DDDs Infinite Breakout Data During Pummel, Data on Human Thresholds & Timed Tests

[Hive]

ok here is the revised data, from colin jfs analysis ^^ i hope this clarifies!

And if there are disagreements, at the very least, I hope it promotes intelligent discussion and promotes corrections to have a more solid understanding of the data in the future, since this is such a big issue concerning the infinites against a few characters. I'm also trying to verify what the human thresholds for hitting a button are as well. if you have any online tests, etc... that will help demonstrate this please show me ^^
However the data seems correct.

Also a big ty to adumbrodeus, Natch, bobson, and Ref for all the help/information! ^^
Much of the work goes to ColinJF though!

Conclusion- on how to mash out properly. Hold the controller in a position that is comfortable for you (try to get all 7 buttons in, the b,a, x, y, l, r, and z). your best bet its to try to start hitting your first button ons frame 10 before the pummel has actually started, since this counts in buffering. try to hit the buttons in a rolling kind of way so that none of the buttons are hit on the same frame as each other but do this as fast as you can. While you are doing this rotate the control stick as fast as you can.

I said before that i expect the average you will see in tournaments is people able to break out before 50% when they are doing it right, due to errors. However, I think potentially someone should be able to break out to around, maybe, 140% if everything goes correctly and with practice.

(made by ColinJF) lol reflex must be pissed that he has to be namedropped in all these convos though lol ^^

List of Assumptions Used
(concrete ones to more debatable ones (for the most part)

-The frame data here is correct and reliable.
-DDDs pummel lasts 29 frames.
-The break total is 90 + 1.7*for every percent.
(so 90 at 0%, and 91.7 at 1%).
-For each frame that passes 1 is subtracted from the break total. so if you are grabbed at 0%, after 90 frames, or 1 and a half second you will be released automatically.
-Each button, or cardinal direction of the control stick you press subtracts 8 from the break
out total. A flick of the control stick takes away 16 frames, although it can't be rotated.
-Pressing buttons enough so that their total > the break total constitutes breaking out of a grab.
-DDD can throw you until 33% from 0 until he is forced to
pummel at least once (undegenerated).
-The pummel hits for 4% damage.
-Each game frame is 1/60th of a second.
-note: The human thresholds used are the MAX of what you should expect, NOT the average.
-There are 10 frames of buffering before the pummel that contribute towards you breaking out. (so the amount of time you need to break out is 39 frames total).
-You can only press 1 button plus 1 cardinal direction of the control stick at any given frame (a maximum of -16 per frame).
Ideally this means you want to be able to skew the buttons you are pressing to have them hit on separeate frames.

-The DDD is able to do the infinite correctly and without
unecessary lag.
-The DDD throws you until it is necessary to pummel.
--The person is able to hit his/her first buttons on the first frame available to.

List of Human Assumptions:

-The average player can hit 6-7 button comfortably (both addressed).
The B, A, X, Y, L, R, and possible Z, as well as the control stick.
-In my opinion, The human capacity to hit a button and c-stick direction is about 10 frames or more. Not
including any sort of reaction time. about 6 times a second roughly.
-In my opinion, It takes an equal amount of time to fully release a button as it
does to press one. (not including the control stick which hits
every interval since its not released). (5 frames to push, 5 frames to release)
-It takes no time to put your hand into position to hit the buttons.



The Key

-In order to find out how many inputs per frame one must press to break out we use the equation

(((90+1.7p)/f)-1)/8= inputs per frame to break out. where p=the percentage you are at, and f is the amount of frames you have to break out in.

so for an example, at 0% damage if you are grabbed and thrown until a pummel is necessary, (about 33%) and then pummeled (4%) the percentage you are at is 37% damage before you are able to escape due to him pummeling. you have 39 frames to escape in. the 29 frames of his pummel plus 10 buffering frames. so the total amount of inputs per frame needed to break out is about 0.36.

at 129% the amount reflex broke out in in the vid below, this would mean that the inputs per frame is about 0.87.

