Hitstun project (frame data people look at this)

[ColinJF]

Leafgreen and friends had previously conducted some research on hitstun, but it turned out to be wrong. Today, I enlisted HotGarbage to collect some preliminary data on hitstun for me to figure out the mechanics behind it. Here is the result:

The number of frames of hitstun is directly proportional to the magnitude of the force with which the character is launched and inversely proportional to the character's specific hitstun constant.

Based on this preliminary data I was able to obtain the specific hitstun constants for a few characters. I hope that by splitting up the work among the different people who can collect frame data we can quickly compute the value for every character. It shouldn't take more than a half hour to do a couple of characters. If you want to help, reply with what characters you are going to do (take several), then do the following procedure:

1. Start a versus game including Mr. Game & Watch and the target.
2. Set the target's handicap to 20% for the first trial.
3. Hit the target with Mr. Game & Watch's up smash.
4. Mash double jump as you fly away.
5. Quit the battle and write down the "max launch speed" from the results screen.
6. Analyse the footage on video and write down the number of frames between the end of hitlag and when you were able to double jump. Do not count the frame the double jump starts. This quantity is the hitstun.
7. Repeat 2-6 for 30% handicap, 40% handicap, 50% handicap, and 60% handicap.
8. Post all of these values (handicap, launch speed, hitstun) for the character.

Specific hitstun constants

Pit: 33
Meta Knight: 34
Fox: 36
Jigglypuff: 32
Sheik: 35

Bowser
Captain Falcon
Diddy Kong
Donkey Kong
Falco
Game & Watch
Ganondorf
Ice Climbers
Ike
King Dedede
Kirby
Link
Lucario
Lucas
Luigi
Mario
Marth
Ness
Olimar
Peach
Pikachu
Squirtle
Charizard
Ivysaur
R.O.B.
Samus
Snake
Sonic
Toon Link
Wario
Wolf
Yoshi
Zelda

These values look close together, but they are in fact different values. Rounding error cannot explain the difference in observations.

Of note: a higher specific hitstun constant means you enjoy less hitstun! Contrary to popular belief, it seems that it is the characters with high fall acceleration who enjoy less hitstun, which actually makes sense, since it compensates for their inherent susceptibility to combos.

If you're going to help, respond with which characters you are going to collect the data for. I figure we should be able to get this out of the way and have something to publish.

 

[Zankoku]

Is mashing accurate? Only so if you can buffer a double jump. In which case, can you do so simply by holding jump or up?

 

[ColinJF]

I'm not sure to be honest. I just had HotGarbage do it a couple of times to make sure he got the same value. I read Magus say that in Melee, you can buffer a double jump only in a small window directly before hitstun ends. If that's the case in Brawl as well, then mashing should be accurate 100% of the time, but you'd have to mash, not hold.

 

[leafgreen386]

And this is what explains why when we performed the test on pit that we got the equation:

h = l / 165

Which got turned into:

h = .4 * l / 66

When we found out that .4 is the constant used in the hitstun code (an added point of interest is that the number 1020 divided by this constant gives the point at which a weak hit becomes a strong hit). L is the launch speed of the move, and as we now know, 66 was... one... half... of... pit's character specific... division constant?

Now I'm totally confused lol

According to my data, characters should be suffering half of the hitstun that they are according to this. But seeing as how this is not the first time this has happened now (going back to steel2nd's data about wolf), I'm gonna lean to believe that the error is on my end. Gonna try to talk to kupo again and figure out exactly what he was doing to get such a huge amount of (yet consistent) error.

Either way, I'm glad I brought up the whole hitstun thing with you again, colin, so that we could uncover this.

As for buffering a double jump... isn't that covered by the game's universal buffering system?