eventually, I settled on the goal of solving the hand health
167epidemic without isolating the Gamecube controller playerbase.
168This seems to resonate with most people in need of the B0XX.
169Among us, there is little-to-no interest in competing on an
170unfair playing field; thus, the model I'll be presenting is
171intended to mimic the performance of a Gamecube controller as
172closely as possible. And while a perfect replica may not exist, I
173believe I’ve achieved something evenly-matched.
174For those who are curious, the advantages of a digital controller
175generally stem from two things. The first is travel time (not the
176same as actuation time, which is discussed in Section 11.1.2).
177Since physical recoil doesn’t exist with directional keys,
178certain motions (i.e. pivot U-tilt) are made reasonable by these
179controllers. The second is precision. Since any coordinate can be
180pinpointed with certainty, there is effectively no difference
181between performing an Ice Climbers desync (a single X or Y-value
182on the entire grid) and a tap jump.
183Wherever possible, I targeted and removed what stood out as
184unrealistic to reliably perform on the Gamecube controller.
185Despite these efforts, it must be understood that the B0XX is not
186obsolete to the Gamecube controller, nor should that be what is
187expected of it. There are some inherent advantages that cannot be
188taken away. In Chapter 10, I will make it clear what these are.
189Similarly, the B0XX has several inherent disadvantages. These
190also tend to fall under two categories. The first is a lack of
1912 
192intuition. Several techniques, such as angling Firefox,
193undoubtedly have a steeper learning curve with this controller.
194The most meaningful disadvantages, though, have to do with the
195B0XX being objectively incapable of performing certain
196techniques. Since the majority of the coordinates on the plane
197aren’t present on this controller (in addition to the rules I’ve
198enforced further reducing what it’s capable of), its potential is
199capped in several areas.
200That being said, it would be foolish to think that taking the
201time to learn a more complex controller with limited options
202ought to be rewarded with the advantages the B0XX grants if it
203isn’t restricted. Perhaps the most obvious example that
204regulation is needed in some capacity is the ability to pinpoint
205the 16.8° (shallowest) wavedash, a feature I consider gamebreaking.
206Some of these advantages aren’t so crystal clear until
207someone has a significant amount of mastery under their belt,
208which is why it took me so long to finalize my rationale.
209Throughout my decisions, I attempted to design the B0XX as
210objectively as possible. Not a single coordinate on the
211controller is arbitrary, though that isn’t to say that the B0XX
212is perfect. The potential to revise this controller, whether for
213the sake of fluidity or game balance, is absolutely still there.
214Nonetheless, I am confident I’ll be providing a starting point
215that is more than sufficient.
216Lastly, we must be aware of the social good that comes alongside
217the solution to our biggest controller-related problem remaining.
218In 2017, we were able to find ways to fix inconsistent
219performance among Gamecube controllers (which, ironically, likely
220stemmed from the digital controller revolution), but health
221concerns still remain. Third-party controllers may be a foreign
222concept to the Melee community, but they are something we have
223overlooked for too long. In most games and sports, several
224different models of equipment are permitted, and understandably
225so; personalization is critical to the enjoyment of any activity.
226The Melee community, however, is faced with something much more
227than just that. For the longevity of our players, it is important
228that we work together to legalize some iteration of the B0XX.
2293 
230*****************************************************************
231[2] Universal Rules
232*****************************************************************
233Parameters that any tournament legal controller should abide by.
234[2.1] Hardware
235[2.1.1] Quantity of Buttons
236All controllers should consist of a 1:1 remapping* of the
237Gamecube controller. For clarification, the actuators I am
238referring to are the A, B, digital L, digital R, X, Y and Z
239buttons, the analog stick/C-stick, and the D-pad. A controller
240with a digital analog stick and/or C-stick should contain one of
241each cardinal direction per stick. The C-stick may also transform
242into the D-pad through the use of a toggle (so long as this
243cannot conflict with the actuation of C-stick inputs).
244A controller may also contain buttons that guarantee certain
245analog L/R-values. This is a logistical requirement for
246controllers that do not have springs, as they would otherwise be
247unable to generate these. While analog L/R buttons inherently
248come with a precision advantage, they are inferior to the full
249analog range of a spring.
250For logistical reasons, the Gamecube controller must resort to a
251spring for analog L/R-values. It should also be noted that the
252Gamecube controller can guarantee certain analog L/R-values
253through the use of either a stopper or calibration exploits (see
254Section 11.2.1).
255*Modifier buttons are only permitted on controllers with at least
256one digital stick, and are only allowed to influence digital
257directional inputs. Modifier buttons may also influence L/R’s
258analog value. There is no harm in having an unlimited amount of
259modifier buttons (so long as they abide by the rules listed in
260Section 4.2), although it is best to keep this number to a
261minimum (the B0XX has two).
2624 
263[2.1.2] Button Locations
264Barring a few exceptions (see Section 2.2.2), controller
265manufacturers should be allowed to arrange their actuators
266however they feel is logical. In general, it would be backwards
267for us to demand anything short of a well-designed controller.
268For the most part, the Gamecube controller’s layout is fine for
269performing in-game techniques frame perfectly. The main exception
270is the C-stick, which is in a much less ideal location. As a
271result, the best way to grip the Gamecube controller is rather
272unorthodox.
273“Javi claw,” a style of grip that has flown under the radar.
274Employing some variation of right hand claw is the only way to
275access the Gamecube controller’s C-stick at all times. While this
276grip isn’t without its downsides, there are ways to mitigate
277nearly all of them. Despite this, right hand claw remains
278unpopular, likely due to its awkward nature. That is to say that
279most players opt not to wield the Gamecube controller optimally
280due to the discomfort that comes with doing so.
2815 
282Unlike the Gamecube controller, any modern controller will have
283its C-stick divided into four easily accessible buttons.
284Sometimes, upholding game purity means compromising quality of
285life to a degree that isn’t worth it. In the same vein that the
286Melee community has come to accept that certain aspects of the
287game (dash back/shield drop) unquestionably warrant redesign, the
288C-stick is an artifact that is unfavorable for competitive play.
289I believe that we should not be opposed to, but supportive of a
290controller that rectifies this.
291As to offer Gamecube controller users the same luxury, the B0XX
292is compatible with the Wii Nunchuk (once plugged in, it replaces
293the analog stick and L button). This will give everyone the
294option to play with the B0XX’s superior right hand layout. The
295Nunchuk utilizes the same stickbox as certain runs of the
296Gamecube controller, making for an easy transition.
297As far as the rest of the B0XX’s layout goes, the elephant in the
298room is surely the Up button’s location. This was chosen by
299process of elimination. It is a must that the left pinky is
300assigned shield (L) and the left thumb operates the modifier
301buttons. If the controller is then given a WASD arrangement, Up
302forces the user to shift their left wrist just to access it. This
303makes the right pinky the most efficient location for Up.
3046 
305Finally, I should clarify L/R/Z’s recommended roles. As I
306mentioned, L is intended to be used for shield, though it should
307also be used to tech. L-cancels should be performed with Z, and
308downwards airdodges (wavedashes) should be performed with R
309(you’ll have to shift your right wrist for these, since R and Y
310aren’t on the home row). Upwards airdodges are a special case
311that will be covered in Section 8.3.4.
312[2.2] Software
313[2.2.1] Override
314Within the context of Melee, opposite cardinal directions must be
315allowed to override each other (Left -> Right = Right. Pressing
316Left + Right on the same frame generates a neutral input). This
317mimics an analog stick most closely, as it shouldn’t be possible
318to receive an unwanted neutral input by failing to release the
319first cardinal before pressing the second (which would happen if
320opposite cardinals negated each other).
321When a cardinal is overriden by the actuation of the opposite
322cardinal, it should not reactivate once the opposite cardinal is
323released. All actuations must be performed manually.
324[2.2.2] Macros & Button Binds
325Macros (inputs occuring on future frames. Can also pertain to
326different outcomes being generated if certain inputs take place
327within a specific # of frames of each other) are illegal.
328Action-action button binds* (more than 1 of A/B/C/L/R/X/Y/Z bound
329to a single button) are illegal.
330Action-direction button binds* (A/B/C/L/R/X/Y/Z bound to an
331impact on the analog X/Y-coordinates) must be evaluated on a
332case-by-case basis. Traditional action-direction button binds
333(i.e. Up bound to A to produce U-tilt) are illegal; however, the
334B0XX uses some of its action buttons in a unique way to pinpoint
335a few niche coordinates (see Section 4.2 and Chapter 8).
3367 
337*While these rules ban button binds on a software level, it
338remains possible to achieve the same effect through hardware. Up
339and A, for example, could be arranged in a yin-yang formation in
340order to guarantee simultaneous actuation. Even though 1:1
341remapping encompasses this exploit to a degree (since it is
342illogical to devote your only Up button to your only A button,
343etc.), it is still worth stating that constructing buttons in a
344manner that guarantees simultaneous actuation is illegal.
345[2.2.3] Order-Dependency
346The B0XX contains two instances of buttons generating different
347outcomes depending on whether or not another button is already
348being held (see Sections 7.3 and 8.3.4). These are needed to
349distinguish the player's intention in situations where
350compatibility with two or more techniques would otherwise
351conflict (i.e. the L button can be used to shield or airdodge).
352Order-dependency is innocuous so long as:
353-All inputs are in line with what is normally legal for an
354actuator to perform.
355-“After” inputs are inclusive of simultaneous presses. For
356example, if Up and L are pressed on the same frame, the
357controller must read this as “Up after L” and “L after Up.”
358This preserves a manual system of inputs, whereas
359recognizing the inputs as “simultaneous” would be a macro.
3608 
361*****************************************************************
362[3] Gamecube Controller Overview
363*****************************************************************
364A thorough understanding of the analog stick’s coordinate plane
365will play a crucial role throughout this document. Excluding
366Section 3.2.3, all of these concepts apply to the C-stick as
367well.
368[3.1] Basics
3699 
370The above diagram contains every coordinate in the game, and
371clearly outlines the 9 sections of the grid (deadzone/4
372cardinals/4 quadrants). The 4 quadrants begin in northeast, and
373ascend counter-clockwise.
374The X and Y axes operate in increments of .0125. Among these
375coordinates, there are some notable ones:
376-.2875 is the minimum value that activates an axis (i.e.
377X .2875 activates east/X .2875 Y .2875 activates northeast).
378-1.0 is the highest magnitude X/Y-value, and can only be
379paired with 0 on the other axis. Due to its microscopic
380range, X or Y +/-1.0 is notorious for being difficult to
381pinpoint on the Gamecube controller (see Chapter 12).
382-X +/-.7000 Y +/-.7000 (45°) are the intended diagonal
383corners.
384-X +/-.9500 Y +/-.2875 (16.8°) and X +/-.2875 Y +/-.9500
385(73.2°) are the shallowest* and steepest** angles
386respectively.
387*Shallow implies that the vector hugs the X-axis.
388**Steep implies that the vector hugs the Y-axis.
38910 
390[3.2] Zones
391[3.2.1] X-Tilt/X-Smash & Y-Tilt/Y-Smash
392On either axis, there are thresholds that serve similar purposes.
393The X-axis has had these documented extensively due to the
394infamous dash back dilemma, the Y-axis not so much. Along the Xaxis,
395.2875 through .7875 is X-tilt, while .8000 through 1.0 is
396X-smash. Along the Y-axis, .2875 through .6500 is Y-tilt,
397while .6625 through 1.0 is Y-smash. X .8000 and Y .6625 aren’t
39811 
399just the cutoffs for tilts/smashes, but several other things as
400well, such as dash/pivot on the X-axis, and tap jump/shield drop/
401fastfall on the Y-axis.
402[3.2.2] 50° Line
403There is a 50° line that separates several techniques in the
404quadrants. Examples include angled F-tilts/vertical tilts, ledge
405get-up/ledgefall, and horizontal/vertical aerials upon pressing
406the A button.
40712 
408[3.2.3] Special Moves
409Neutral-B is unconditionally |X| <= .5875 with |Y| <= .5375. When
410your character is airborne, vertical-B is |Y| => .5500, and sideB
411is |X| => .6000 with |Y| <= .5375.
41213 
413When your character is grounded, the zones for vertical-B and
414side-B are usually swapped as shown. This is true for most, but
415not all grounded states. Crouch, for example, uses the same zones
416as airborne.
417|X| <= .5875 with Y -.5500 will not produce a B move when these
418zones are used. This was likely a programming oversight.
41914 
420[3.2.4] Roll & Spotdodge
421|X| => .7000 causes roll and Y <= -.7000 causes spotdodge. Y <=
422-.7000 also causes crouch.
42315 
424*****************************************************************
425[4] Digital Controller Overview
426*****************************************************************
427[4.1] Modifiers
428Through the use of modifiers, digital inputs are able to
429adequately mimic an analog stick. Without them, the 4 arrow keys
430can only pinpoint the highest magnitude cardinals (X or Y +/-1.0)
431and 45° angles (X +/-.7000 Y +/-.7000). Modifiers can be thought
43216 
433of as shift keys: when held alongside the arrow keys, they can be
434used to select any coordinate within the designated section of
435the grid.
436The provided diagram shows examples of techniques that require
437the use of modifiers. Slight presses of the stick and angles that
438aren’t 45° can’t be performed with the 4 arrow keys alone.
439[4.2] Restrictions
440All modifiers on the B0XX abide by three overarching rules.
441Modifiers cannot change the section of the grid you’re in
442(deadzone/4 cardinals/4 quadrants). This job is reserved for the
4434 arrow keys.
444Modifiers cannot pinpoint banned coordinates (see Section 5.2).
445These are the areas of the game where digital inputs have too
446much of a precision advantage over an analog stick otherwise.
447Action buttons (A/B/C/L/R/X/Y/Z) can possess modifier properties
448and influence the analog X/Y-coordinates. In doing so, they must
449abide by several rules that ensure they can’t cause disingenuous
450behavior (see Chapter 8). For clarification, this feature is not
451meant to permit action inputs bound to directional inputs in the
452traditional way (i.e. U-tilt button). It is simply meant to
453accommodate niche coordinates that Modifiers 1 and 2 don’t have
454room for. For example, the Firefox angle pictured in Section 4.1
455is comprised of Up + Left + Modifier 2 + C-Up. While C-Up remains
456actuated throughout this interaction, its action input bears no
457significance: C-Up merely serves as the physical button needed to
458identify these coordinates.
459The third rule (also known as non-dedicated modifiers) can be
460considered the hallmark of the B0XX. Without it, the controller’s
461minimalistic button layout would be unattainable.
46217 
463*****************************************************************
464[5] Nerfs
465*****************************************************************
466[5.1] Travel Time
467Banning button sequences that involve physically impossible
468analog stick motions.
469[5.1.1] Smash DI
47018 
471On the Gamecube controller, the go-to method for SDI is quartercircle
472SDI. This is when you start in a cardinal, then go into an
473adjacent diagonal to SDI twice by no later than frame 4.
474Due to the lack of physical recoil on the B0XX, an additional
475quarter-circle SDI motion is feasible. This makes for a total of
4763 SDI inputs, which can also occur as rapidly as a frame apart
477from each other (although not spacing them out increases the
478chance that they overlap). Compared to the Gamecube controller,
479this is excessive.
48019 
481To balance things out, the second quarter-circle SDI motion must
482be removed. Before this can happen, we must establish the hitlag
483windows that are consistent with competitive play. Usually, these
484last for <= 9 frames, since almost all of the viable moves in the
485game hit for <= 20% (18%/19%/20% = 9 frames of hitlag). From
486there, it is relevant that you cannot SDI on the first frame of
487hitlag. This leaves us with SDI windows of <= 8 frames.
488Frame 1: Cannot SDI
489Frame 2: East (SDI)
490Frame 3: Northeast (SDI)
491Frame 4: Southeast (Banned)
492Frame 5: Southeast (Banned)
493Frame 6: Southeast (Banned)
494Frame 7: Southeast (Banned)
495Frame 8: Southeast (Banned)
496Frame 9: Southeast (Banned)
497In accordance with the Gamecube controller, the B0XX has been
498programmed to limit the player to only one quarter-circle SDI
499motion (2 SDI inputs) per 9-frame hitlag window. When a cardinal
500is followed by a diagonal on a later frame, the correct diagonal
501to go to next is banned for 6 frames. This input, if attempted,
502is not “pushed” to frame 10 (as that would be a macro); it is
503killed entirely.
504Following this nerf, the only remaining concern involves a
505technique known as double-tapping. This is when the middle and
506index fingers (in that order) are used to actuate a single button
507twice in succession. In theory, this can be used to bypass the
508current SDI nerf. By double-tapping a cardinal, then going to an
509adjacent diagonal, 3 SDI inputs within 9 frames of hitlag remains
510possible.
511For the time being, I chose not to address double-tapping due to
512my inability to recreate it in practice (double-tapping has a
513steep learning curve). If double-tapping proves to be
514exploitable, I will remove it from the B0XX.
51520 
516[5.1.2] Pivot Tilts
517Due to the physical recoil caused by a pivot motion, it is nearly
518impossible to pinpoint the ideal zones for certain tilts shortly
519afterwards on the Gamecube controller. Determining which of these
520the B0XX can have will require case-by-case evaluation.
521Pivot F-tilt isn’t a culprit. It can reliably be done by flicking
522the stick and pressing A as the stick makes its return across Xtilt.