-To find the total amount of inputs you need to press in the allotted time, you use the equation ((90+1.7p)-f)/8 this means that at 37% by the 39th frame you need to have had pressed 15 inputs. (rounded up to the nearest whole input).

again, for 129% this tallies up to around 34 buttons.

-To apply this to human thresholds if you want (i will be using my opinions that the avg human capacity to hit a button takes maybe 10 frames to hit and release. the c-stick hits a cardinal direction every 5 frames. obviously differences in opinions are bound to happen (see data sheet and summary). 5 frames means that you can press a single button about 6 times a second though.) for my opinions on what someone should be expected to do in 39 frames this means that a person will be able to hit 36 inputs in 39 frames. this means that in order to achieve a breakout in 39 frames the inputs needed to be pressed are less than 36 in order for it to be within this estimate.



Data Sheet for Human Threshold Equations
its lengthy

in order to make an equation to figure out using your own estimates you have to find out how many frames after hitting a button it takes to hit that button again, if you use the idea that it takes an equal amount of time to hit a button for example, it would then be then be the amount of time it takes to hit a button times 2.
so the equation to figure out how many buttons are pressed in a given frame period is given by
c*(f/(2a) (rounded down to the nearest whole number)), where f, is the amount of frames to do it in, and a as stated above "a" is the amount of frames it takes to hit a button from rest on that button, and c is the number of buttons you can hit comfortably on the controller (and triggers) in one hand position. also 2a>c. why? if 2a is below c the buttons that you are pressing are fully released and ready to be pressed again before the rest of the buttons have been hit and the equation will start to overestimate it.

The above equation assumes that you are hitting all the buttons at equal intervals through the space of 2a, which doesn't happen (which is why it was rounded down). to compensate for hitting them in a cascading pattern,
we take the equation f/2a -(f/2a rounded to the nearest whole number), bc we are going to focus on any parts of the equation that get "fractured" by the frame period. for the rest the above equation produces the same answer.
we then take c/2a which is the fraction of the interval that the buttons c are hit in interval.
for example if it takes you 10 frames to hit a button and return to your original positon (2a) and you are hitting 5 buttons comfortably, only in the 1/2 or the first half of the ten frames you are hitting the buttons.
therefore: if f/2a-(f/2a rd)>c/2a, you add c to your total (since all the buttons are able to hit)(rd=rounded down to the nearest whole number)
if f/2a-(f/2a rd)<c/2a, though, you must find what percent/fraction of buttons are hit within that period. to do this you take (c*(f/2a- (f/2a rd))/(c/2a) rd) and add that to the above equation

Finally if you add the above equations to the control stick equation 1+(f-1)/b, rounded down to the nearest whole number, where b is the time it takes to to hit a cardinal direction on the control stick after just hitting a previous direction, we get the amount of inputs entered by the frame amount, you should geat the total.

Summary
(putting it all together)

Human Thresholds equation (to find out how many inputs are pressed in a given period, f)

a=the amount of time it takes to press a button from rest.
2a=the amount of time it takes to press a button and return to your original position.
(2 comes from the assumption that it takes as long to release a button as it does to hit it. but the important thing is 2a- which is the time it takes to get back regardless)
f=Is the number of frames you are expected to enter the inputs in. (in ddds case 39 frames)
c=The amount of buttons on the control stick that you can hit comfortably with one hand position (most people say 6 or 7).
b=The amount of time it takes you to rotate the control stick from one cardinal direction to the next. (90 degrees.)( is usually related to a)
rd= is not at variable! this means rounded down to the nearest whole number and will usually have parenthesis around something to show which part is being rounded.

Step 1: c*((f/2a) rd), but only if 2a>c. If 2a<c the amount of buttons pressed =f (if this is true skip step 2)

Step 2: now use the equations x=f/2a-(f/2a rd), and y=c/2a.
If x>=y. add c to your total in step 1.
If x<y add c*[(f/2a- (f/2a rd))/((c/2a) rd)] to your total in step 1.