52321 
524Likewise, pivot UF/DF-tilt can be made very reliable. This is due
525to the fact that their ideal zones come into contact with the
526case of the Gamecube controller, making it possible to notch for
527them. By centering your corners on Y .6125 through .6500, you’ll
528be able to perform these by quarter-circle pivoting.
52922 
530The vertical pivot tilts are where the Gamecube controller runs
531into problems. It is nearly impossible to reach the ideal zones
532for these immediately after the pivot; however, there is still a
533good method for pivot D-tilt. By quarter-circling downwards into
534Y-smash + > 50° territory and waiting 4 frames (this timer begins
535upon entering Y-tilt), you can perform a D-tilt afterwards. Pivot
536U-tilt cannot be performed in this manner due to tap jump.
537On the B0XX, the vertical pivot tilts have been brought in line
538with the Gamecube controller. When any of the coordinates capable
53923 
540of performing a dash (X => .8000) are actuated for specifically 1
541frame (this is how a pivot is performed), the A button won’t work
542within the ideal zones for these tilts for a set period of time.
543Pivot U-tilt: 15 frames (removed)
544Pivot D-tilt: 4 frames
545[5.1.3] Notch Integrity
546In Section 5.1.2, I brought up some rather uncommon notches that
547assist with pivot UF/DF-tilt. While these notches do exist, they
548are ultimately bad to have. This is because it is impossible to
549construct them without losing out on better notches.
550There are several points of interest in proximity of the
551NE/NW/SW/SE grooves.
552In all 4 quadrants, the pivot UF/DF-tilt notch sites happen to be
553near X +/-.7000 Y +/-.7000. These are significant for being the
554best* trajectory DI coordinates in the game against moves that
555cause trajectory 361 knockback (nearly half of the moves in the
556game do this). Therefore, any notch must be weighed against how
55724 
558far it takes you away from X +/-.7000 Y +/-.7000; in other words,
559how much DI it costs you. Pivot UF/DF-tilt notches both take you
560at least 4 degrees away from these coordinates. This alone isn’t
561a worthwhile trade-off.
562*Technically, X +/-.7125 Y +/-.7000 and X +/-.7000 Y +/-.7125 are
563the best trajectory DI coordinates. These aren’t listed because
564they have special traits that make them difficult to pinpoint.
565In quadrants 3 and 4, there are two more techniques that benefit
566from notches. The first is jab cancel, which spans from Y -.7000
567to -.7500. This is when you cancel a jab’s IASA frames with a
568crouch in the opposite direction, then press A to jab again. The
569second is shield drop, which spans from Y -.6625 to -.6875 on
570vanilla and Y -.6625 to -.7875 on Universal Controller Fix (a mod
571that has become the tournament standard). Pivot DF-tilt notches
572(Y -.6125 through -.6500) cannot coincide with the notches for
573either of these techniques (regardless of game version).
574Based on this information, the best coordinates to center your
575corners on (if playing on UCF) are:
576Quadrants 1, 2, 3 and 4: X +/-.7000 Y +/-.7000
577These coordinates give you the best trajectory DI in quadrants 1
578and 2, and the best trajectory DI, jab cancel, and UCF shield
579drop in quadrants 3 and 4.
580Having your corners on these values means not having pivot UF/DFtilt
581notches, which I believe the B0XX should remain true to. At
582this point, both of these pivot tilts have to be re-evaluated
583based on how well a Gamecube controller with X +/-.7000 Y
584+/-.7000 corners can perform them.
58525 
586In the same fashion as pivot U/D-tilt, there is still a good
587method for pivot DF-tilt, but not UF-tilt. Whereas X +/-.7000 Y
588-.7000 is valid for a DF-tilt after 4 frames, X +/-.7000 Y .7000
589will prompt tap jump. Pressing the A button within the ideal
590zones for these tilts (after pivoting) has been restricted
591accordingly.
592Pivot UF-tilt: 15 frames (removed)
593Pivot DF-tilt: 4 frames
59426 
595[5.1.4] Dash Back Out of Crouch
596Disclaimer: UCF does not affect dash back out of crouch (dash
597back and dash back out of crouch are two separate techniques).
598In today’s metagame, an increasingly popular choice when
599techchasing is crouching in front of a knocked down opponent.
600This removes the need to react to their get-up attack, since
601crouch cancel renders get-up attack invalid. From there, dash out
602of crouch can be used to follow your opponent’s wake-up roll.
603While dash out of crouch isn’t without execution requirements of
604its own, many players find the most success techchasing by
605rinsing and repeating this strategy.
606Unfortunately, this strategy has a glaring weakness: it is
607humanly impossible to successfully dash back out of crouch 100%
608of the time. While dashing forwards and backwards out of crouch
609share certain criteria, there is a key difference between them
610that attaches a failure rate to the latter.
61127 
612The execution test incurred by dashing in either direction out of
613crouch is being able to traverse from crouch to dash (X-smash)
614within 2 frames (if this motion takes 3 or more frames to
615complete, your character will walk). Through the use of one
616method or another, this criteria isn’t hard to satisfy. Some
617players swear by rolling the analog stick along the bottom of the
618rim in order to minimize its travel route, while others prefer to
619return the stick to its centerpoint before pressing it
620horizontally. With practice, either of these methods should
621ensure a 100% success rate on dash forward out of crouch.
622Along the way to X-smash, X +/-.7750 Y -.6125 and X +/-.7875 Y
623-.6125 (squatRV) cannot be avoided 100% of the time.
624When it comes to dash back out of crouch specifically, any method
625that does not involve hugging the bottom of the rim is made
626unviable by squatRV. Located at Y -.6125, squatRV will cause your
627character to stand up from crouch. SquatRV can be disregarded
628when dashing forwards out of crouch, as it can be cancelled into
62928 
630dash forward; however, squatRV cannot be cancelled into dash
631back. This wouldn’t have been a problem if not for the fact that
632squatRV cannot be avoided 100% of the time.
633Even if the most direct route from crouch to dash is taken, two
634dreaded coordinates must be traversed in the process: X +/-.7750
635Y -.6125 and X +/-.7875 Y -.6125. If you are polled in these
636coordinates, dash back out of crouch will fail. Since this is
637impossible to account for, there is always an element of luck
638when attempting dash back out of crouch with an analog stick.
639By crouching, then dashing in a quadrant, it is impossible to
640fail dash back out of crouch with digital inputs.
641Because digital inputs don’t have a travel route, they are able
642to skip from crouch to dash without traversing squatRV. While
643removing this advantage may seem hopeless, it is actually just a
64429 
645matter of banning the button sequence shown in the diagram. This
646button sequence in particular is concerning because it consists
647of Down -> Down-Forward, which means the player never has to
648release Down in order to perform dash back out of crouch. For as
649long as this is true, dash back out of crouch cannot fail;
650however, if Down must be released, then digital inputs are forced
651to incur risk. In order to create the need to release Down, it
652must be illegal for the B0XX to travel from crouch territory to
653diagonal dash territory. While the B0XX’s programming doesn’t
654outright ban this button sequence, it is made impossible through
655means that will be revealed in Sections 8.2.1 and 8.3.3. For now,
656operate under the assumption that the B0XX must fall back on a
657secondary dash back out of crouch method.
658Once Down has to be released, the combination of squatRV and the
6592-frame dash window creates an execution test for dash back out
660of crouch.
66130 
662With dashing in the quadrants out of the picture, the B0XX is
663forced to perform dash out of crouch in a cardinal direction.
664There are only two valid button sequences for this:
665Sequence A (Success)
666Frame 0: Crouch
667Frame 1: Dash (X +/-1.0 Y 0)
668Sequence B (Success)
669Frame 0: Crouch
670Frame 1: X +/-.7000 Y -.7000
671Frame 2: Dash (X +/-1.0 Y 0)
672Having to release down also makes for the possibility of a third
673button sequence. This will cause dash back out of crouch to fail:
674Sequence C (Failure)
675Frame 0: Crouch
676Frame 1: SquatRV (X 0 Y 0)
677Sequence A is a one-frame link, while Sequence B is a pseudo-twoframe
678link (Down and Forward can overlap for one frame, but Down
679must be released on the next frame for dash to occur). Meanwhile,
680Sequence C is an accidental series of inputs that can result from
681having to perform a skillful motion. All in all, removing the
682B0XX’s ability to crouch, then dash in a quadrant forces the
683player to think twice about performing dash back out of crouch.
68431 
685[5.2] Precision
686Removing the ability to pinpoint certain coordinates with 100%
687accuracy.
688[5.2.1] Y-Tilt + > 50°
689Within the quadrants, > 50° territory can be used to perform
690turnaround vertical tilts. In doing so, it is relevant whether
691you are pointing in Y-tilt or Y-smash. Whereas the former can be
692used in any situation, the latter can only be used in buffered
693situations (i.e. L-cancel lag). This is because Y-smash is
694associated with techniques that conflict with your ability to
695perform tilts, such as tap jump and smash attacks.
696On the Gamecube controller, the elusive Y-tilt + > 50° is
697extremely difficult to pinpoint. As a result, people usually
698perform non-buffered turnaround vertical tilts by turning around,
699then pointing vertically (in the north or south corridor), which
70032 
701can only be equal to or slower than pointing directly at Y-tilt +
702> 50°. To recreate this inconvenience, the B0XX cannot pinpoint
703Y-tilt + > 50°.
704[5.2.2] Neutral-B Integrity
705One of the Gamecube controller’s intrinsic risks is overshooting
706into side-B territory when attempting to neutral-B in the other
707direction.
70833 
709Digital inputs can circumvent this risk by situating the analog
710stick in X .2875 through .5875 (the zone that ensures a neutral-B
711in the chosen direction).
712Since this would otherwise feel disingenuous, attempting to
713neutral-B with the analog stick in X .2875 through .5875 will
714result in your X-value being pushed into side-B territory.
715X .7375, the coordinate this pushes you to, is a non-arbitrary
716one that the B0XX already contains elsewhere. This means that the
717only way to neutral-B is to completely release the horizontal(s).
71834 
719[5.2.3] Shield Drop Down
720With the Gamecube controller, there are two drastically different
721ways to shield drop.
722The first shield drop method involves shutting off roll, then
723going to the SW/SE corner of the stick (which has been notched).
724This has become well-known for its ease and reliability in recent
725years. Colloquially, it is called the “Axe method.”
72635 
727The second method involves pointing directly down at shield
728drop’s Y-values. This lets you shield drop as early as frame 2
729(if you haven’t shielded yet). In theory, there is no reason not
730to shield drop this way every time. In reality, however, this
731method is far too difficult to perform consistently.
732The key to removing shield drop down from the B0XX lies in the
733order of priority within the game. If chosen on the same frame,
734the game will prioritize:
73536 
736Spotdodge (Y -.7000) > Roll (X +/-.7000) > Shield Drop (Y -.6625)
737Therefore, if the only shield drop Y-values on the B0XX are
738paired with roll X-values, shield dropping without shutting off
739roll is impossible*. The B0XX uses the coordinates X +/-.7250 Y
740-.6875, which abide by this rule.
741*This nerf removes the B0XX’s ability to shield drop down on
742frames 2, 3 and 4. There remains a method that involves tilting
743shield downwards (in Y-tilt) for 4 frames, which then turns the
744entire spotdodge range (Y <= -.7000) into shield drop on
745specifically frames 5 and 6. This does not need to be targeted,
746as it can reliably be performed with the Gamecube controller.
747[5.2.4] Ambiguous DI
748When you are thrown vertically, the best option is often to DI
749ambiguously. This is when you make it difficult for your opponent
750to discern which side of them you’re on. In these situations,
751there are always “perfect coordinates;” ones that make it as hard
752as possible for your opponent to tell.
753In this situation, for example, X .4000/.4125 is the perfect mixup.
754In theory, if the B0XX had access to every X-value, it would
755be able to pinpoint both of these coordinates; therefore, the
756line has to be drawn somewhere.
75737 
758The clear choice is 4 coordinates along the X-axis. Since less
759than 4 (2) is grossly underpowered, while more than 4 (8) digs
760into user-friendliness, this is the correct amount.
761The significance of X +/-.7375, the final coordinate in this
762range, will be explained in Section 7.1.
763[5.2.5] Ice Climbers Desyncs
764The B0XX does not contain any of the coordinates that cause Popo
765and/or Nana to perform isolated actions. These are:
766X +/-.8000 (Popo X-Smash/Nana X-Tilt)
767Y +/-.6625 (Popo Y-Smash/Nana Y-Tilt)
768X +/-.7000 Y Not Along Rim (Popo Roll)*
769X Not Along Rim Y -.7000 (Popo Spotdodge/Nana Shield Drop)*
770X Along Rim Y -.8000 (Popo Spotdodge/Nana Shield Drop) (UCF
771Only)**
772X .6250 (Popo Run/Nana Runbrake)***
773X .7500 (Popo Teeter Break/Nana Teeter)***
774Y .5625 (Popo Jump out of Dash/Run/Runbrake/Turnrun)
775|X| <= .5875 Y -.5500 (Nana Neutral-B) (Grounded Only)
776X Y or C X Y in Proximity of 50° Line (2 Different Aerials)
777C X +/-.8000 (Popo F-Smash)
778C Y +/-.6625 (Popo U/D-Smash)
77938 
780*Based on the logic presented in Section 5.1.3, it is acceptable
781to pinpoint X +/-.7000 and/or Y -.7000 (the sources of these
782desyncs) along the rim.
783**This desync is a byproduct of UCF’s modified shield drop range.
784***These desyncs only work facing east. This is due to the
785Gamecube controller’s X-axis spanning from -128 to 127 before
786conversion to Melee values occurs, resulting in imbalances in
787absolute values.
788[5.2.6] Airdodge
789If unrestricted, the B0XX’s most blatantly overpowered feature is
790its ability to pinpoint the shallowest wavedash angle in the
791game. It doesn’t take much experience with this controller to
792realize that this needs to be nerfed in some capacity.
79339 
794Initially, the B0XX had five airdodge angles. Just like the 16.8°
795airdodge, the 73.2° airdodge was very unrealistic (albeit not
796nearly as competitively advantageous). The lower and upper limits
797clearly needed to be kept healthy distances away from the perfect
798angles.
799Following this realization, there wasn’t a need for more than
800three airdodge angles. Choosing wavedash distances is extremely
801unintuitive past the point of “short”/”medium”/”long,” anyway, so
802this worked out well. These wavedashes were clearly around the
803right lengths, but I eventually wanted to settle on angles that
804weren’t just benchmark numbers. 30.1° and 59.9° had been chosen
805because they were closest to 30° and 60°; the final product
806needed a better explanation than that.
80740 
808This would prove to be the most complex nerf of all, but I was
809eventually able to find non-arbitrary airdodge angles within the
810game. Coincidentally, these were only a coordinate off of the
811ones I started with. They also mirrored each other, making for
812the most elegant outcome I could’ve hoped for. The coordinates
813for these angles are:
814X +/-.8500 Y +/-.5000 (30.5°)
815X +/-.5000 Y +/-.8500 (59.5°)
816To learn about what makes these coordinates significant, refer to
817Chapter 6.
818I would also end up finding a reason to use X +/-.6500 Y +/-.6500
819as the coordinates for the 45° airdodge (instead of X +/-.7000 Y
820+/-.7000). This will be revealed in Section 8.3.2.
82141 
822[5.2.7] Other
823There are a few cases of techniques requiring incredible
824precision that don’t fall under any of the aforementioned
825categories. All of the coordinates listed in this section are
826illegal for the B0XX to have.
827Middle UF/DF-Tilt & Middle UF/DF-Smash
828One of the most heated debates during Melee’s design phase was
829how the deadzone ought to be shaped and sized. There are remnants
830on the game disc suggesting that circular deadzones, as well as
831smaller deadzones were considered at some point. When these
832ultimately didn’t make the cut, a technique that was shafted in
833the process was the ability to F-tilt/F-smash at two more angles.
834These less well-known F-tilt/F-smash angles still made it to the
835final version, seemingly as an Easter egg. In each quadrant, a
836single set of coordinates will let you perform them. These
837coordinates are X +/-.9500 Y +/-.2875* (and C X +/-.9500 Y
838+/-.2875), the ones that hug the X-deadzone corridor hardest.
839*X +/-.9500 Y +/-.2875 (on the analog stick) doesn’t necessarily
840have to be banned; however, it is illegal for the A button to
841actuate alongside these coordinates.
84242 
843Pikachu & Pichu’s Double Upwards Up-B
844The equation for Pikachu and Pichu’s up-B pulses is ridden with
845loopholes. While there’s no need to go into detail, the takeaway
846is that there are exploitative coordinates that allow you to up-B
847directly upwards twice. These are:
848X +/-.5000 Y 0
849X 0 Y -.5000
850X +/-.4000 Y +/-.3000
851X +/-.3000 Y -.4000
85243 
853[5.3] Summary
854Smash DI
855Only 2 smash DI inputs can be performed during a hitlag window.
856Pivot Tilts
857Pivot U-tilt is removed / Pivot D-tilt (fast method) is removed.
858Notch Integrity
859Pivot UF-tilt is removed / Pivot DF-tilt (fast method) is
860removed.
861Dash Back Out of Crouch
862The risk of a failed dash back out of crouch is incurred.