Step 3: Finally add ((1+((f-1)/b)) rd) to the other two (or one) totals to find the answer.


Example: How many buttons are pressed if you can hit a button or control-stick direction in 8 frames each, in an interval of 39 frames, if you can hit 7 buttons comfortably?
(14+ (.4375=.4375)so +7 so +5)=26 buttons.

Examle2:How fast do you have to hit the buttons to break out at 129%, pressing 7 buttons comfortably over 39 frame interval? (the amount you need to reach is 34 buttons). if a and b are hit at the same rates?
7*((39/2a) rd), if 2a>7. what happens when 2a=7 though? (35+(.5714-1) +4+6)=45. so we can say that yes, 2a>c.

now use 7*((39/2a) rd) + step 2 + ((1+((38)/a)) rd)= 34. we have to guess and check to account for the round offs. if a=5 (21 + 7 + 8)= 36 buttons. You need to be able to press the buttons a little faster than about once every 10 frames (5 to push and 5 to release). (I think the actual number is about 9.6 total). IMO this is about what you'd expect for the edge of human capacity. its pretty hard to skew those numbers evenly as well and hit them on the first frame possible ^^ but it IS possible for some people, which is the point, just very hard since any small mistake will cost you..

IMO the average result you would expect for most people though is that they will be able to break out until 50% in a tourney. people won't skew their buttons to hit on cascading frames, they won't be able to hit their first buttons on the 10th frame exactly before the pummel, they won't be able to mash all of the buttons at below 10 frames a button, especially the triggers etc...
However I would consider the max of what you can expect people to break out if they learn how to mash at their full potentials is to around 140%. Very hard to do, but as you can see it can make a big difference ^^



Some Resources and Tests

edit: here is a website that tests reaction time for fun ^^ http://www.humanbenchmark.com/tests/...time/index.php
please feel free to post your results here so i can get an idea of what to expect ^^

edit2:
This website tests how many times you can press the 0 button in a 10 second interval. (its in japanese though) ^^
http://www.hudson.co.jp/hde/vol006/omake/tr16/index.html
I suggest using your thumb to see how much since you will be using your thumb on the controller. to find out how many frames that you are able to hit a button in take your total amount and divide it by 10 this is the amount of inputs you hit in a second. take that number and divide it into 60. this will give you the amount of time in frames it takes to hit a button. (this number is 2a in the equations)

edit3: the video for Reflex breaking out at 129% is here->
http://www.youtube.com/watch?v=Fm8z4O24CvI
he breaks out near the end.

 

[Amazing Ampharos]

Reflex broke out of that pummel at 129% in the now famous video. Adding in the 33% King Dedede gets to do with his down throw before he has to pummel, that means that absolute minimum percentage that you should be caught in an infinite with Mario, Luigi, or Samus is 96%, and that assumes (probably incorrectly) that Reflex was at the absolute limit of ability in that video. Reaction time is a non-factor since the non-stale down throws give you plenty of time to start mashing (Reflex claimed to have predicted the grab in that video so it's the same case really). Your calculations of optimal human mashing speed can't be right just because we have direct video evidence of significantly faster mashing. Hitting a bunch of buttons at once isn't the optimal strategy which explains the error; rotating the control stick (starting with a diagonal input) while mashing whatever buttons you can that don't slow down your rotation (strumming your index finger along BAX would be amazing if you had the coordination, but if it slows you down on the control stick, it's not worth it) is the human optimal strategy. Also, if I recall (though it doesn't really matter), c-stick flicks are worth 16, not 8. I say it doesn't matter because the c-stick has to be returned to neutral before being input again so it's a lot slower to deal with even with double worth than the control stick or even the buttons.

There is one condition in which an infinite can be done lower, and that's if King Dedede does his 33% and then does his running chaingrab across the stage before going for the pummel. This actually makes the infinite percentage variable based on stage position, but when you remember that these down throws across the stage are already going to be pretty stale, it's probably not going to add that much more damage.

 

[Hive]

yes but in the video where reflex breaks out, the DDD also doesn't throw him until it is required to pummel, which changes the data for conditions A) thru D). if the ddd misses his pummel by 1/6th of a second or even 1/2th of a second (from the time before AND after the pummel before the next throw) it is not only possible but probable that he would have broken out. both cases being relatively small, and the conclusion imo being overall more plausible then the data being wrong. 8 frames to hit the buttons is not an underestimat imo, if you can hit the buttons over 5 times per second (and rotate the control stick about 4 times a second as stated in condition C) then please show me. but under the circumstances i think they are plausible for some of the better mashers in the community).

Edit: also i forgot before, but if the ddd did the standing infinite until 33% and then tried to dash and grab them again, usually the other person will stil be out of range due to the longer trajectory from the staled throw. its just... you usually don't see this bc normal cgs don't last this long without them either ending or without the ddd pummeling. however if i'm wrong please show me ^^ (though even if it is, the data will still show ddds infinite without room to dash)

 

[ColinJF]

It's really hard to "mess up" the pummel because buffering is in the game.

How long it takes you to press down the buttons doesn't actually matter for anything. It would matter if you pressed all the buttons down at the same time, let go of them all at the same time, etc. but if you did that it would only count as one input anyway. What you actually do is stagger the presses so that on any arbitrary frame, you are pressing one thing, and the higher ratio of those things that are C stick flicks, the better.

 

[Hive]

its actually pretty easy imo to miss small amounts of frames imo before and after the pummel before the next throw. 1/6th-1/2 of a second is a very understandable condition.

and also i did account for all the buttons being pressed at the same time (except in some circumstances the c-stick had different timings). if you pressed all the buttons together it wouldn't just count as 1 input as well btw, at least not for breaking out (for doing actual moves it does though. i think is where the confusion is)

edit: also its hard to buffer the pummel since you must press it directly after grabbing the person and there's not much time to transition (especially from a shield grab, which is most often used)... either way, pummeling at the first frame you are able to is still shown above.

 

[ColinJF]

If you can press multiple buttons per frame then the figure should be even higher than it would be just making the correction I described above.

 

[Hive]

no it wouldn't, because that should be what i accounted for to begin with :/

edit: also moving your thumb to c-stick potentially loses the b and z buttons. and you must flick it to get the 16 frames, so the net effect=0 to the conditions above, possibly less since it takes more to flick it and return to its initial position than to press a button.

 

[ColinJF]

I just conducted some research on grab breaks. Here are the results

- multiple buttons in the same frame only count as one button and take off 8 frames
- you can press a direction concurrently with a button to get an extra 8 frames off
- you cannot grab break in the first 18 frames, even if you run the counter down, unless you are being pummelled

Dedede gets approximately 33% (it's not exactly 33%) from a grab before he has to pummel. Each frame, the counter can be decreased by at most 8+8+1=17, so you can break out up to

(39*17 - 90)/1.7 - 33 = 304% before the first grab.

We have observed Reflex breaking out at 129%. That translates into a breaking rate of (((90+1.7*129)/39-1))/8 = 0.87 inputs per frame, which is obviously possible.

The results given in the original post of this topic are, unfortunately, very wrong.

 

[PKNintendo]

Who was Reflex using?

 

[ColinJF]

He was using Wario, but my initial tests suggest grab break rate mechanics are the same for all characters.

Here is a graph that tells you what percent (before the first grab) you can get out at for an arbitrary number of inputs per frame (note that this number has to be in [0, 2]). This includes the damage of the initial throws before he has to pummel.

[Edit: The first version of this graph had an error, but it's fixed now.]