863Y-Tilt + > 50°
864Y-tilt + > 50° cannot be pinpointed.
865Neutral-B Integrity
866The risk of an accidental side-B is incurred.
867Shield Drop Down
868Shield drop down (fast method) is removed.
869Ambiguous DI
870There are no ambiguous DI mixups.
871Ice Climbers Desyncs
872There are no Popo/Nana desync coordinates.
873Airdodge
874The shallowest/steepest airdodge angles are each kept 13.6° from
875the respective perfect angles. 30.5° and 59.5° are non-arbitrary
876airdodge angles that will be discussed in Chapter 6.
877Other
878Various obscure coordinates are illegal.
87944 
880*****************************************************************
881[6] Wavedash Mechanics
882*****************************************************************
883From the beginning, 45° was the only airdodge angle set in stone.
884With the shallowest/steepest airdodges angles in the game (as
885well as anything near them) out of the question, I had two left
886to find. These needed to be in the vicinities of 30° and 60°.
887Given how subjective the act of choosing these airdodge angles
888might seem, yet how directly they correlate with the B0XX’s
889strength, I felt the need to study this area of the game
890thoroughly. I kept my hopes up that the most important
891coordinates on the entire controller would have some sort of
892justification behind them. If not, the B0XX would forever remain
893arbitrarily designed.
894[6.1] The Ledge
895My first instinct was to investigate the high tier characters’
896(Fox/Falco/Sheik/Ice Climbers/Pikachu/Luigi) ledgedashes. This
897isn’t as obvious with an analog stick, but digital inputs make it
898easy to tell that airdodge angles have an impact on ledgedash
899frame data.
900As it turned out, I had a very shallow understanding of the ledge
901all along. It is a mechanic with many subtleties I wasn’t
902formerly aware of.
903[6.1.1] Intangibility Thresholds
904I first learned that, when ledgedashing, steeper airdodges cause
905you to make contact with the floor faster. This means that
906intangibility is inversely correlated with distance. Despite this
907being overlooked by the scene at large, it is often better to
908prioritize the former.
90945 
910The frame Sheik will land on when she falls on frame 1, jumps on
911frame 2, then airdodges at various angles on frame 9.
912Sheik is unique in that her most intangible ledgedash requires an
913extremely steep airdodge: => 59.5°. This allows her to maximize
914her actionable frames by landing on frame 9 (if Sheik airdodges
915at an angle shallower than 59.5°, she cannot land until frame 10
916or later).
917X +/-.5000 Y -.8500 (59.5°) are the last valid coordinates along
918the rim for Sheik’s frame 9 ledgedash. I figured Sheik ought to
919be able to land on frame 9 with the B0XX, so I went with these
920coordinates. Even though this is a bit of a TAS airdodge angle
921for Sheik, distance isn’t important in this particular case. The
922purpose of her frame 9 ledgedash is intangibility, and nothing
923else.
92446 
925Intangibility thresholds exist for all characters who can
926ledgedash, and are worth being aware of. Finding the B0XX’s final
927airdodge angle, however, would require yet another discovery.
928[6.1.2] Ledge Elevation
929A little-known fact about the stages in Melee is that
930ledgedashing is not uniform across them. This is because the
931elevation of the stage relative to the ledge varies. For whatever
932reason, the game developers made this decision. They also went on
933to reuse certain ledge positionings.
934Easier Stages (Ledge is Higher):
935Harder Stages (Ledge is Lower):
936Conveniently, half of the legal stages share an easier difficulty
937(BF, DL64, and FoD have a higher ledge), while the other half
938share a harder difficulty (FD, YS, and PS have a lower ledge). By
939difficulty, I quite literally mean that certain ledgedash motions
940will succeed on the easier stages and SD on the harder ones.
94147 
942[6.1.3] Airdodge Angle
943As far as the criteria that determine whether a ledgedash
944succeeds or not goes, there are two components. The first is the
945airdodge angle, which (usually) must be shallow enough to rise
946above the stage. With this in mind, Fox’s frame data led me to
947the final set of airdodge coordinates.
948Fox’s frame 6 ledgedash consists of falling (without fastfalling)
949for 2 frames, jumping on frame 3, then airdodging on frame 6.
950This ledgedash requires a shallower airdodge angle than usual to
951succeed. On the easier stages, Fox needs a <= 34.7° airdodge to
952make it onstage. On the harder stages, he needs <= 30.5°.
953Due to certain intricacies of the B0XX’s ledgedash methods, it is
954mandatory that Fox is able to ledgedash on frame 6 on the harder
955stages with this controller (it is overly difficult to ledgedash
956with Fox otherwise). These intricacies will be explained in full
957in Section 11.1.1, although you’ll probably figure out what they
958are by the end of this chapter. For the time being, this means
959Fox needs access to X +/-.8500 Y -.5000 (30.5°), the first angle
960along the rim that enables this ledgedash.
96148 
962[6.1.4] Jump Trajectory
963The other component that the success of a ledgedash hinges on is
964the horizontal trajectory of the midair jump. This is calculated
965on the jump frame (as early as frame 2) of a ledgedash sequence.
966To demonstrate this, I will analyze Fox’s frame 5 ledgedash.
967X .7000 aerial drift/jump trajectory, and a 45° airdodge (X
968+/-.7000 Y -.7000) will be used.
969Frame 0:
970Fox hangs from the ledge. Along Battlefield’s X-axis, he is
971located at 70.32.
972Frame 1:
973Fox falls from the ledge. If he falls diagonally, he’ll advance a
974small amount in that direction. In this case, he falls straight
975down, which doesn’t result in any horizontal movement.
97649 
977Frames 2, 4 & 5 (No Jump Trajectory):
978On frame 2, Fox jumps, but he has failed to do so with horizontal
979trajectory. He remains at 70.32.
980On frames 3 and 4, Fox drifts at X -.7000, but this has a
981negligible impact on his positioning. He only makes it to 70.13.
982Then, on frame 5, Fox airdodges at 45°, but this results in a SD.
98350 
984Frames 2, 4 & 5 (Jump Trajectory):
985This time, Fox jumps at X -.7000 trajectory on frame 2. He has
986already made it to 69.71.
987By frame 4, he has made it to 69.31.
988Then, on frame 5, the 45° airdodge makes it onstage.
98951 
990In the first example, Fox sealed his fate on frame 2 when he
991jumped without any trajectory. At that point, the 45° airdodge
992couldn’t have succeeded. In the second example, Fox jumped with
993trajectory. This brought him closer to the stage, granting him
994more leniency on his airdodge angle. All characters’ ledgedashes
995benefit from jump trajectory in this manner.
996[6.2] Skill System
997When designing the B0XX, I had to take all of the subtleties of
998ledgedashing into consideration. This wasn’t just a matter of
999choosing airdodge angles, but preserving the Gamecube
1000controller’s intrinsic challenges. So far, I have yet to show
1001parallels for some of these. For example, the stage a Gamecube
1002controller user is playing on should influence the airdodge angle
1003they attempt, as certain angles won’t work on every stage.
1004Similarly, the all-important role of jump trajectory on the B0XX
1005has neither been demonstrated, nor explored in full. Rest
1006assured, these subtleties are all intact.
1007[6.2.1] Difficulty
1008On the B0XX, the relationship between stage difficulty (ledge
1009elevation) and jump trajectory is preserved across the cast. To
1010demonstrate this, I’ll once again analyze Fox’s frame 5
1011ledgedash.
1012Frames 2 & 5 (Easier Stage / No Jump Trajectory)
1013On the B0XX, it is fine for Fox to jump without any trajectory
1014(on frame 2) on the easier stages.
101552 
1016He’ll need a single frame of horizontal drift (can be done as
1017late as frame 4), but he’ll still be able to make it onstage with
1018the 30.5° airdodge.
1019Frames 2 & 5 (Harder Stage / No Jump Trajectory)
1020On the harder stages, this isn’t the case. If Fox fails to jump
1021with trajectory, no matter how well he drifts from that point on,
1022he will die if he attempts a ledgedash on frame 5. For anyone
1023curious, a <= 28.8° airdodge would have been needed for Fox to
1024survive from this position.
102553 
1026Harder stages increase the risk of an unintentional ledge getup.
1027In summary: a Gamecube controller user can afford to steer clear
1028of ledge get-up (< 50° territory) on the easier stages because
1029jump trajectory isn’t as necessary. On the harder stages,
1030however, they should look to jump with trajectory in order to
1031reduce the shallowness needed on their airdodge. But in doing so,
1032the risk of an unintentional ledge get-up becomes very real.
103354 
1034On the B0XX, this challenge is preserved. The player must avoid
1035pressing Forward on the fall frame of a ledgedash*, otherwise,
1036they too will receive a ledge get-up. To perform your character’s
1037most intangible ledgedash with jump trajectory, you will have to
1038one-frame link the Forward press after falling (either frame 1
1039Down, frame 2 Down-Forward or frame 1 Back, frame 2 Forward).
1040*This one-frame link is only needed for < 50° ledgedashes. When
1041performing > 50° ledgedashes, you can ledgefall diagonally (see
1042Section 8.1.2).
1043[6.2.2] Distance
1044Not all ledgedashes that occur on the same frame are created
1045equal. Regardless of stage, there is always incentive to perform
1046a more skillful motion and jump with trajectory.
1047Fox’s Frame 5 Ledgedash (No Jump Trajectory):
104855 
1049If Fox jumps without trajectory, then airdodges at 30.5°, he only
1050makes it to 44.46 along Battlefield’s X-axis.
1051Fox’s Frame 5 Ledgedash (Jump Trajectory):
1052If Fox jumps with X 1.0 trajectory, he makes it to 43.25. This is
1053the best ledgedash the B0XX is capable of.
1054This extra 1.2 units along the X-axis equates to airdodging at a
1055~2° shallower angle, which, believe it or not, is meaningful. If
1056you take a close look at Fox’s feet, you’ll be able to tell that
1057jumping with trajectory caused him to travel further. This is a
1058prime example of why the B0XX’s wavedash mechanics had to be
1059fine-tuned so carefully. Seemingly small details like these can
1060make a difference.
1061Due to the variable that is jump trajectory, ledgedashing on the
1062B0XX is by no means static. Skill is always rewarded with longer
1063ledgedashes, and in some cases, more intangible ones.
106456 
1065[6.3] Traction Anomaly
1066There is one final aspect of Melee’s physics engine that is worth
1067mentioning.
1068Conventional wisdom has it that the shallower the airdodge, the
1069further the wavedash will travel. This is not universally true.
1070When a character’s velocity exceeds their maximum walk speed (a
1071seemingly unrelated stat), their traction is doubled, reducing
1072their effectiveness at sliding across the floor. For most
1073characters, this creates a breakpoint where certain angles aren't
1074shallow enough to make up for doubling the character’s traction,
1075resulting in a net loss in distance traveled. It is appropriate
1076that I use Fox to demonstrate this anomaly, since the airdodge
1077coordinates I’ve given the B0XX cause him to be impeded by it.
1078Fox’s maximum walk speed is 1.6 units/frame. His base traction is
1079.08. This means that if Fox’s velocity ever exceeds 1.6
1080units/frame, his traction will double to .16.
1081This is how far some of the shallower angles cause Fox to travel:
1082 X +/-.8375 Y -.5375 (32.7°) = 25.6 units
1083 X +/-.8375 Y -.5250 (32.1°) = 26.3 units
1084 X +/-.8500 Y -.5250 (31.7°) = 26.6 units
1085 X +/-.8500 Y -.5125 (31.1°) = 26.9 units
1086X +/-.8500 Y -.5000 (30.5°) = 25.8 units*
1087X +/-.8625 Y -.5000 (30.1°) = 26.0 units*
1088X +/-.8625 Y -.4875 (29.5°) = 26.4 units*
1089X +/-.8625 Y -.4750 (28.8°) = 26.8 units*
1090 X +/-.8750 Y -.4750 (28.5°) = 27.0 units
1091*Despite being shallower, these angles are actually worse than
1092the ones directly before them. This is because they cause Fox’s
1093velocity to exceed 1.6 units/frame, but not by enough to be worth
1094it. Here is a diagram of these gimped angles:
109557 
1096Interestingly, the first airdodge angle along the rim that
1097enables Fox’s frame 6 ledgedash on the harder stages (X +/-.8500
1098Y -.5000) is also the first one that doubles Fox’s traction. This
1099means Fox’s wavedash on the B0XX is a bit weaker than it would
1100seem (since the B0XX uses these coordinates). Whereas the
1101airdodge angle itself is 13.6° shy of perfect, Fox’s wavedash is
1102effectively 15° to 16° shy, since airdodges ~2° steeper than this
1103one cause him to travel further.
110458 
1105*****************************************************************
1106[7] Modifier Buttons
1107*****************************************************************
1108Modifier buttons are the glue that hold the B0XX together. They
1109are what make it viable to play Melee with a completely digital
1110controller. That being said, modifier buttons take up valuable
1111real estate, which is why one of my highest priorities was
1112needing as few of them as possible.
11134 modifier buttons (and a WASD arrangement) is a no-go.
1114The downfall of most digital Melee controllers is the inclusion
1115of Modifiers 3 and 4. Though these might seem necessary at first,
1116the problems they cause are just as readily apparent. Like poison
1117to the controller, Modifiers 3 and 4 push other buttons out of
1118good locations only to station themselves in lackluster ones.
1119Then, they force you to memorize several new button combinations,
1120exacerbating an already uninviting lack of intuition.
1121From early on, I knew that the B0XX had to be kept to two
1122modifier buttons. Fortunately, two is all it needs. The secret is
1123to fill each of them to the brim with functionality.
112459 
1125[7.1] Modifier 1
1126When Modifier 1 is pressed in conjunction with the cardinal
1127directions, you will receive:
1128Horizontals: X +/-.7375*
1129Verticals: Y +/-.6500
1130Quadrants: X +/-.7375 Y +/-.2875
113160 
1132On the B0XX, both modifier buttons enable slight presses of all 4
1133cardinal directions as well as all 4 quadrants. This is their
1134most basic function.
1135Along the X-axis, Modifier 1 will produce X .7375. While X-tilt
1136spans up until X .7875, X .7375 is the upper extremity a digital
1137controller should use. This is because X .7375 is the greatest Xvalue
1138that does not break teeter, the mechanic that protects your
1139character from falling when they walk near the end of the stage.
1140X-values that do not exceed .7375 are therefore most efficient,
1141since they not only let you slight DI when thrown, but also
1142safely walk and perform tilts.
1143It is worth mentioning that X .7375 isn’t ideal for intentionally
1144teetering (to set up a teeter-drop, for example). The best way to
1145do that is to initiate walk (X-tilt), then immediately release
1146the modifier button to continue walking at X 1.0 (maximum) speed.
1147This allows you reach the end of the stage as fast as possible.
1148*Modifier 1 won’t always cause the horizontals to be modified to
1149X .7375 (see Section 7.3).
1150Along the Y-axis, the only worthwhile modification is Y .6500.
1151The lower extremity (Y .2875) is obsolete in comparison, since it
1152causes you to stand up from crouch (Y => -.6125), performs
1153neutral-B (|Y| <= .5375) instead of vertical-B, etc.
1154Within the quadrants, Modifier 1 will produce X .7375 Y .2875. X
1155is kept consistent with the horizontals so that alternating the
1156horizontals and quadrants doesn’t awkwardly change your movement
1157speed, while Y is given the non-arbitrary value of .2875. These
1158coordinates allow you to point in < 50° territory and perform
1159various techniques, such as UF/DF-tilt. In particular, it was
1160imperative for me to assign this side of the 50° line to the
1161modifier button in the best resting position for the user’s left
1162thumb. This is because Modifier 1 enables the 30.5° wavedash,
1163making it the most important modifier button. I will explain how
1164it does this in Section 8.3.3.
116561 
1166Analog L/R 43, the largest lightshield in the game.
1167Lastly, when pressed in conjunction with the L button, Modifier 1
1168will produce analog L 43. In Melee, analog L/R-values span from 0
1169to 140, with 43 being the lightest press that generates a shield.
1170A full analysis of the B0XX and Gamecube controller’s analog L/R
1171capabilities will be conducted in Section 11.2.1 (once more
1172information has been conveyed).
117362 
1174[7.2] Modifier 2
1175When Modifier 2 is pressed in conjunction with the cardinal
1176directions, you will receive:
1177Horizontals: X +/-.2875*
1178Verticals: Y +/-.7375
1179Quadrants: X +/-.2875 Y +/-.7375
118063 
1181Originally intended to mirror Modifier 1, Modifier 2 is a watered
1182down version of what it could have been. Within the quadrants,
1183Modifier 2 is banned from pinpointing Y-tilt + > 50°, and must
1184settle for pinpointing Y-smash + > 50° instead. Still, this is
1185useful for performing turnaround U-tilt in buffered situations (Y
1186<= -.7000 will cause crouch and prevent turnaround D-tilt). In
1187the cardinals, X +/-.2875 usually serves as another slight DI
1188option, while Y +/-.7375 isn’t chosen for any particular reason
1189other than symmetry with Modifier 1.
1190*Modifier 2 won’t always cause the horizontals to be modified to
1191X .2875 (see Section 7.3).
1192Analog L/R 140. Visibly identical to, and incurs the same shield
1193stun as a digital shield, but cannot powershield.