 

[Hive]

I just conducted some research on grab breaks. Here are the results

- multiple buttons in the same frame only count as one button and take off 8 frames
- you can press a direction concurrently with a button to get an extra 8 frames off
- you cannot grab break in the first 18 frames, even if you run the counter down, unless you are being pummelled

Dedede gets approximately 33% (it's not exactly 33%) from a grab before he has to pummel. Each frame, the counter can be decreased by at most 8+8+1=17, so you can break out up to

(39*17 - 90)/1.7 - 33 = 304% before the first grab.

We have observed Reflex breaking out at 129%. That translates into a breaking rate of (((90+1.7*129)/39-1))/8 = 0.87 inputs per frame, which is obviously possible.

The results given in the original post of this topic are, unfortunately, very wrong.

first off, let me say that i appreciate your criticisms ^^ they can only help to find the truth about the infinite which was my reason for posting it here

however, i would like to point out a few things:
Our numbers are actually much more similare then they appear. obviously 304% is a lot higher than my numbers, but why is that?
Its not actually major differences in our calculations, its the fact that my entire focus is based around estimating and approximating human thresholds while your conclusions are focused around a robotic mashing threshold. Naturally mine will be much lower.

You do ring up one really good point though that only one input + a button can be pressed at one time. I did not know this, and actually had thought of it when it was first brought up, but trusted the available data to me at the time that would seem to suggest otherwise. thank you however for confirming this.
How will pressing the buttons in a cascading pattern differ from that of an all together pattern though? if it takes 8 frames to press a button the difference is this for 39 frames if you are hitting 7 buttons +the control stick, and the control stick in a direction every 8 frames:
my calculations used (7 buttons + c on first frame, c on 9th frame, 7 buttons + c on 17th frame, c on 25th frame, 7 buttons plus c on 33rd frame)
For your calculations this same button rate is (1 button +c on first frame, 1 button on second frame, third button on third frame, etc... no buttons on 8th frame, 1 button and c on 9th frame)
what is the overall difference then? the difference for 39 frames is 2 buttons. my calculations press 2 more buttons for a total difference of (16/1.7) a 10% difference.
so its not horribly off, and considering you used the same equation as i (90+1.7p) and the same amount of frames to overcome, is it any surprise?

Your error though is your calculation for 0.87. According to your own calculation (90+1.7*129)/8/38= 1.01 inputs per frame. this means that the person is able to press over 39 buttons + control stick directions in 39 frames. which is above the human threshold. my assumptions on human thresholds are very modest and using the same rates that i have used the entire time and applying them to your calculations the person would not be able to break out of the percent that you say reflex did. By changing the way we view the human threshold (putting it into a ratio out of two) it makes the illusion that 1.01 is a small amount. but it is actually very fast, and your results are actually very similar to mine. (a max difference for most equations by about 10%)

 

[ColinJF]

Its not actually major differences in our calculations, its the fact that my entire focus is based around estimating and approximating human thresholds while your conclusions are focused around a robotic mashing threshold. Naturally mine will be much lower.

Your entire focus is nonsense. How long it takes to press the buttons does not matter, and I described why in my very first post in this topic:

How long it takes you to press down the buttons doesn't actually matter for anything. It would matter if you pressed all the buttons down at the same time, let go of them all at the same time, etc. but if you did that it would only count as one input anyway. What you actually do is stagger the presses so that on any arbitrary frame, you are pressing one thing.

Your error though is your calculation for 0.87. According to your own calculation (90+1.7*129)/8/38= 1.01 inputs per frame.

I didn't calculate it wrong. You calculated it wrong! Your mistake is that 1/1+1=2, not 1/2. Divison has precedence over subtraction.

0.87 is the correct figure.

 

[Hive]

you misregard actual estimates on human thresholds though and make assumptions that aren't based on as much. mine is not a lame focus at all, it is to account for what you should expect people to do at a human threshold.