1194When pressed in conjunction with the L button, Modifier 2 will
1195produce analog L 140. Very few people are aware of the
1196significance of this technique. On the Gamecube controller, it is
1197done by situating the L/R slider atop its digital press (with
1198proper calibration). In doing so, the intention is to not
1199powershield.
1200Contrary to popular belief, powershielding physical attacks can
1201actually be disadvantageous. Because a successful powershield
1202pushes your character back, it can push them out of range for
120364 
1204certain out of shield options. Analog L/R 140 can be useful if
1205you’re able to recognize these situations, since it incurs the
1206same amount of shield stun as a digital shield but lacks the
1207ability to powershield.
1208[7.3] Horizontal Modification Conditionals
1209While this mechanic is neither a macro nor a button bind, it
1210still falls under a separate category. On the B0XX, the 2
1211modifier buttons don’t always modify the horizontals to
1212X .7375/.2875. Instead, they present the option to modify the
1213horizontals when doing so would be useful (based on the number of
1214horizontals currently held). Without this mechanic, the modifier
1215buttons’ X-axis modification can actually be a hindrance in
1216certain scenarios.
1217Only the horizontals are subject to this mechanic (the quadrants
1218are exempt). This is also the only instance on the entire
1219controller of an overriden cardinal mattering (if 2 horizontals
1220are held).
12211 Horizontal Held
1222X is modified to .7375/.2875. If only 1 horizontal is held, this
1223will always match the player’s intention.
12242 Horizontals Held
1225X is not modified (it remains 1.0). This is because when 1
1226horizontal is already held, modifying the second horizontal is
1227almost always unwanted.
1228The best example would be ledgedashing with Modifier 1. Modifier
12291 enables the 30.5° wavedash, meaning it’s safe to assume the
1230player’s intention is to travel as far as the B0XX permits. In
1231this situation, an experienced B0XX user would want to jump with
1232X 1.0 trajectory to gain as much distance as possible, but
1233Modifier 1’s modification to X .7375 would normally interfere. By
1234making it so that the modifier buttons do not cause an X-axis
1235modification when the second horizontal is pressed, the option to
1236press Back to fall from the ledge, Forward (X 1.0) + jump, then
123765 
1238Down-Forward + airdodge is no longer arbitrarily compromised.
1239This is the most difficult (and rewarding) ledgedash motion that
1240can be performed on the B0XX.
1241Modifier is Pressed When 2 Horizontals are Already Held
1242As mentioned in Section 2.2.3, this is the first of two instances
1243of order-dependency on the B0XX. It is implemented to account for
1244the one scenario in which the player would want to modify the Xaxis
1245while (already) holding 2 horizontals: pivot F-tilt.
1246In Section 5.1.2, I showed the best pivot F-tilt method on the
1247Gamecube controller. This method involves flick pivoting,
1248followed by pressing A while the stick is in X-tilt during its
1249return. Since digital inputs do not traverse X-tilt upon release,
1250they cannot use this method, which means they need an
1251alternative.
1252This alternative is dashing twice, followed by modifying the Xaxis
1253to X-tilt (valid territory for the pivot as well as the Ftilt).
1254In order to support this sequence without compromising the
1255ledgedash sequence shown a moment ago, this X-axis modification
1256will only occur if a modifier button is pressed when 2
1257horizontals are already held.
1258Even without this interaction, the B0XX would still be able to
1259perform pivot F-tilt (since X-tilt would still be producible by
1260holding 1 horizontal). This interaction merely makes it so that
1261you do not have to arbitrarily release the second horizontal in
1262order to do so. The B0XX should recognize that whenever a
1263modifier button is pressed after a horizontal, the player’s
1264intention is always to modify the X-axis to X-tilt.
126566 
1266*****************************************************************
1267[8] Non-Dedicated Modifiers
1268*****************************************************************
1269Squeezing as much functionality as possible into Modifiers 1 and
12702 was only the first step. The real breakthrough that kept the
1271number of modifier buttons on the B0XX to two was allowing A/B/C/
1272L/R/X/Y/Z to possess modifier properties. I realized this was a
1273necessary feature when I couldn’t find ways to accommodate
1274certain coordinates with two modifier buttons alone, but I
1275couldn’t crowd the layout of the controller, either. At that
1276point, the solution was to use existing buttons to pinpoint the
1277remaining coordinates the B0XX needed. This removed the need to
1278install additional modifier buttons.
1279Once I established that non-dedicated modifiers were necessary in
1280some capacity, I quickly realized that A/B/C/L/R/X/Y/Z's action
1281input had to remain actuated at all times (even while the button
1282served as a non-dedicated modifier). At first, I experimented
1283with the A/B/C/L/R/X/Y/Z buttons influencing the analog X/Ycoordinates
1284without generating an action input at all (just like
1285modifier buttons). This seemed like the obvious way to
1286accommodate the coordinates the B0XX needed without turning
1287A/B/C/L/R/X/Y/Z into action-direction button binds, but it
1288resulted in several sources of conflict. Most notably, it had
1289become possible to “flub” certain action inputs if the buttons
1290they interacted with were being held. For example, if Up + Left +
1291Modifier 2 were held and C-Up was pressed with the intention of
1292U-smashing, the B0XX would interpret this as an attempt to angle
1293Firefox and fail to generate a C-Up input. I dismissed this idea
1294almost as it came for this reason alone.
1295The need to accommodate niche coordinates without compromising
1296their corresponding action inputs led to my next epiphany:
1297action-direction button binds didn’t have to be harmful.
1298Modifiers already weren’t allowed to generate an Up, Down, Left,
1299or Right input on their own, which took most of the concerns
1300raised by action-direction button binds out of the picture (i.e.
1301U-tilt button, Shine button, etc.). This meant that the remaining
1302concerns laid within the cardinals and quadrants themselves.
1303Melee’s coordinate plane was bound to have more than just eight
130467 
1305meaningful thresholds (4 cardinals/4 quadrants); however, if I
1306could manage to define the rest of these thresholds and deem it
1307illegal for non-dedicated modifiers to traverse them, then the
1308concerns raised by action-direction button binds would be fully
1309alleviated. At that point, the B0XX’s “action-direction button
1310binds” would merely be buttons that interchangeably served as
1311action inputs or directional inputs, but never both at the same
1312time in a meaningful way. This would allow the controller to
1313operate on the minimum # of buttons possible, a quality of life
1314improvement that cannot be overstated.
1315[8.1] Restrictions
1316This section will define the thresholds within the cardinals and
1317quadrants that dictate what constitutes meaningful execution, and
1318then deem it illegal for non-dedicated modifiers to traverse
1319them. For the most part, this means paying respect to the zones
1320of the grid I covered in Section 3.2.
132168 
1322[8.1.1] Tilt/Smash Integrity
1323X .8000 and Y .6625 serve as the dividing lines for many
1324techniques. As a result, manually entering, then leaving these
1325zones is commonly required. I will use fastfalling (Y <= -.6625),
1326followed by wavelanding with two different airdodge angles to
1327illustrate the importance of the tilt/smash thresholds.
1328Fastfalling, then wavelanding with airdodge coordinates that
1329contain Y <= -.6625 can be performed with a single stick motion.
133069 
1331Wavelanding with Y > -.6625, however, requires Y -.6625 to be
1332manually traversed with a second stick motion. This concept
1333applies to various techniques along either axis.
1334The B0XX must respect these thresholds as well. A/B/C/L/R/X/Y/Z
1335may not cause the analog stick to traverse X .8000 or Y .6625 in
1336a manner that meaningfully circumvents having to perform a stick
1337motion. The clause at the end is included because several of the
1338non-dedicated modifiers can traverse the tilt/smash thresholds in
1339situations where they are of no relevance (i.e. angling up-B).
134070 
1341[8.1.2] 50° Line Integrity
1342The division caused by the 50° line is most notable for dictating
1343what constitutes genuine ledgedash behavior.
1344When attempting to ledgefall with the Gamecube controller, < 50°
1345territory must be avoided (as it will cause regular get-up). >
134650° territory, however, is valid for a ledgefall. This means that
1347a > 50° ledgedash (ledgefall -> jump -> airdodge at > 50°) can
1348all be performed with a single stick motion.
134971 
1350Ledgedashing at < 50°, however, requires the 50° line to be
1351manually traversed with a second stick motion.
1352As far as the B0XX goes, this means that pointing in > 50°
1353territory must be paired with actions that are in > 50° territory
1354as well (and vice versa). In other words, A/B/C/L/R/X/Y/Z may not
1355cause the analog stick to traverse the 50° line in a manner that
1356meaningfully circumvents having to perform a stick motion.
135772 
1358This diagram embodies 50° Line Integrity. The two green sequences
1359are examples of valid ledgedash motions: the player ledgefalls in
1360either west, southwest, or south, then manually points at
1361southeast in order to airdodge at < 50°. The red sequence is an
1362example of an illegal ledgedash: the player ledgefalls in
1363southeast, then, the L/R button (airdodge) automatically
1364traverses the 50° line for them.
136573 
1366[8.1.3] Down-B/Side-B Integrity
1367Pointing in either down-B or side-B territory means being unable
1368to perform certain other actions. This usually isn’t relevant
1369(since most special moves are not cancellable), except for with
1370Fox and Falco’s Shine.
1371If you want to Shine, then wavedash at an angle in down-B
1372territory, a single stick motion will do the job.
137374 
1374However, if you want to Shine, then wavedash at an angle in sideB
1375territory, a second stick motion will be needed; therefore, if
1376pressing the B button would produce a down-B, A/L/R/X/Y/Z may not
1377redirect the analog stick to side-B territory (and vice versa)*.
1378C is exempt, since the concerns here are action inputs, not
1379necessarily stick motions.
1380*The only exception to this rule is shield (not airdodge)
1381pointing at ~45° regardless of the sitation. This is because the
1382B0XX cannot support a variety of shield angles.
138375 
1384[8.2] Definitely Not Action-Direction Button Binds
1385With Section 8.1’s rules in effect, there is very little that
1386non-dedicated modifiers can still do. Nonetheless, they are
1387needed to tap into a few aspects of Melee’s analog control.
1388Section 8.2 covers cases in which it is not only irrelevant, but
1389strictly disadvantageous that A/B/C/L/R/X/Y/Z must be actuated in
1390order to prompt its corresponding directional modification. These
1391action inputs are inconsequential, however, since your character
1392isn’t actionable in any of these situations.
1393[8.2.1] Firefox
139476 
1395To angle Firefox (and other up-B’s), two modifications must be
1396made to the diagonal directional inputs. First, Modifier 1 or 2
1397is used to denote whether the X or Y-axis is being hugged. Then,
1398the C-stick buttons allow you to choose from angles approximately
13997° apart from each other. The C-stick buttons’ semi-circle
1400arrangement makes this surprisingly intuitive.
1401Firefox Angles:
1402C-Right: +/-.9500 and +/-.2875 (16.8°/73.2°)* ** ***
1403C-Up: +/-.9000 and +/-.4000 (24.0°/66.0°)***
1404C-Left: +/-.8500 and +/-.5125 (31.1°/58.9°)***
1405C-Down: +/-.7875 and +/-.6125 (37.9°/52.1°)****
1406Markings showing where a controller’s “hybrid gate” has been
1407constructed.
1408*The justification for these coordinates hinges on the legality
1409of a case mod that is increasingly common among professional
1410players. On a controller that contains the notches shown in the
1411picture, there is no risk of overshooting into the cardinals when
1412attempting the shallowest/steepest angles in the game. While
1413notches are subject to arbitrary factors (i.e. degradation) that
141477 
1415digital buttons are exempt from, it is theoretically legal to
1416center these on the perfect angles (16.8°/73.2°).
1417**The A button will not work when X +/-.9500 Y +/-.2875 are being
1418pinpointed so that the 4th/5th F-tilt/F-smash angles cannot be
1419performed.
1420***As explained in Section 5.1.4, it is illegal for the B0XX to
1421travel from crouch territory to diagonal dash territory. When
1422held alongside Modifier 1 and diagonal directional inputs, CRight,
1423C-Up, and C-Left are capable of generating a dash (X
1424=> .8000) within the quadrants. As to comply with the dash back
1425out of crouch nerf, Down + Modifier 1 + C-Right/C-Up/C-Left will
1426produce Y -.6125 (squatRV). This makes it impossible to dash back
1427out of crouch with these Firefox coordinates.
1428****These are valid coordinates for pivot UF/DF-tilt. They have
1429been accounted for in the Notch Integrity nerf’s code.
1430[8.2.2] Slight DI
1431Normally, Modifiers 1 and 2 will modify the horizontals to
1432X .7375 and .2875 respectively. To access X .5875 and .4375, hold
1433the A button as well. These coordinates will still produce an Ftilt
1434in the neutral game, but they’ll let you DI less predictably
1435when being thrown. The A button is opposite of Modifier 1, which
1436makes A feel exactly like a modifier button when used for DI.
143778 
1438[8.3] Sort of Action-Direction Button Binds
1439In Section 8.2, there was no relationship between the action
1440inputs and their corresponding directional modifications. C
1441(Firefox) and A (slight DI) were entirely chosen based on their
1442button locations.
1443In Section 8.3, this is not the case: these modifications are
1444prompted by action inputs that directly need them to occur. While
1445this is worth distinguishing, it ultimately doesn’t make a
1446difference. The ability to influence the analog X/Y-coordinates
1447as needed without circumventing meaningful execution remains
1448consistent with these modifications.
1449All of the modifications within this section pertain to the L and
1450R buttons, which are inherently problematic for digital
1451controllers. Since shielding, wavedashing, and airdodging
1452defensively – three very different techniques – are all linked to
1453these buttons, several interactions must be created in order for
1454the B0XX to offer its user the correct options to choose from.
145579 
1456[8.3.1] Shield Tilt (Automatic)
1457It is best that I begin this section with a diagram of the
1458actions that can be performed with the analog stick while
1459shielding:
1460Shield tilt can always occur up until the thresholds for roll,
1461spotdodge, and tap jump, meaning |X| <= .6875 with |Y| <= .6500
1462is always valid. On the ground floor specifically, shield can be
1463tilted down to Y -.6875, since shield drop cannot be performed.
146480 
1465With digital inputs, making the most of shield tilt requires two
1466separate features. This section will focus on automatic shield
1467tilt, which looks to correct the B0XX’s default quadrant
1468coordinates of X +/-.7000 Y +/-.7000. These coordinates are
1469strictly undesirable when shielding because they’ll cause tap
1470jump or spotdodge. Had the player intended to perform these
1471actions, they would have simply pointed in a cardinal direction;
1472therefore, it is redundant for the quadrants to perform them. The
1473L button will make these corrections, as it is the B0XX’s primary
1474shield trigger. Z is sometimes useful as a shield, so it will
1475make these corrections as well.
1476When L or Z is pressed in conjunction with quadrant 1 or 2, you
1477will receive:
1478X +/-.7500 Y .6500*
1479Automatic shield tilt is intended to be used in situations where
1480your ability to roll has been shut off (i.e. dashing then
1481buffering a shield horizontally). Since these situations
1482encompass the vast majority of those that call for shield tilt,
1483automatic shield tilt is usually all you need. In quadrants 1 and
148481 
14852, the values X +/-.7500 Y .6500 are obtained by maxing out the
1486Y-axis (such that it does not cause tap jump), followed by the Xaxis.
1487Because we’re operating under the assumption that roll has
1488been shut off, it is fine for X to be equal to or greater
1489than .7000.
1490*L in conjunction with quadrant 1 or 2 won’t always produce X
1491+/-.7500 Y .6500 (see Section 8.3.4).
1492When L or Z is pressed in conjunction with quadrant 3 or 4, you
1493will receive:
1494X +/-.7250 Y -.6875
1495In quadrants 3 and 4, it is best for shield to default to Y
1496-.6875. Not only does this Y-value avoid spotdodge on the ground
1497floor, but it also causes shield drop on platforms. From there,
1498the X-axis is maxed out. These coordinates (X +/-.7250 Y -.6875)
1499comply with the shield drop down nerf from Section 5.2.3, which
1500stated that all shield drop Y-values on the B0XX needed to be
1501paired with roll X-values.
150282 
1503For transparency’s sake, there is a small complication involving
1504the coordinates X +/-.7250 Y -.6875 that has to do with the
1505dynamics of notches (precision, physical construction, etc.) as
1506explained in Section 5.1.3. Based on the logic conveyed within
1507that section, one could make the argument that pinpointing shield
1508drop in quadrants 3 and 4 requires a degree of precision that is
1509only permissible if the B0XX opts to use its SW/SE “notches” on
1510one of shield drop’s 3 Y-values. To be exact, this is to say that
1511the B0XX should be able to pinpoint either Y -.6625 through
1512-.6875 (shield drop) or Y -.7000 through -.7500 (jab cancel), but
1513not both. Therefore, L and Z modifying to Y -.6875 (shield drop)
1514is problematic, since the B0XX is already capable of pinpointing
1515X +/-.7000 Y -.7000 (jab cancel). This argument would have been
1516sound, had the B0XX been geared for vanilla Melee.
1517On UCF, shield drop and jab cancel overlap.
1518As shown in the diagram, an unavoidable consequence of UCF’s
1519modified shield drop range is shield drop overlapping with jab
1520cancel, making it possible to notch for both techniques on the
1521same controller. This is one of the more subtle ways in which UCF
1522surpasses traditional Gamecube controller hardware.
152383 
1524Under these circumstances, Y -.6625 through -.6875 are reduced to
1525generic shield tilt coordinates, as they are no longer needed for
1526shield drop. It is, therefore, fine for the B0XX to retain X
1527+/-.7000 Y -.7000 (jab cancel/UCF shield drop) as its unmodified
1528quadrant values while modifying to X +/-.7250 Y -.6875 when L or
1529Z is held (to avoid spotdodge). To put this in perspective: had
1530UCF not been the tournament standard, the B0XX would have had to
1531use X +/-.7250 Y -.6875 as its unmodified quadrant values in
1532order to retain shield drop, at which point it would not have
1533been able to jab cancel.
1534[8.3.2] Shield Tilt (Manual)
1535Manual shield tilt, which completes the shield tilt duo by
1536blocking actions such as roll, spotdodge, and shield drop, is far
1537less straightforward to implement than its counterpart. In a
1538perfect world, either Modifier 1 or 2 would have performed manual
1539shield tilt, but they are occupied with analog L 43 and 140
1540respectively. To make matters worse, most non-dedicated modifiers
1541aren’t viable candidates. Unlike Firefox (C) and slight DI (A),
1542manual shield tilt is performed in situations where your
1543character is actionable; therefore, if A/B/C/X/Y was required for
1544manual shield tilt, its action input would surely conflict. This
1545leaves R as the only eligible button for manual shield tilt,
1546since its action input is a shield itself.
1547The first thing to know about manual shield tilt is that R’s
1548modifications will override L and/or Z’s if R is held alongside
1549them. Since manual shield tilt’s coordinates are by all means
1550secondary to those of automatic shield tilt, the intention to use
1551manual shield tilt is clear whenever R is held. Giving R priority
1552allows it to not only serve as a tilted shield on its own, but
1553also a “modifier button” of sorts when used with L or Z.
155484 
1555Having to press Up and R at once would have been poor design.
1556The second thing to know about manual shield tilt is that it
1557cannot be performed in north, quadrant 1, or quadrant 2. These
1558three sections of the grid could have been supported, but I chose
1559not to do so due to Up and R being on separate rows. From a
1560design standpoint, it would have been a mistake to encourage the
1561player to tilt their wrist diagonally to simultaneously press
1562these buttons. Conveniently, shield tilt in north is useless,
1563while automatic shield tilt suffices for quadrants 1 and 2 in
1564almost every situation.
156585 
1566When R is pressed in conjunction with Left or Right and no
1567modifier buttons are held, you will receive:
1568X +/-.6875 Y 0
1569Of R’s three manual shield tilt modifications, this one is most
1570straightforward: X +/-.6875 avoids roll in the horizontals.
1571When R is pressed in conjunction with Down and no modifier
1572buttons are held, you will receive:
1573X 0 Y -.6500
1574Aside from avoiding spotdodge, X 0 Y -.6500 allows you to perform
1575a hidden shield drop method that I mentioned briefly in Section
15765.2.3. By tilting shield downwards in Y-tilt (the light gray
1577area) for 4 frames, the entirety of spotdodge (pink) will turn
1578into shield drop on specifically frames 5 and 6. While there
1579isn’t much reason to use this method over the “Axe method,”
1580it is still nifty that the B0XX is able to accommodate it.
158186 
1582When R is pressed in conjunction with quadrant 3 or 4 and no
1583modifier buttons are held, you will receive:
1584X +/-.6500 Y -.6500
1585Even though the ideal set of coordinates to use within quadrants
15863 and 4 is X +/-.6875 Y -.6500 (since X can extend up
1587until .6875), X +/-.6500 Y -.6500 is used so that R can produce a
158845° vector. This is an unnoticeable hindrance to manual shield
1589tilt that allows R to excel at its other duties (these are
1590wavedash-related; see Section 8.3.3). As far as manual shield
1591tilt goes, these coordinates allow you to tilt your shield
1592diagonally downwards on a platform without shield dropping.
1593Throughout this section, you may have noticed that manual shield
1594tilt does not occur when either Modifier 1 or Modifier 2 is held.
1595The primary reason for this is holding an analog trigger and a
1596digital trigger at once causes the digital trigger to take
1597priority. Since manual shield tilt is tied to digital R, a
1598digital shield would always override analog L 43 or 140 and
1599defeat the purpose of tying them to this feature.
160087 
1601[8.3.3] Wavedash
1602The altered wavedash angles (30.5° especially) are the most
1603frequently used modifications on the B0XX. As with everything
1604else in this chapter, the goal is to incorporate these angles in
1605a manner that doesn’t involve installing additional buttons.
1606Since there are two non-45°* angles and two modifier buttons to
1607work with, achieving this is relatively straightforward.
1608Modifier 1 is associated with < 50° territory, and Modifier 2 is
1609associated with > 50° territory, which makes distributing the
161030.5° and 59.5° wavedashes easy enough. The only complication
1611stems from the fact that the modified quadrants have to preserve
1612certain X and Y coordinates (+/-.2875 and +/-.7375) in order to
1613perform other duties. Since the wavedash angles don’t consist of
1614these coordinates, another modification must be prompted by their
1615action button: R.
1616*As shown in Section 8.3.2, R (in conjunction with no modifier
1617buttons) modifies the analog X/Y-coordinates from X +/-.7000 Y
1618-.7000 to X +/-.6500 Y -.6500 on its own so that it can assist
1619with shield tilting in addition to wavedashing at 45°.
162088 
1621When the modifier buttons and R are pressed in conjunction with
1622quadrant 1, 2, 3 or 4, you will receive:
1623Modifier 1: X +/-.8500 Y +/-.5000* (30.5°) (pictured)
1624Modifier 2: X +/-.5000 Y +/-.8500 (59.5°)
1625At first glance, this would appear to be a blatantly disingenuous
1626action-direction button bind; however, looks can be deceiving.
1627Once this diagram is viewed from the perspective of what these
1628coordinates actually do, it is clear that there was no
1629directional modification of any significance.
1630*These coordinates cannot be exploited to perform dash back out
1631of crouch because the R button will generate a shield.
163289 
1633Theoretical coordinates the B0XX could have used.
1634When an airdodge is calculated, the magnitudes of X and Y are
1635ignored (only the angle between them is relevant). This means
1636that, in theory, Modifier 1 could have given coordinates that
1637produce a 30.5° vector, point in X-tilt + Y-tilt territory, and
1638walk at X .7375 to begin with. X +/-.7375 Y +/-.4375, for
1639example, satisfies all of this criteria (its wavedash angle is
164030.6°, but that’s besides the point). These coordinates can
1641therefore be considered the analog stick’s effective location
1642whenever the modified quadrants (with or without R) are
1643pinpointed on the B0XX.
1644I opted not to use coordinates like X +/-.7375 Y +/-.4375 for the
1645sole purpose of design elegance. Wavedashes are usually performed
1646with the analog stick pressed against the rim, which I wanted the
164790 
1648airdodge coordinates I gave the B0XX to stay true to (not to
1649mention how ridiculous testing all the coordinates that aren’t
1650along the rim would have been). I also wanted the raw modified
1651quadrants to contain X or Y +/-.2875, as hugging the X or Y-axis
1652seemed least arbitrary. R’s modifications should be understood as
1653a small touch I gave the controller so that it could consist of
1654relatable coordinates like these. As far as gameplay goes, they
1655do not make a difference.
1656Lastly, I should revisit the notches I introduced in Section
16578.2.1. While these are excellent for wavedashing at
1658shallower/steeper angles, assessing this correlation gets far too
1659subjective. Whereas up-B’s tend to give you a generous amount of
1660time to make use of these notches, wavedashes must be performed
1661almost instantaneously. Under these circumstances, notches are
1662much less effective than digital inputs. For this reason, I
1663decided not to factor notches into the rationale for my airdodge
1664angles at all.
1665Even if notches hadn’t existed, this logic would still have been
1666sound. Anyone’s angles will, on average, be shallower/steeper
1667when given 42 frames to aim (Firefox) as opposed to 4 (Fox’s
1668wavedash). In a sense, it is fortunate that wavedashes take off
166991 
1670on the same frame as their action inputs (L/R), as this allows
1671the B0XX to restrict them to less shallow/steep angles than its
1672up-B’s. Had wavedashes behaved like up-B’s, differentiating the
1673two techniques wouldn’t have been possible.
1674[8.3.4] Home Row Upwards Airdodge
1675Throughout Section 8.3.3, it was implied that R was only meant to
1676be used for wavedashing (which pertains to airdodging in quadrant
16773 or 4 specifically). This may have come across as an oversight,
1678but it was very much intentional. While R does cause its
1679modifications in quadrants 1 and 2, these modifications are only
1680meant to restrict the B0XX’s airdodges to 30.5°/59.5° (X .7375
1681Y .2875 and X .2875 Y .7375 would have airdodged at 21.3° and
168268.7° respectively). As far as actually performing upwards
1683airdodges goes, using R is discouraged.
1684Once again, this is because it would have been a mistake to
1685formally support the simultaneous use of buttons that are on
1686different rows with the same hand. R was placed in this location
1687because the index finger’s strength is needed for swift and
1688accurate wavedashes (similar reasoning went into B’s placement).
1689When it comes to upwards airdodges, R is unideal.
169092 
1691L, on the other hand (no pun intended...), is in a great
1692location for upwards airdodges.
1693The potential to correct this design flaw lies in the L button,
1694which is in an eligible location for upwards airdodges. The only
1695hurdle is that analog L 43/140 and automatic shield tilt cannot
1696be compromised in the process. In order for L to accommodate
1697upwards airdodges while preserving the rest of its interactions,
1698the second and last instance of order-dependency is needed.
1699In accordance with the three downwards airdodge angles, the goal
1700is to give the B0XX the ability to comfortably airdodge at 30.5°,
170145°, and 59.5° in quadrants 1 and 2. For the L button to make
1702this possible, it must adjust for these duties when it is pressed
1703after Up (this is inclusive of a simultaneous Up + L press).
1704Since directional airdodges can only be performed by pressing L/R
1705after (or alongside) the directional input, these adjustments
1706will always match the player’s intention to airdodge.
1707This means that the only drawbacks to L’s adjustments involve
1708shielding (with L) and shield tilting in quadrant 1 or 2 on
1709specifically the same frame. Due to the inclusiveness of “after”
1710inputs, the B0XX can mistakenly interpret quadrant 1/2 + L on the
1711same frame as an upwards airdodge when the intention was to
171293 
1713shield tilt. Luckily, there is a way to heavily mitigate this
1714interference when it occurs.
1715When L is pressed in conjunction with no modifier buttons and
1716quadrant 1 or 2 and Up is already held, you will receive:
1717X +/-.6500 Y .6500 (45°)
1718Once again, X +/-.6500 Y .6500 are invaluable in that they are
1719compatible with airdodge and shield. These coordinates not only
1720allow an upwards airdodge to take place at 45°, but they also
1721nullify the most common occurrence of the same-frame shield tilt
1722interference. When a digital shield is simultaneously actuated
1723and tilted in quadrant 1 or 2, the only consequence is pointing
1724at slightly less optimal coordinates (X +/-.6500 Y .6500 instead
1725of X +/-.7500 Y .6500). This is unnoticeable.
172694 
1727When L is pressed in conjunction with Modifier 1 and quadrant 1
1728or 2 and Up is already held, you will receive:
1729Digital L + X +/-.8500 Y .5000 (30.5°)
1730Aside from the airdodge angle being modified to 30.5°, L will now
1731produce a digital press (for airdodge) instead of analog L 43.
1732This time around, the drawbacks to the same-frame shield tilt
1733interference are more severe, since this outcome is radically
1734different from the expected lightshield. Due to the infrequency
1735of tilting a lightshield in quadrant 1 or 2 (let alone
1736simultaneously), however, this isn’t a concern.
173795 
1738When L is pressed in conjunction with Modifier 2 and quadrant 1
1739or 2 and Up is already held, you will receive:
1740Digital L + X +/-.5000 Y .8500 (59.5°)
1741Again, L will produce a digital press (instead of L 140), and the
1742airdodge angle is modified to 59.5° as expected. The same-frame
1743shield tilt interference is infrequent to the point that it isn’t
1744a concern.
174596 
1746*****************************************************************
1747[9] Other Interactions
1748*****************************************************************
1749[9.1] UF/DF-Smash
1750X-smash + Y-tilt is UF/DF-smash territory on both the analog
1751stick and C-stick.
175297 
1753Throughout this document, I failed to show a way to pinpoint the
1754regions where X-smash overlaps with Y-tilt (aside from
1755Firefox/airdodge angles, which require C/L/R to be actuated).
1756This is because the B0XX does not possess one. These parts of the
1757grid are responsible for very few functions: they can be used to
1758dash/run with ASDI down, but the C-stick is better for that. They
1759can also be used to guarantee that dash back out of crouch
1760succeeds, but that was banned in Section 5.1.4. The only
1761technique the B0XX truly needs these regions for is UF/DF-smash,
1762which can be performed by 8 characters in the game. Even still,
1763neither Modifier 1 nor 2 can afford to accommodate these regions,
1764as the ones they pinpoint are more important. This leaves the
1765option of using the C-stick to perform UF/DF-smash.
1766With the analog stick, the B0XX inherently has to hold two
1767cardinal directions in order to produce diagonal vectors. Due to
1768the long list of techniques the analog stick has to do with
1769(smash DI comes to mind), it is essential that this remains true.
1770With the C-stick, however, we can afford to break this rule since
1771there is nothing remotely exploitative that could result from
1772doing so. This is especially necessary considering the
1773arrangement of the C-stick buttons; since they are meant for the
1774right thumb, having to press two of them at once isn’t ideal.
1775To prompt these diagonal C-stick vectors, Modifier 1 must first
1776be held in conjunction with Up or Down on the analog stick. This
1777will produce X 0 Y +/-.6500, which won’t inform your opponent
1778that you are pointing in either direction (since these
1779coordinates cause neither tap jump nor crouch). Then, press CLeft
1780or C-Right. Instead of the usual C X +/-1.0 Y 0, this will
1781produce C X +/-.9500 Y +/-.3000, which are valid coordinates for
1782UF/DF-smash. These coordinates comply with the restrictions on
1783Popo’s F-smash desync, and the 4th/5th F-smash angles.
1784[9.2] D-Pad
1785When Modifiers 1 and 2 are held simultaneously and the C-stick
1786has yet to be actuated, the C-stick transforms into the D-pad.
1787Holding Modifiers 1 and 2 simultaneously also shuts both of their
1788directional, analog L, and C-stick modifications off.
178998 
1790*****************************************************************
1791[10] B0XX Advantages
1792*****************************************************************
1793[10.1] Travel Time
1794Most of the remaining advantages on the B0XX stem from the
1795ability to press certain arrow keys in succession without
1796experiencing physical recoil. These can fall under:
1797-Cardinal to diagonal. The B0XX can perform these inputs on frame
17981 then 2. This is 100% unremovable, since any sort of timingbased
1799lockout would conflict with the ability to perform basic
1800functions that involve pointing in the quadrants.
1801-Cardinal to opposite cardinal. The B0XX can perform these inputs
1802on frame 1 then 2 as well. Even though this sequence could have
1803been toned down through the use of lockouts, I opted not to do so
1804for a few reasons. For one, unlike the SDI nerf (which bans the
18053rd SDI input for 6 frames in order to fill up <= 9 frame hitlag
1806windows), enforcing a lockout on opposite cardinals would have
1807been arbitrary, since there isn’t a definitive number of frames
1808in this case. Furthermore, enforcing a lockout would have been
1809overkill. Whereas the SDI nerf completely disincentivizes the
1810button sequence it targets, an opposite cardinal lockout would
1811have incentivized performing a button sequence as quickly as
1812permitted. This would have caused the player to compete against
1813their own controller’s lockouts, an entirely artificial
1814diversion.
1815It should also be noted that both of the aforementioned sequences
1816are risky to attempt so quickly (frame 1 then 2), as they won’t
1817be sequences at all if both arrow keys are input on the same
1818frame. Frame 1 then 3 generally gives much better risk/reward.
1819Lastly, the following analog stick motions/difficulties should be
1820kept in mind for Sections 10.1.4, 10.1.5, 10.1.6, and 10.1.7:
1821-Frame 1 West (X -1.0), Frame 5 East (X 1.0): Easy
1822-Frame 1 West (X -1.0), Frame 4 East (X 1.0): Difficult
1823-Frame 1 West (X -1.0), Frame 3 East (X 1.0): Humanly unrealistic
1824-Frame 1 West (X -1.0), Frame 2 East (X 1.0): Impossible
182599 
1826[10.1.1] Quarter-Circle Smash DI
1827With an analog stick, quarter-circle SDI inputs should take place
1828on frame 1 then either 3 or 4. With digital inputs, frame 1 then
18292 is possible. This can sometimes result in an additional SDI
1830input within a hitlag window.
1831100 
1832[10.1.2] Fastfall -> Side-B
1833Digital inputs make it easier to fastfall on frame 1 then side-B
1834on frame 2. This is strictly relevant for Samus, who only has a
18352-frame window to perform short hop fastfall missile.