(I used 39-1 as (39-1), sorry, but that's bc the other doesn't really make sense). you used the 1 in the passing of time wrong. think about it. if you need to break out at 0% it should technically take 90, or 11.25 buttons over the course of 39 frames (or .29 buttons per frame to break out on the 39th frame). 11.25 inputs at -8 each=90
but according to your equation it is 0.16 inputs per frame. or 6.24 buttons. there is a discrepency bc according to you pressing 6.24 buttons which will take aways about 50 of the 90 will somehow overcome the 90.

edit: we're using the same data with slight variations! there shouldn't be any major differences. don't you see that?

edit2: the passage of time doesn't need to be there. (90+1.7p) is the amount you need to break out of. since
each input does 8 you divide the percent by 8 to find how many buttons you need to press total. the total amount of buttons divided into 39 frames or 38 frames = the amount if buttons you need to press per frame.

 

[ColinJF]

I'll post something later.

 

[Hive]

kwl ^^ anyways before i go to bed, ty for the criticisms btw i appreciate the effort. and i hope there's no ill will towards me

edit: this should be right, correct me if i'm wrong->
you need a total of 39 buttons total to break the amount (90+1.7(129)=309.3 regardless of frames. (breakout number/8 rounded up to the nearest 1 since you can't press half a button). In order to press 39 inputs in 39 frames you need to press 1 button per frame. so the total amount is 1 input per frame or more. which is beyond pressing the buttons and c-stick at a rate of once every 8 frames

 

[ColinJF]

I'm right in everything I've said.

0.16346153846154 inputs per frame, or 6.375 inputs is enough to avoid being throw after a pummel by Dedede at 0%. The total grab counter is 90, but if you press nothing, when he throws you, the counter will have lowered to 90-39=51 all by itself. The amount you need to take off is only 51. 6.375*8 = 51. So everything I've said is correct and there are no discrepancies.

And by the way, I've verified that this is how the grab counter works; 90 is not the amount you need to take off, it's the amount of time that needs to pass when you press nothing.

Consider this: If you press one input each frame for the first 9 frames of being grabbed, you will break on frame 19 (i.e. after 18 frames). Why? Because (90-18)/8 = 9. If you don't believe me, test it yourself.

In other words, my figure for Reflex (0.87 inputs per frame) and my graph are both correct.

 

[Hive]

The total grab counter is 90, but if you press nothing, when he throws you, the counter will have lowered to 90-39=51 all by itself.

ok so what you are saying is that if you need to break out at 0% each frame that goes by will reduce the amount you need to break out of?
so technically if you are at 0% and 90 frames go by, 1 and a half seconds you should break out automatically yes? without pressing any buttons.

edit: example at 0% (90+1.7p)=90. now if your equation is ((90/frames needed to break out) -1)/8
it follows that if you need to break out in 90 frames, the buttons you should press per frame to break out is exactly 0 and beyond that is negative. if you don't press anything though do you actually break out automatically? careful if you tested this in training (i'm not sure how you did) the cpu does have some amount of breaking out.

 

[ColinJF]

ok so what you are saying is that if you need to break out at 0% each frame that goes by will reduce the amount you need to break out of?
so technically if you are at 0% and 90 frames go by, 1 and a half seconds you should break out automatically yes? without pressing any buttons.

edit: example at 0% (90+1.7p)=90. now if your equation is ((90/frames needed to break out) -1)/8
it follows that if you need to break out in 90 frames, the buttons you should press per frame to break out is exactly 0 and beyond that is negative.

This is correct.

I tested these mechanics using the frame by frame hack.

 

[Hive]

ah i just tested this ^^ and you are correct i wish you would have told me this sooner lol, i had no idea you were using that conclusion- but your right, and me not knowing this is why i have been arguing til now lol^^

Edit: i'll try to use this data to fix the OP by sometime tomorrow, ok? ^^ its good to have a more definite answer.

let me see if i got this, kk? ^^ according to your equation then 309.3 frames as opposed to 90 frames to break out automatically right at 129%? if you are at 129% (309.3 to break out) after 39 frames the the amount you need to break out is 270.3 then, right? so the total amount of buttons you need to press is 34 (rounded up to the nearest button). i got that right, i hope?