1836[10.1.3] Run -> Crouch -> U/UF-Tilt
1837Run can be cancelled with crouch, which can then be cancelled
1838into any attack on the very next frame. Whereas U/UF-tilt are
1839physically far away from crouch on an analog stick, they are
1840immediately accessible with digital inputs.
1841[10.1.4] Moonwalk
1842Digital inputs allow you to alternate the horizontals on frame 1
1843then 2 to perform the best moonwalk in the game.
1844101 
1845[10.1.5] Dash Back -> Dash Back
1846Once a dash forward is initiated, it can be cancelled into a dash
1847back on frame 5. A dash back, however, can be cancelled into
1848another dash back on frame 3. Digital inputs make a single
1849repetition of dash back (frame 1) -> dash back (frame 3) entirely
1850possible (the rate of subsequent repetitions will be on par with
1851that of an analog stick).
1852[10.1.6] Dash -> Jump With Backwards Trajectory
1853With characters who have 3 frames of jumpsquat, it is difficult
1854to dash in one direction (frame 1) then reach the opposite
1855direction in time for trajectory to be calculated on the final
1856frame of jumpsquat (frame 4) with an analog stick. Digital inputs
1857make this a breeze, since they can already be there by frame 2.
1858102 
1859Surprisingly, dashing then jumping with backwards trajectory is
1860almost completely useless. This is because it forces you to
1861commit to an overly defensive jump arc. These situations are
1862characterized by launching a counterattack on your opponent,
1863which neutral trajectory is more well-suited for. Neutral
1864trajectory allows you to react, then make your decision, whereas
1865backwards trajectory locks you into relinquishing all pressure.
1866[10.1.7] Aerial Drift
1867The impact digital inputs have on aerial drift is largely
1868misunderstood. While they are better overall, this is not nearly
1869to the degree some people believe, nor for the reasons that might
1870appear to be true.
1871Most assessments of aerial drift fail to account for jump
1872trajectory, which is an entirely separate mechanic. As shown in
1873Section 10.1.6, jump trajectory is calculated based on your Xvalue
1874on the final frame of jumpsquat. This locks your character
1875into an arc that cannot be exceeded once it has been determined.
1876For example, X .2875 trajectory followed by X 1.0 aerial drift
1877(on every airborne frame) will travel significantly less far than
1878X 1.0 trajectory followed by X 1.0 aerial drift. In competitive
1879play, it is usually best to takeoff with either X 0 or X +/-1.0
1880trajectory. The former allows you to assess the situation before
1881making a decision, while the latter gives you the most potential
1882to drift in either direction.
1883Neither digital inputs nor an analog stick ever struggle to
1884takeoff with X 1.0 trajectory (excluding the situation I covered
1885in Section 10.1.6); however, preparing to perform aerials with
1886the A button can arbitrarily cause conflict. The coordinates with
1887the greatest X-value that produce a D-air, for example, are X
1888+/-.6375 Y -.7625; therefore, pointing in D-air territory prior
1889to takeoff results in your jump trajectory being stunted.
1890For this reason, it is always best to C-stick the 4 directional
1891aerials. The analog X/Y-coordinates can then be aimed as desired,
1892leaving N-air as the only aerial that conflicts with jump
1893trajectory. This is pertinent to the matter at hand because the
1894B0XX’s button layout encourages you to always C-stick your
1895103 
1896directional aerials, whereas the Gamecube controller’s does not.
1897I mentioned this in Section 2.1.2 when I criticized the Gamecube
1898controller for its C-stick being inaccessible without a grip most
1899players aren’t willing to employ. This tends to result in B0XX
1900users having a jump trajectory advantage over Gamecube controller
1901users despite the option being there to nullify this in full. The
1902B0XX’s Nunchuk-compatibility will likely open people’s eyes to
1903just how much of an inconvenience this is.
1904As far as aerial drift itself goes, both controllers have their
1905advantages. Unlike with jump trajectory, which favors polarizing
1906coordinates like X 0 and X +/-1.0, an analog stick has merit when
1907it comes to aerial drift. Through its intuitive design and full
1908range of X-values, an analog stick allows you to effortlessly
1909point at where you want to go. This is usually much more
1910efficient than repeatedly tapping X -1.0 and X 1.0 to station
1911yourself with digital inputs (since using the X-values in between
1912is even more difficult).
1913On the flipside, an analog stick cannot compare to how effective
1914digital inputs are at alternating one horizontal, then the other.
1915Whereas an analog stick will typically complete this sequence by
1916frame 4 or 5, digital inputs can complete it by frame 2. This
1917shines when aerial drifting, since it is the only movementrelated
1918area of the game where directional inputs will always
1919produce immediate results.
1920Currently, I am inclined to believe that the Gamecube controller
1921is advantaged at aerial drifting in one direction (i.e. East
1922only), while the B0XX is clearly advantaged at alternating two
1923directions. Even though it doesn’t always come into play, the
1924B0XX’s advantage is less replaceable, making it more valuable
1925overall. This is because digital inputs can simulate fine analog
1926control (though this requires a high degree of skill), whereas
1927their lack of recoil truly cannot be replicated with an analog
1928stick.
1929104 
1930[10.2] Precision
1931In most cases, it is easy to remove precision-related advantages
1932from the B0XX. This is usually a one-step process that involves
1933banning the coordinates in question. Despite this tendency, there
1934remains a notable instance of the B0XX inherently being more
1935precise. This pertains to a basic function that cannot be
1936removed.
1937[10.2.1] No-Fastfall from Ledge
1938When ledgedashing, it is theoretically always best to fall from
1939the ledge (on frame 1) without fastfalling on frame 2. This
1940causes your character to remain at nearly the same elevation on
1941frame 2, which can be used to create 2-frame windows on your
1942fall, jump, and airdodge inputs (as opposed to 1-frame windows
1943had you fastfell).
1944Fox’s Ledgedashes:
1945 Frame 1 Fall Fall Fall
1946 Frame 2 Jump No-Fastfall Fastfall
1947 Frame 3 -- Jump Jump
1948 Frame 4 -- -- --
1949 Frame 5 Airdodge -- --
1950 Frame 6 Airdodge --
1951Frame 7 Airdodge
1952With Fox, for example, ledgedashing is most consistent* if you
1953time your no-fastfall for frame 1.5, jump for frame 2.5, and
1954airdodge for frame 5.5. This is because Fox’s frame 5 and frame 6
1955ledgedashes are interchangeable (since they both require the
1956airdodge to take place 3 frames after the jump), which makes it
1957best to time your inputs for in between frames (to pad yourself
1958with leniency in either direction). A no-fastfall is needed on
1959frame 2 to make the frame 6 ledgedash a part of this equation.
1960*These are the most consistent inputs timing-wise. Jump
1961trajectory and airdodge shallowness are separate variables that
1962this chart fails to account for.
1963105 
1964With the Gamecube controller, there are three ways a no-fastfall
1965from ledge can be performed. Since each of these methods have
1966unique characteristics, I’ll go over them one at a time.
1967Analog Stick / Back
1968Despite being widely perceived as the best way to ledgefall,
1969pointing the analog stick backwards is the worst of the three nofastfall
1970from ledge methods, since it defeats its own purpose.
1971This is because it compromises your ability to jump with neutral
1972or forwards trajectory on frame 2, making it nearly impossible to
1973perform the most intangible ledgedash. Since jumping with
1974backwards trajectory automatically results in a SD when
1975ledgedashing (because it prompts the backflip animation), this
1976method usually forces you to wait until frame 3 to jump safely,
1977resulting in only 1-frame leniency on your jump and airdodge.
1978Analog Stick / Down (Y -.2875 through -.6500)
1979Slightly pressing downwards on the analog stick is probably tied
1980for the best way to no-fastfall from ledge but isn’t without its
1981downsides. Since overshooting into Y <= -.6625 will cause you to
1982fastfall, there is a learning curve to this method. Jump
1983trajectory is also compromised by this method due to the gentle
1984nature of the motion it requires (your analog stick probably
1985won’t be extended that far horizontally by the time you jump).
1986C-Stick / Back or Down
1987A claw grip-exclusive: using the C-stick to ledgefall is also
1988perfectly viable but suffers from similar problems. When using
1989this method, you must be wary of alternating between the Cstick’s
1990cardinals and quadrants, which will cause your character
1991to perform an aerial and (most likely) SD. For example, if you
1992fall with C-Left on frame 1, then shift to C-Down-Left on frame
19932, a B-air or D-air will come out. As a result, this method
1994requires a degree of precision comparable to that of gently
1995pressing the analog stick downwards. Jump trajectory is also
1996compromised by this method, since you’ll have to refrain from
1997jamming the analog stick horizontally until frame 2 or later (if
1998you simultaneously point in ledge get-up territory on the analog
1999106 
2000stick and ledgefall territory on the C-stick on frame 1, the
2001ledge get-up takes priority).
2002All in all, there are two viable no-fastfall from ledge methods
2003on the Gamecube controller, both of which incur risk. On the
2004B0XX, all three no-fastfall from ledge methods are viable, and
2005none of them incur risk. Pressing the analog stick backwards
2006becomes viable since physical recoil no longer exists, while the
2007other two methods cannot fail (since overshooting into
2008undesirable territory is impossible). This makes the B0XX more
2009consistent at no-fastfall from ledge than the Gamecube
2010controller.
2011It is also worth noting that a byproduct of the Nunchuk B0XX is
2012the third method becoming risk-free. Since its C-stick is
2013comprised of digital buttons, the Nunchuk B0XX isn’t subject to
2014the challenge of not shifting between the C-stick’s cardinals and
2015quadrants. This allows a “Gamecube controller” to guarantee
2016itself no-fastfalls from ledge.
2017Despite this, ledgefalling with the C-stick on the Nunchuk B0XX
2018isn’t always the best option. For that matter, none of the nofastfall
2019from ledge methods (on the Gamecube controller, B0XX, or
2020Nunchuk B0XX) covered in this section are unequivocally the
2021“best” way to ledgedash. This is because the most potent strategy
2022is to disregard not fastfalling and jam the analog stick into Xvalue
2023territory in order to perform the most intangible ledgedash
2024with as much jump trajectory as possible (see Section 11.1.1).
2025107 
2026*****************************************************************
2027[11] Gamecube Controller Advantages
2028*****************************************************************
2029[11.1] Hardware
2030The Gamecube controller has scattered advantages stemming from
2031the physical construction and/or inner workings of its analog
2032stick. There is no common denominator among these other than that
2033they are hardware-related.
2034[11.1.1] Most Intangible Ledgedash
2035In Section 10.2.1, I examined the theoretical best timings for
2036the (no-fast)fall, jump, and airdodge of a ledgedash. While
2037following the steps provided in Section 10.2.1 will increase your
2038ledgedash consistency*, doing so will stunt your ledgedash
2039potency. This is because, on a human level, all of the nofastfall
2040from ledge methods (on the Gamecube controller, B0XX,
2041and Nunchuk B0XX) are suboptimal for jumping with trajectory
2042(refer to Section 10.2.1); therefore, the most potent ledgedash
2043method does not look to no-fastfall from ledge.
2044As explained in Chapter 6, jump trajectory is a vital component
2045of ledgedashing. Jump trajectory will always contribute to the
2046distance of a ledgedash, but in some cases it can even determine
2047a ledgedash’s success. For these reasons, it is best to
2048prioritize jumping with trajectory (preferably as much as
2049possible) in some situations. This is done by jamming the analog
2050stick into > 50° territory in order to avoid ledge get-up (< 50°)
2051while also picking up an X-value either prior to, or alongside
2052the jump frame of your ledgedash. In the process, you will most
2053likely traverse Y -.6625 (fastfall) due to the forceful nature of
2054this motion, which means the second most intangible ledgedash
2055will not be an option. This is inconsequential, however, if you
2056are confident in your ability to perform the most intangible
2057ledgedash.
2058*Sometimes, a lack of jump trajectory will cause a ledgedash to
2059fail; therefore, the various no-fastfall from ledge methods only
2060108 
2061increase the consistency of ledgedashes that don’t require jump
2062trajectory to succeed.
2063In particular, it is crucial that you jump with trajectory on the
2064harder stages (FD, YS, PS) in order to reduce the shallowness
2065needed on your airdodge. On the B0XX, this correlation still
2066exists, but it behaves in a much more fixed manner. Since the
2067B0XX’s airdodge angles are set in stone, jumping with trajectory
2068isn’t just recommended, but required for certain ledgedashes.
2069109 
2070Whereas the Gamecube controller can always initiate a fall in >
207150° territory, the B0XX must fall elsewhere to perform < 50°
2072ledgedashes. This makes the B0XX less effective than the
2073Gamecube controller at jumping with trajectory.
2074The dilemma with this is the B0XX is worse at jumping with
2075trajectory than the Gamecube controller. With an analog stick,
2076you’re always able to aim for X-value territory as early as the
2077fall frame (frame 1) of a ledgedash. Since > 50° territory is
2078valid for both the fall and the X-value, it is possible to
2079satisfy all of your criteria before the jump frame (frame 2 or
2080later) even takes place. This effectively creates a 2-frame
2081window (at minimum) to jump with trajectory.
2082On the B0XX, falling in > 50° territory during a < 50° ledgedash
2083sequence is impossible. 50° Line Integrity (one of the nondedicated
2084modifier restrictions) ensures this in order to
2085preserve the challenge of jumping with trajectory; however, this
2086results in it becoming more difficult to jump with trajectory on
2087the B0XX than on the Gamecube controller. Since the B0XX cannot
2088fall with the analog stick pointed forward and perform a < 50°
2089ledgedash, it must point the analog stick forward on the jump
2090frame. This gives the B0XX only a 1-frame window to jump with
2091trajectory when performing the most intangible < 50° ledgedashes.
2092Fox in particular is crippled by this 1-frame window because his
2093frame 5 30.5° ledgedash requires him to jump with trajectory on
2094the harder stages (see Section 6.2.1). This is mitigated by the
2095fact that the 30.5° airdodge allows Fox to perform his second*
2096most intangible ledgedash on frame 6. The jump frame for this
2097ledgedash takes place on frame 3, which gives the player a much
2098more reasonable 2-frame window (frames 2 and 3) to jump with
2099trajectory. It is a must that the B0XX is permitted the airdodge
2100coordinates X +/-.8500 Y -.5000 (30.5°), as Fox’s ledgedash
2101capabilities on the harder stages are astonishingly poor without
2102them. In general, the B0XX is more dependent on the second most
2103intangible ledgedash than the Gamecube controller.
2104*Technically, a frame 6 ledgedash is Fox’s third most intangible
2105ledgedash (and a frame 5 ledgedash is his second). Through ECB
2106(environmental collision box) manipulation, Fox is able to
2107110 
2108ledgedash on frame 4. This ledgedash has not been mentioned
2109throughout this document because it cannot be performed in most
2110situations.
2111When performing Fox’s frame 5 ledgedash, don’t airdodge in the
2112final ~7°!
2113So long as a Gamecube controller user is aware of their
2114character’s intangibility thresholds, their controller is
2115inherently better than the B0XX at performing the most intangible
2116ledgedashes. Jumping with trajectory will not only cause these to
2117travel further, but in some cases, make it on-stage.
2118111 
2119[11.1.2] Actuation Time
2120One of the biggest misconceptions about digital inputs is that
2121their lack of a travel route grants them an actuation time
2122advantage over an analog stick. This would appear to be the case
2123based on a comparison of the two input methods’ analog X/Y
2124readings. Whereas an analog stick motion gets polled at several
2125points along the way, digital inputs skip directly from point A
2126to point B. This creates the illusion that digital inputs actuate
2127faster.
2128The error in this reasoning is that travel time and actuation
2129time aren’t the same thing. Travel time specifically pertains to
2130how long it takes to skip from point A to B once the analog X/Ycoordinates
2131have been actuated. Actuation time, on the other
2132hand, measures how long it takes for the decision to influence
2133the analog X/Y-coordinates to occur in-game in the first place.
2134The latter not only can’t be measured in-game, but most certainly
2135has a travel route; it just happens to be on a physical level.
2136In this case, the actuation time in question is the duration of a
2137B0XX button being physically reached and pressed. While there are
2138tools that can be used to measure an individual controller’s
2139actuation time (such as an oscilloscope), this isn’t a relevant
2140statistic. Since we are only interested in comparing two
2141controllers, the best method is to simply have the same person
2142actuate both of them simultaneously.
2143In December 2017, I conducted several Gamecube controller vs.
2144B0XX actuation time tests to compare how quickly the two
2145controllers initiated a dash (X-smash) (50 attempts per trial).
2146To ensure accurate results, the following assumptions were made:
2147-The Gamecube controller did not have P.O.D.E. (a
2148potentiometer malfunction that causes wonky X/Y-axis
2149readings).
2150-The B0XX uses Sanwa OBSF-24 buttons. These have a 30g
2151actuation force (the lowest of any arcade button on the
2152market).
2153112 
2154-Starting positions were standard. My left thumb rested atop
2155the center of the Gamecube controller’s analog stick, while
2156my right index finger hovered 1cm* above the B0XX buttons.
2157*Maintaining this distance is necessary when playing on the
2158B0XX. Without it, you cannot alternate the horizontals
2159effectively.
2160-Both controllers were actuated swiftly and comfortably.
2161There are also some key pieces of information to convey:
2162-Only the Gamecube controller can be polled in X-tilt when
2163attempting a dash. This is because an analog stick must
2164traverse X-tilt to reach X-smash, whereas digital inputs
2165skip from the deadzone to X-smash. X-tilt actuates faster
2166than X-smash, since it is closer to the center of the analog
2167stick. X-tilt inputs are relevant, and will be tallied.
2168-Only dash forward was performed. Dash forward actuation
2169time was then extrapolated to assess dash back actuation
2170time with 100% accuracy (since dash forward actuation time
2171tells us X-tilt and X-smash actuation time).
2172-Our dash back assessments will operate under the assumption
2173that UCF is on. UCF makes X-tilt valid for the turnaround
2174frame of a dash back, effectively reducing dash back’s
2175actuation time on the Gamecube controller.
2176113 
2177To put these actuation time differences in perspective, the
2178advantages in “units” traveled along the X-axis will be listed.
2179Here is an image for reference:
2180Lastly, Fox was the character used. Here are the the possible
2181outcomes he can receive (the units traveled are approximations):
2182Dash forward (X-tilt)
2183+.2 units on frame 1
2184+.2 units on frame 2
2185+2 units on frame => 3
2186Dash forward (X-smash)
2187+0 units on frame 1
2188+2 units on frame => 2
2189Dash back (UCF) (X-tilt)
2190+0 units on frame 1
2191+2 units on frame => 2
2192Dash back (UCF) (X-smash)
2193+0 units on frame 1
2194+2 units on frame => 2
2195114 
2196Here are the results from the first trial:
2197GCC X-tilt 2 frames before B0XX X-smash: 7 occurrences
2198Dash forward: GCC 2.4 units advantage
2199Dash back (UCF): GCC 4 units advantage
2200GCC X-smash 1 frame before B0XX X-smash: 5 occurrences
2201Dash forward: GCC 2 units advantage
2202Dash back (UCF): GCC 2 units advantage
2203GCC X-tilt 1 frame before B0XX X-smash: 23 occurrences
2204Dash forward: GCC .4 units advantage
2205Dash back (UCF): GCC 2 units advantage
2206GCC / B0XX X-smash on same frame: 5 occurrences
2207Dash forward: Tie
2208Dash back (UCF): Tie
2209GCC X-tilt / B0XX X-smash on same frame: 10 occurrences
2210Dash forward: B0XX 1.6 units advantage
2211Dash back (UCF): Tie
2212In this particular trial, the Gamecube controller averaged a 1.04
2213units advantage on dash forward, and a 1.68 units advantage on
2214dash back (UCF).
2215All trials were consistent with these results. While this can’t
2216be seen on-screen, the Gamecube controller has a clear actuation
2217time advantage over the B0XX on UCF. At tournaments running on
2218Arduino adapters (which fix dash back by killing 1 frame of Xtilt),
2219this advantage is mostly neutralized.
2220115 
2221[11.1.3] Wank Smash DI
2222Popular within the Smash 64 community, wank SDI is an aptly named
2223technique that allows you to generate SDI inputs as rapidly as
2224humanly possible. Recently, Wizzrobe has shown that while it may
2225not be as necessary, wank SDI is just as effective in Melee.
2226The only way to survive in Smash 64.
2227Wank SDI is performed by situating the analog stick against the
2228rim with your left thumb, then vibrating the controller back and
2229forth with your right hand. This requires you to slightly adjust
2230your grip on the controller, but the reward is worth it. Wank SDI
2231will generate SDI inputs every 3-4 frames (i.e. frame 1, 5, 9,
223212, 16...) with the Gamecube controller, and can be performed
2233indefinitely. In terms of speed, this is on par with a burst of
2234quarter-circle SDI, but it comes with the advantage of removing
2235the need to time your SDI inputs.
2236The B0XX does not have a wank SDI equivalent.
2237116 
2238[11.2] Analog
2239The Gamecube controller has higher potential in several areas due
2240to it having the full range of analog X/Y, C X/Y and L/R-values.
2241[11.2.1] Lightshield
2242Whereas the Gamecube controller contains the full range of analog
2243L/R 43 through 140, the B0XX is only able to pinpoint L/R 43 and
2244140. This would have been a one-way advantage if not for the
2245B0XX’s ability to pinpoint L/R 43, which normally requires a
2246precise input on the Gamecube controller.
2247With the Gamecube controller, there are two ways to bypass the
2248precision needed to pinpoint L/R 43; however, neither of them are
2249without their flaws. Most commonly, a physical stopper is
2250inserted in the L/R trigger. This lets the player press down
2251forcefully and be halted at the analog L/R-value of choice. While
2252stoppers serve their intended purpose, they come with the
2253downside of preventing the trigger from being digital pressed.
2254This is why, in my opinion, the better method is to manipulate
2255the trigger’s calibration upon plugin. In doing so, L/R 44
2256through 140 can be disabled on a software level. This allows L/R
225743 to be pinpointed by resting the slider atop its digital press.
2258The second method allows you to pinpoint L/R 43 with one trigger
2259and have access to the entire analog L/R range with the other.
2260This would have been flawless if not for the existence of analogdigital
2261transition, a game mechanic that can cause your shield to
2262fail to protect you from physical attacks. Analog-digital
2263transition occurs when you are polled in L/R 43 through 140,
2264followed by digital L/R on the next frame. This results in an
2265analog shield protecting your character on frame 1, no shield
2266protecting your character from physical attacks on frames 2 and
22673, then the expected digital shield on frame 4 (despite a shield
2268being visibly displayed the entire time). Similar to vanilla
2269Melee’s dash back dilemma, this polling sequence is impossible to
2270account for.
2271Luckily, a Gamecube controller can (and should) be made immune to
2272analog-digital transition. By removing the spring from an L/R
2273117 
2274trigger (or manipulating its calibration), the entirety of L/R 43
2275through 140 can be disabled. This guarantees digital L/R presses
2276with the modified trigger, but conflicts with being able to
2277devote a trigger to analog L/R 43.
2278Based on this information, there are two L/R trigger arrangements
2279worth considering:
2280Arrangement A
22811 trigger without a spring
22821 trigger with a spring / calibrated to pinpoint L/R 43
2283Arrangement B
22841 trigger without a spring
22851 trigger with a spring / regular calibration
2286If arrangement A is chosen, then the B0XX is advantaged over the
2287Gamecube controller due to its ability to select L/R 140 (which
2288arrangement A cannot select); however, this situation should not
2289exist, as the stronger choice is easily arrangement B*. While
2290arrangement B lacks the ability to pinpoint L/R 43, this is more
2291than made up for by its ability to select from the entire analog
2292range (especially since the Z button can be used to pinpoint an
2293L/R 49 shield).
2294*Arrangement A is probably stronger if your game plan is to camp
2295the top platform.
2296Compared to arrangement B, the B0XX does have the advantage of
2297being able to pinpoint L/R 43; however, nerfing this is
2298unwarranted. The B0XX’s analog L/R capabilities are already
2299heavily disadvantaged due to their complete lack of a middle
2300range, which being able to pinpoint L/R 43 does not make up for.
2301Furthermore, a nerf would have nearly no impact, since the only
2302non-arbitrary L/R-value to give the B0XX would be 49 (the
2303strength of the shield generated by the Z button).
2304118 
2305[11.2.2] Shield Tilt
2306The B0XX is only able to support one to two sets of shield tilt
2307coordinates per section of the grid. While this is mitigated by
2308the fact that most of these coordinates extend the B0XX’s shield
2309as far as the game permits, the Gamecube controller’s ability to
2310tilt its shield in any direction within |X| <= .6875 |Y| <= .6500
2311is ultimately superior.
2312[11.2.3] Trajectory DI
2313While the B0XX can TDI (trajectory DI) at angles that aren’t 45°
2314(through the use of Firefox coordinates, wavedash coordinates,
2315etc.), this feature not only isn’t formally supported, but
2316suffers from a massive intuition disadvantage. Overall, this
2317isn’t a concern, as TDI’ing in increments of 45° is surprisingly
2318adequate.
2319The B0XX’s main TDI-related disadvantages stem from its
2320horizontal TDI capabilities (or lack thereof). While these are
2321similarly unintuitive, the real problem is the shortage of
2322options. Since the B0XX only contains
2323X .2875, .4375, .5875, .7375, and 1.0, it cannot perform
2324ambiguous DI mix-ups. Perhaps even more importantly, slide-off DI
2325(influencing your character to slide off the end of a stage or
2326platform in order to regain actionability) is often impossible.
2327These two disadvantages simplify the punish game for your
2328opponent very frequently.
2329119 
2330[11.2.4] Automatic Smash DI
2331The B0XX’s C-stick can only pinpoint C X or Y +/-1.0, C X
2332+/-.7000 Y +/-.7000, and C X +/-.9500 Y +/-.3000 (for UF/DFsmash),
2333which means it is missing the vast majority of its
2334coordinate plane. This limits the B0XX’s ASDI (automatic smash
2335DI) options.
2336Fox’s ECB (the orange diamond) is not above the platform during
2337frames 31-40 of his forwards techroll.
2338For the most part, this isn’t an issue, as it is usually best to
2339ASDI in one of the four cardinal directions with C X or Y +/-1.0.
2340However, there is a notable exception that pops up frequently in
2341competitive play. In techroll situations like the one shown in
2342the image, it is common to use DSDI (double-stick DI, a technique
2343that utilizes TDI from the analog stick and ASDI from the C-stick
2344120 
2345at once to prompt a platform slide-off) in response to being
2346attacked by your opponent. Usually, C Y -1.0 (straight down) is
2347the ideal ASDI coordinate for this, but in certain cases this
2348won’t work. With Fox’s forwards techroll, for example, Fox’s ECB
2349shifts slightly off the platform on frames 31-40. If C Y -1.0 is
2350used here, a platform slide-off will not occur.
2351ASDI’ing in these regions (with the C-stick) will shift Fox’s
2352ECB back onto the platform while preserving as much downwards
2353influence as possible.
2354A few years ago, tauKhan deduced that downwards diagonal ASDI
2355could be used to DSDI with Fox (and potentially other characters)
2356in this situation. C X +/-.2875 Y -.9500, for example, causes
2357downwards influence in addition to X .2875 horizontal influence.
2358This horizontal influence is usually enough (more is needed at
2359higher %’s) to shift Fox’s ECB onto the platform and cause DSDI
2360to succeed.
2361Although the B0XX can pinpoint C X +/-.7000 Y -.7000, these
2362coordinates are not ideal for this technique, since their Y-value
2363of -.7000 only shifts Fox by 2.1 units along the Y-axis (ASDI =
2364X/Y * 3) when he is attacked. The maximum diagonal Y-value of
2365-.9500 (which shifts Fox by 2.85 units) causes DSDI to succeed
2366until much higher percents.
2367For now, the B0XX is incapable of performing DSDI with steep ASDI
2368angles. If I find a way to incorporate this feature, then the
2369B0XX will no longer be disadvantaged in this area. I abstained
2370from doing so for the time being, as I believe the initial
2371release should focus on the fundamentals.
2372121 
2373[11.2.5] Walk/Run
2374Whereas the Gamecube controller can walk with X .2875 through
23751.0, the B0XX can only walk with X .2875, .7375, and 1.0 (without
2376the use of A/B/C/L/R/X/Y/Z).
2377Additionally, X <= .6125 will cause runbrake (the mechanic that
2378causes run to cease) once your character is in the run state.
2379Whereas the Gamecube controller can vary its run between X .6250
2380and 1.0, the B0XX can only modify run to X .7375.
2381[11.2.6] Firefox
2382With the Gamecube controller, you are able to choose from
2383hundreds of Firefox angles. These allow you to weave around
2384edgeguards and sweet-spot the ledge whenever you are in range.
2385With the B0XX, you are only able to choose from 40* Firefox
2386angles. Since these angles are approximately 7° apart from each
2387other, weaving around edgeguards usually means settling for an
2388imperfect one. Similarly, sweet-spotting the ledge is hindered,
2389although measures can be taken to mitigate this. Most of the
2390time, the best strategy is to begin your up-B in a location that
2391offers you the option to sweet-spot the ledge. This solves one
2392problem, but not without creating another: the act of getting to
2393this location can give your opponent the extra few frames they
2394need to set up a successful edgeguard.
2395*Technically, the modified quadrants (X +/-.7375 Y +/-.2875 and X
2396+/-.2875 Y +/-.7375), shield tilt coordinates (X +/-.7500 Y .6500
2397and X +/-.7250 Y -.6875), and airdodge coordinates (X +/-.8500 Y
2398+/-.5000 and X +/-.5000 Y +/-.8500) make for a total of 60
2399Firefox angles.
2400122 
2401[11.2.7] Airdodge
2402The biggest disparity between the Gamecube controller and B0XX is
2403their airdodge capabilities. Between inherent disadvantages and
2404the need for a hard cap on its upper/lower limits, airdodge
2405brings out the worst in a controller that lacks analog control.
2406Usually, I am of the opinion that the B0XX’s intuition-related
2407disadvantages become nonfactors as one gains mastery over the
2408controller. Wavedashing is the main exception to this rule. Since
2409a wavedash must be aimed instantaneously, it is extremely
2410difficult to determine when a 45° or 59.5° airdodge should be
2411chosen over 30.5° (which most players will find themselves
2412defaulting to). An arbitrary disadvantage also comes in the form
2413of being unable to change your mind once you’ve committed to a
2414wavedash angle. On the Gamecube controller, a spontaneous
2415decision can be made to readjust the analog stick during
2416jumpsquat. This cannot be done on the B0XX, since it is usually
2417impossible to shift between the modifier buttons so quickly.
2418Similar to sweet-spotting the ledge with Firefox, certain
2419starting points are preferable when airdodging with the B0XX.
2420When recovering, the especially useful 59.5° (steepest) airdodge
2421must begin at a specific elevation relative to the stage in order
2422to minimize the amount of time your character hovers in the air
2423before landing. Grapple characters (Samus, Link, and Young Link)
2424are hindered in their ability to align their recoveries with the
2425ledge for similar reasons. Most notably, microspacing with
2426wavedashes does not exist. Since the B0XX is only capable of
2427airdodging at three different angles, the ideal wavedash length
2428usually isn’t present. It is recommended that you adapt a neutral
2429game centered around microspacing with dashes instead.
2430Finally, none of the aforementioned disadvantages would have
2431mattered had the B0XX been allowed to wavedash at 16.8°. In
2432giving the B0XX its most necessary nerf, its shallowest wavedash
2433angle goes from being humanly unrealistic to underwhelming. 30.5°
2434may be viable, but any pro would have faith in their ability to
2435consistently beat this benchmark with the Gamecube controller.
2436123 
2437*****************************************************************
2438[12] 1.0 Cardinal
2439*****************************************************************
2440[12.1] Overview
2441Throughout this document, I refrained from acknowledging the
24421.0/.9875 cardinal disparity due to how insignificant, yet
2443lengthy of a subject it is. Ultimately, this is a game mechanic
2444where digital inputs are marginally better than the ideal analog
2445stick as it stands. It is unfair to isolate them, however, due to
2446the glaring differences in 1.0 cardinal efficacy among Gamecube
2447controllers.
2448The root of the problem is a decision made by the game developers
2449that can be considered questionable at best. Within each cardinal
2450(on both sticks), there is a generous range assigned to the value
2451.9875, yet only a single set of coordinates assigned to 1.0.
2452Being the greatest X/Y-value in the game, the latter is naturally
2453more desirable in most situations. When running, for example, Fox
2454will reach a peak acceleration of 2.17 units/frame with .9875, as
2455opposed to 2.20 units/frame with 1.0. While this is an almost
2456unnoticeable difference, the fact remains.
2457124 
2458A Gamecube controller’s analog X/Y-values being read in 20XX
24594.07.
2460In a perfect world, the 1.0 cardinal could have added an element
2461of skill to the game. In reality, however, this isn’t the case.
2462Due to how tiny the 1.0 cardinal’s range is, pinpointing it is
2463almost entirely dependent on hardware. The Gamecube controller
2464shown in the picture, for example, will consistently receive X
2465-.9875 Y -.0125 when its analog stick is pointed to the left.
2466This consistency stems from a plastic case with sharp corners
2467(these help situate the stick in a specific location), while X
2468-.9875 Y -.0125 stems from misalignment due to manufacturing
2469variance. Had it not been for this misalignment, this controller
2470could have consistently pinpointed the 1.0 cardinal.
2471Based on this information, it is clear that finding the 1.0
2472cardinal varies from controller to controller. The next issue,
2473ironically, has to do with another developer’s decision.
2474At the moment, the competitive Melee scene has displayed
2475widespread acceptance of Universal Controller Fix, a game mod
2476intended to fix inconsistencies in Gamecube controller
2477performance. Generally, these inconsistencies stem from poorly
2478designed game mechanics. Dash back, which gives a 1-frame window
2479for an analog input, is an oversight by any developer’s
2480standards. Likewise, shield drop, which has a miniscule range of
2481125 
24823 Y-values, was presumably shafted late into Melee’s development
2483by the inclusion of spotdodge. These are the Gamecube
2484controller’s two most notorious sources of inconsistency, as well
2485as the ones most desperately in need of repair.
2486UCF fixes these two mechanics then closes its doors, despite it
2487not necessarily being correct to do so. Lesser sources of
2488inconsistency, namely the 1.0 cardinal, remain unaddressed as of
2489the current version. This decision appears to have been made for
2490two reasons.