 

[Magik0722]

Is there a link to this video?

 

[ColinJF]

let me see if i got this, kk? ^^ according to your equation then 309.3 frames as opposed to 90 frames to break out automatically right at 129%? if you are at 129% (309.3 to break out) after 39 frames the the amount you need to break out is 270.3 then, right? so the total amount of buttons you need to press is 34 (rounded up to the nearest button). i got that right, i hope?

This is correct, except it's "inputs", not "buttons". The distinction matters because you can press a button and a direction concurrently to get two inputs in one frame.

 

[Hive]

one more thing before i change the op, colin, if that's ok (for clarification, not bc i disagree with you ^^)->
i just realized while editing the op to your data-
0.87 inputs per frame is a bit harder than what i expected (or what i'll be using is 34 buttons per 39 frames which is needed)

say you press the buttons in a cascading example.
I've said earlier that the amount of time it takes a human to take on average to press a button is 8 frames, and 8 frames to release it again, before the process repeats.
I'll try to do this manually so as to clarify: 7 buttons with one hand position can be pressed comfortably on the controller.
so this means that with these requirements for human threshold this should occur
(1 button +1 c stick direction on the first frame, 1 button on the second frame, 1 button on the third frame, etc.., no buttons on the 8th frame, c-stick direction on the ninth frame (1st button is fully released at this point), 2nd button fully released on frame 10, etc...) (the main point being that it takes an equal amount of time to release the buttons as it does to press them)
all in all at 8 frames to press an input this will mean that after 39 frames only 26 buttons total will have been pressed. in order to actually get to the 34 buttons pressed which is the desired amount you would need to press the buttons in about 6 frames, and release in the same amount of time. All in all this means that you are able to press a single button or trigger on the controller about 6 times a second. Is this actually plausible, what are your thoughts?

however, i was thinking if the ddd missed a few frames in the pummel it would increase how quickly would be needed and account for that.

oh, and also i meant inputs in the above post lol sorry ^^ i make that mistake a lot lol

 

[ColinJF]

Are you considering that you are rotating the joystick while pressing these buttons?

As for whether it's humanly possible, seeing as Reflex breaks out at high damages all the time, I am convinced it is possible. Maybe 129% was a fluke, but he breaks out at over 110% in many videos. I can't do it myself, but it's clearly possible. Maybe just take 110% as the benchmark or something.

 

[Hive]

yes i have accounted for the c-stick being rotated, that is why after 7 frames 8 inputs have been added, and that the c-stick hits again on the 9th frame while the other buttons are being released. however, i am also assuming that the amount of time it takes to rotate the c-stick 90 degrees is equal to the same amount of time it takes to press a button down. being that the distance the control stick needs to be moved is larger than the distance needed to press a button i thought it was a fair assumption. I've seen the vids of reflex you are talking about ^^ what i am saying though is (since we do not know the actual frames of him breaking out) According to your data, which is more plausible?
that reflex can press a button about 6 times per second, or that the ddd missed some frames from his grab to his pummel, and from the end of his pummel to his grab?


ps magik the vid of reflex breaking out should be here-> http://www.youtube.com/watch?v=Fm8z4O24CvI
towards the end.

 

[ColinJF]

It cannot be inferred from the grab break mechanics data alone which is more plausible obviously, but considering Reflex's ability to do it consistently against top level opponents, I'm hesitant to attribute it to opponent error. Dedede's role in this is not difficult to do properly.

You can just discuss the possibilities in your opening post.

 

[Hive]

i agree. if it were just from the grab break data though i think it would be implausible due to how fast i've seen people able to hit buttons on tests online. however there may be something we are missing, i messed up the assumption, the ddd might have messed up by a small amount, or reflex is an incredibly talented button masher.
I'll post my thoughts on the human threshold in the op, but i'll try to specify they are my opinions as well.