2491For one, fixing the 1.0 cardinal simply isn’t in demand. Unlike
2492with high dash back % and shield drop notches, there has never
2493been widespread incentive to seek out and/or vend controllers
2494that possess 1.0 cardinals (despite the fact that they can be
2495notched). Those who don’t believe UCF should operate on an
2496objective metric tend to dismiss 1.0 cardinals on this basis
2497alone. In patching a game, it is common practice to target
2498mechanics that necessitate the patch in the first place rather
2499than make every conceivable improvement. From this perspective,
25001.0 cardinals do not make the cut.
2501The issue with this view is that it practically concedes that the
25021.0 cardinal meets UCF's criteria, yet dismisses the 1.0 cardinal
2503due to a lack of demand, an entirely arbitrary factor. This
2504decision is then justified by a philosophy that applies to
2505modern-day games that often have thousands of imperfections to
2506attend to, which should not apply to UCF, a mod intended to serve
2507a niche purpose within a game that has a handful of controllerrelated
2508problems at most. The lack of demand is especially
2509meaningless given the lack of knowledge surrounding the 1.0
2510cardinal. Had any effort been made to educate the public about
2511this mechanic, demand for it very well could have been there.
2512This leads to the second counterargument to the 1.0 cardinal’s
2513inclusion, which is founded on falsehoods. Many proponents of
2514UCF, including the dev team themselves, have propagated the
2515belief that UCF merely causes controllers to tie the theoretical
2516best hardware (without exceeding it in any capacity). Under this
2517premise, the 1.0 cardinal receiving a redesign can be dismissed
2518on the basis of the 1.0 cardinal becoming easier to pinpoint than
2519126 
2520on vanilla. The issue with this is that UCF already exceeds the
2521best vanilla controller in several ways.
2522One way to debunk the claim that UCF doesn't exceed the best
2523hardware is to examine the unavoidable byproducts of fixing dash
2524back and shield drop. In Section 5.1.3, for example, I showed why
2525it is impossible for shield drop notches and jab cancel notches
2526to coexist on vanilla. By increasing shield drop’s range, it
2527becomes possible to notch for both on UCF. Even more striking are
2528the implications of fixing dash back, a change that flips the
2529controller lottery on its head. On vanilla, highly elusive
2530controllers “suffer” from a potentiometer malfunction known as
2531P.O.D.E. This gives them high dash back % at a cost. P.O.D.E.
2532compromises pivoting, dashing out of crouch, and several other
2533areas of the game. On UCF, no such trade-off is necessary. Since
2534dash back is distributed on a software level, it is possible to
2535reap the benefits of P.O.D.E. without actually having it. This
2536not only devalues P.O.D.E., but also lets you have the best of
2537both worlds.
2538While these byproducts are telling on their own, there is no need
2539to delve so deeply. The easiest way to debunk the UCF team’s
2540claim is to examine UCF's redesigns of dash back and shield drop
2541themselves. Formerly, no Gamecube controller could successfully
2542dash back 100% of the time or perform shield drops with as little
2543finesse as UCF requires. Even though UCF isn’t that big of a jump
2544from a high P.O.D.E controller (95-98% dash back) with Y -.6750
2545notches (extremely easy shield drops), this is more so meant to
2546illustrate a point. In order to fix inconsistencies, UCF exceeds
2547the best possible hardware.
2548How else could UCF fix inconsistencies? If a mechanic is broken,
2549its redesign inherently has to set the bar higher. Once this has
2550been established, UCF’s rationale must be re-evaluated entirely.
2551The essence of this mod isn’t to tie the best hardware, but to
2552redesign mechanics that make it impossible to achieve equal
2553performance through hardware. Under this premise, the fact that a
2554controller with eight 1.0 cardinals (four on each stick) is
2555nearly impossible to maintain is no longer a relevant detail. The
2556bottom line is that by not redesigning the 1.0 cardinal, we are
2557knowingly accepting a less fair version of Melee.
2558127 
2559[12.2] Redesign
2560Before I can show how the 1.0 cardinal should be redesigned,
2561there is a crucial concept I must illustrate.
2562Shield drop ranges on vanilla (Y -.6625 through -.6875) and the
2563current version of UCF (Y -.6625 through -.7875).
2564When UCF was first introduced, a common complaint was that
2565“shield dropping [had been made] too easy.” After all, UCF had
2566given it a whopping 11 Y-values to vanilla’s 3. While this change
2567allowed everyone to experience the joy of shield dropping, it was
2568initially subject to criticism. Many people felt that shield
2569drop's range should have been increased, but not by this many Yvalues.
2570The hole in this logic is that it evaluates the shield drop mod
2571based on the number of Y-values it converts, rather than how
2572difficult it is to shield drop under the mod. The former is an
2573entirely superficial statistic that can be misleading at face
2574value. This is because, just like on vanilla, two controllers
2575that have the same shield drop range on a software level aren’t
2576necessarily on an even playing field.
2577128 
2578A theoretical shield drop range UCF could have used. 2 Y-values
2579(-.7000 and -.7125) are added to vanilla’s 3. This would have
2580failed to achieve equality among controllers.
2581Say, for example, UCF increased shield drop’s range by only 2 Yvalues.
2582This would have had a drastic effect on some controllers,
2583but none on others. Whereas controllers with corners centered on
2584Y -.7000/-.7125 would have gained the ability to shield drop,
2585controllers with Y <= -.7250 would have been unaffected. This mod
2586would have failed to encompass a significant percentage of
2587controllers.
2588129 
2589Despite their differences, these shield drop ranges are equally
2590easy to pinpoint on most controllers. Either of these would have
2591been acceptable.
2592For this reason, a redesign can only be correct if it guarantees
2593all controllers a 100% success rate. UCF achieves this with its
2594overwhelmingly generous shield drop range of 11 Y-values (top
2595left). It should be understood, however, that there would have
2596been nothing wrong with UCF going even further (top right). The
2597only thing to be wary of when redesigning shield drop is
2598shrinking spotdodge’s range in exchange. This justifies keeping
2599shield drop’s redesign to a minimum.
2600130 
26011.0 cardinal redesigns that I do not recommend.
2602I took the time to establish this concept in order to convey why
2603neither of the above 1.0 cardinal redesigns make sense. It should
2604be obvious that Redesign A, which converts an additional 9 sets
2605of coordinates to 1.0, is a far cry from being acceptable. Sharp
2606corners are still entirely necessary under this redesign, and
2607misalignment remains a significant disadvantage. A controller
2608that begins with a worn down case and X .9875 Y .0375 (to the
2609right), for example, is far more likely to exit this modified 1.0
2610range than one that begins with a sharp case and X .9875 Y .0125.
2611Redesign A would be adequate for some controllers, but not most.
2612131 
2613Redesign B, on the other hand, is by no means bad - it just isn’t
2614logically consistent. Its additional 25 sets of coordinates would
2615encompass the majority of controllers, but performance would
2616still vary ever so slightly. Redesign B is suggestive of wanting
2617to preserve some sort of difficulty in pinpointing the 1.0
2618cardinal, when in reality very few controllers would still be
2619subject to this. Most controllers would already be well outside
2620the realm of ever missing.
2621All X/Y and C X/Y-values that are => .9625 should be converted
2622to 1.0. This is a non-arbitrary cutoff, since => .9625 is
2623cardinal-exclusive (.9500 can be reached in the quadrants).
2624Eventually, things slippery slope their way into justifying
2625Redesign C. Unlike shield drop’s redesign, there is no trade-off
2626(spotdodge’s range) to be wary of when redesigning the 1.0
2627cardinal; Redesign C sacrifices the values .9625 and .9750, which
2628are obsolete to 1.0. There is no reason not to do this, since it
2629helps ensure a 100% 1.0 cardinal success rate across all
2630controllers; therefore, Redesign C is the 1.0 cardinal redesign I
2631encourage the UCF team to implement. If Redesign C (or even
2632Redesign B) went through, the Gamecube controller’s 1.0 cardinal
2633efficacy would tie that of digital inputs, at which point the 1.0
2634cardinal would no longer be a point of contention.
2635132 
2636[12.3] Plan B
2637While I believe the most logical course of action is for the 1.0
2638cardinal to receive a redesign, I am aware that this doesn’t
2639necessarily mean the UCF team will implement one. Political
2640decisions like these aren’t always about correctness. If the UCF
2641team chose to redesign the 1.0 cardinal, it would (ironically) be
2642viewed as the biggest update to their mod yet. This may or may
2643not be a risk they are willing to take.
2644In the event that the 1.0 cardinal is not redesigned, the
2645Gamecube controller’s ability to pinpoint the 1.0 cardinal could
2646only be equal or worse to that of a digital controller. Despite
2647this, there are several reasons the B0XX should be permitted 1.0
2648cardinals (rather than .9875) regardless of the UCF team’s
2649decision. Some of these are relatively straightforward, while
2650others involve more complex interactions.
2651The most direct way to justify the B0XX being permitted 1.0
2652cardinals is to assess the ideal Gamecube controller’s 1.0
2653cardinal efficacy. This is to ask, “How much worse than the B0XX
2654is the Gamecube controller [with the best hardware] at
2655pinpointing the 1.0 cardinal?” The answer is: not much. Despite
2656the 1.0 cardinal’s microscopic range, lucky alignment and sharp
2657corners will make a Gamecube controller incredibly consistent at
2658pinpointing it. While this document does not contain video
2659evidence, this can be surmised by verifying that a Gamecube
2660controller with sharp corners does in fact get polled at the same
2661set of coordinates consistently; therefore, it is within the
2662rules to use a Gamecube controller that most frequently receives
2663the 1.0 cardinal in all four directions (on both sticks). If we
2664are to then base the B0XX off of what is theoretically legal (the
2665only rational approach to take), restricting its cardinals
2666to .9875 would be far more extreme than guaranteeing them 1.0.
2667At this point, the remaining counterargument is that the B0XX is
2668still more efficient than the Gamecube controller at pinpointing
2669the 1.0 cardinal. While this is true, this shouldn’t be taken at
2670face value. Similar to how I encouraged you to evaluate the
2671redesigned shield drop ranges in Section 12.2 based on their
2672133 
2673difficulty rather than their size, the 1.0 cardinal should be
2674evaluated based on its implications, not its magnitude.
2675At first glance, giving the B0XX .9875 cardinals would appear to
2676be consistent with the trend of making the B0XX equal to or worse
2677than the Gamecube controller wherever possible; however, this
2678line of reasoning fails to consider what makes 1.0 cardinals a
2679competitive concern in the first place. Among the areas of the
2680game affected by the 1.0 cardinals, horizontal movement speed is
2681paramount. As shown in Section 11.1.2, the B0XX is already at an
2682actuation time disadvantage in this area that outweighs the
2683advantage gained by having 1.0 cardinals. Based on this statistic
2684alone, it is reasonable to permit the B0XX 1.0 cardinals as
2685compensation.
2686Fox’s frame 6 ledgedash does not succeed on the harder stages
2687with a 30.5° airdodge and X .9875 jump trajectory.
2688Furthermore, there is an interaction involving Fox that tips the
2689scales even more heavily in favor of the B0XX being permitted 1.0
2690cardinals. In Section 6.1.3, I explained that the B0XX was given
2691134 
2692the airdodge coordinates X +/-.8500 Y -.5000 (30.5°) so that Fox
2693could ledgedash on frame 6 on the harder stages, and in Section
269411.1.1 I elaborated on the necessity of this technique. Something
2695I didn’t mention in either of those sections, however, is that
2696Fox’s frame 6 ledgedash only succeeds with a 30.5° airdodge on
2697the harder stages if Fox jumps with X 1.0 trajectory (X .9875
2698jump trajectory will not work). This is due to the inverse
2699correlation between airdodge shallowness and jump trajectory.
2700Since 30.5° is the steepest eligible angle for Fox’s frame 6
2701ledgedash, it is only enabled by the strongest jump trajectory;
2702therefore, if Fox’s jump trajectory is nerfed to X .9875, his
2703airdodge angle must be buffed to X +/-.8625 Y -.5000 (30.1°) in
2704order for his frame 6 ledgedash to succeed.
2705This means that nerfing the B0XX’s cardinals to .9875 would have
2706to coincide with a buff that would roughly negate this nerf’s
2707impact. It must be understood that whether the combination of 1.0
2708cardinals and a 30.5° airdodge or .9875 cardinals and a 30.1°
2709airdodge is chosen makes a negligible difference overall. Neither
2710combination is better across the board; dash-reliant characters
2711(i.e. Captain Falcon) would surely prefer the former combination,
2712while wavedash-reliant characters (i.e. Luigi) would prefer the
2713latter. It is, therefore, reasonable to endorse one or the other
2714based on an assessment of game balance, at which point the
2715combination of 1.0 cardinals and a 30.5° airdodge is clearly the
2716healthier choice. This is single-handedly due to the fact that
2717this combination encourages onstage gameplay with its stronger
2718cardinals, whereas the combination of .9875 cardinals and a 30.1°
2719airdodge encourages ledgedashing with its stronger airdodge
2720angle. Naturally, the former favors the skill set we should be
2721looking to test.
2722For all of the aforementioned reasons, it is clear that the B0XX
2723should be permitted 1.0 cardinals.
2724135 
2725*****************************************************************
2726[13] Conclusion
2727*****************************************************************
2728There are many genres of games in which analog inputs would never
2729be able to coexist with digital inputs. It is a blessing that
2730this is not the case with Super Smash Bros. Melee, the game
2731perhaps most in need of them. As I found ways to replicate most
2732of the Gamecube controller’s intrinsic challenges over the course
2733of the B0XX’s development, it became increasingly apparent that
2734this project was destined to happen. I can’t say that this
2735surprised me; having had this game since the day it came out, I
2736have witnessed its divinity countless times. That being said, I
2737am glad that this worked out.
2738There is no doubt in my mind that the B0XX should be tournament
2739legal in its current iteration. Following its fine-tuning, the
2740B0XX is a well-balanced controller that suffers from several
2741inherent disadvantages but compensates for them with new flavor
2742in other areas. If anything, this is the most we could have asked
2743for from a third-party controller; with a physics engine as
2744complex as Melee’s, it is fitting that the B0XX provides players
2745with a fresh experience.
2746As evidenced by its overwhelming demand, the B0XX will likely go
2747down as one of the biggest leaps forward in the history of Melee.
2748HAL Laboratory may have created a masterpiece back in 2001, but
2749nothing is perfect; in recent years, we have come to embrace the
2750fact that the future of this game lies in our hands. Software
2751modifications and third-party controllers have gone from being
2752radical ideas to household names as Melee remains alive as ever
2753in 2018. It is paramount that the evolution of this industry
2754continues to prosper with the legalization of the B0XX.
2755Thank you for reading,
2756Aziz “Hax$” Al-Yami
2757136 
2758*****************************************************************
2759[14] F.A.Q.
2760*****************************************************************
2761Q: If the B0XX is legal, are any Gamecube controller PCB mods
2762legal?
2763A: No.
2764Q: What does all of this mean for the legality of other digital
2765controller brands?
2766A: This document does not endorse the legality of any controller
2767that does not utilize the latest version of the B0XX software and
2768abide by the B0XX ruleset.
2769Q: How will future updates to the B0XX’s software be carried out?
2770A: Presumably, the Melee It On Me Competition Committee will
2771serve as the hub for the B0XX’s tournament legal parameters.
2772Q: How will tournament organizers verify that someone is running
2773the latest version of the B0XX software?
2774A: At the moment, only a handful of people have a B0XX. Once the
2775product is commercially available, verifying a controller’s PCB
2776will be a one-step process through a downloadable client on our
2777website. The only tools needed will be a USB Type-C cable, a
2778computer, and an internet connection.
2779Q: Should it be standard protocol to verify a B0XX user’s PCB?
2780A: No. Unless there is reason to believe that someone is
2781cheating, their controller should not be examined.
2782Q: Will analog ranges (i.e. Firefox angles) be customizable on
2783the tournament patch?
2784A: No. Even if allowing for customization within the parameters
2785I’ve set wouldn’t raise any concerns, I believe the distributions
2786I’ve chosen are most efficient from a logistical perspective.
2787Q: Can I rearrange my button locations?
2788A: Yes.
2789Q: Will I be able to play on an un-nerfed version(s) of the B0XX?
2790A: Yes. We will provide plenty of recreational patches.
2791137 
2792Q: Will the B0XX support other games?
2793A: Yes. More information TBA.
2794Q: Will a WiiU adapter be needed to use the B0XX on a PC?
2795A: No. More information TBA.
2796Q: When will the B0XX be commercially available, and how much
2797will it retail for?
2798A: July 2018 / $199.99 USD + shipping/tax. Wii Nunchuk sold
2799separately.
2800Q: Will the initial run of the B0XX be sold through a
2801Kickstarter?
2802A: No. The initial run will already have been manufactured by the
2803time it goes on sale.
2804Q: Where can I stay tuned for B0XX-related updates?
2805A: All major updates will be made through my Twitter account
2806(https://twitter.com/ssbmhax) and our website (http://20XX.gg).
2807You can also subscribe to our newsletter through our website, or
2808join our Discord channel (http://20XX.gg/discord).
2809138 
2810*****************************************************************
2811[15] Patch Notes
2812*****************************************************************
281304/16/2018
2814Initial release.
2815139