Lucas Kill Move %s, a quantitative measure of knockback

FightingGameGuy · Mar 16, 2008

Intro

This newer, better section of this guide is under construction. You can still find the old data below based on the approach Famitsu used, but the new data will be far more versatile.

Launch Speed Formulas [Under Construction - Non-charged smashes done]

These are the formulas the game uses to determine the speed at which Mario is launched by Lucas' moves. Launch speed correspondes proportionally to flight distance by the following formula:

FightingGameGuy said:
Distance Traveled = 1*(10^(-5))*((Launch Speed)^1.8432)

In the following formulas X is percent, after the damage of the move has been taken into consideration. All of these formulas are 0% stale moves.

Non-charged Smashes
FSmash: 35.466x + 2466.6
UpSmash: 42.864x + 2040.3
DSmash: 41.025x + 1764.5

Specials
NeutralB: 19.988x + 1859.8
SideB: 16.011x + 2018.1
DownB: 25.654x + 1812.9
UPB(2): 20.017x + 3260.1

Tilts
Ftilt: 29.871x + 1115.7 (R^2= 0.9967: it might be have some sort of sweetspot)
Uptilt: 24.958x + 1855.5

Aerials
Fair: 33.441x + 852.56
nair: 15.824x + 3025 (R^2 = 0.9985)
bair: Too sweetspot dependant to establish a correlation
upair: 35.378x + 929.55

Launch Speed to Distance Traveled Conversion

FightingGameGuy said:
Distance Traveled = 1*(10^(-5))*((Launch Speed)^1.8432)

Here are some tables giving Distance traveled (Note: these tables give distance traveled in any direction)

100%
Upsmash 101
Fsmash 92.4
Dsmash 88.3
UpB(2) 72.2
Nair 56.6
Upair 53.4
DownB 51.5
Uptilt 50.9
Fair 47.6
Ftilt 45.7
NeutralB 40.8
SideB 36.2

200%
Upsmash 263
Dsmash 235
Fsmash 217
Upair 157
Fair 140
UpB(2) 131
Ftilt 125
DownB 120
Uptilt 117
Nair 97.5
NeutralB 88
SideB 71.2

Analysis (Under Construction)

Up-Smash Is by far Lucas' strongest move.

DownB has impressive launch gain (highest of his B-moves) with percent. When you match that the move's nearly perfectly horizontal trajectory you have a great KO move at high percents.

Upair has low a low constant launch meaning you can combo with it a low percents. But it also has a high launch gain so you can also kill with it!

Nair's launch doesn't not scale much with percent but instead is starts high. So it is not in fact a good KO move.

Ftilt and Fair have strong launch gain so they're decent launchers at high percents.

FightingGameGuy's old Data said:
I was inspired by what I saw translated in the famitsu magazine/strategy guide/promo thing. I ran the same tests, using Mario at the default CPU level 3 seeing at what percent moves could kill Mario when he stands in the center of final destination. Note that DI (the CPU's DI seems to be almost totally negligible) and weight are not taken into consideration by these numbers!

Organized by lowest % for a kill
Upsmash 75
Dsmash 96
Fsmash 95 (not instant kill, instant kill at 97)
DownB 115 (not instant kill, instant kill at 147)
UpB(2) 120
Dthrow 141
NeutralB 147
Upthrow 151
DashA 151 (Sweetspot Dependant)
Upair 152
Fair 154
Bair 158 (near horizontal trajectory, non-instant kill, sweetspot dependant)
Uptilt 160
Bthrow 172
Ftilt 177
Nair 194
Fthrow 200
SideB 289
NeutralA combo 300-320ish
Dair 370 (last hit only, grounded foe, kills by ceiling)
Dtilt NEVER Kills

Organized by type
Smashes
Upsmash 75
Dsmash 96
Fsmash 95 (not instant kill, instant kill at 97)

Tilts
Uptilt 160
Ftilt 177
Dtilt NEVER Kills

Aerials
Upair 152
Fair 154
Bair 158
Nair 194 (last hit only)
Dair 370 (last hit only, grounded foe, kills by ceiling)

Other As
DashA 151
A combo 300-315

Specials
DownB 115 (not instant kill, instant kill at 147)
UpB(2) 120
NeutralB 147
SideB 289

Throws
Dthrow 141
Upthrow 151
Bthrow 172
Fthrow 200

Notes
Other Factors Weight, DI, and positioning all effect whether a move can kill. These %s are intended as a quantitative measure of a moves knockback not as a memorize-this-percent-for-when-you-can expect-a-kill.

Smashes%s come from using the C-stick. Non C-sticked smashes kill at slightly lower percents (1-4).

Non-instant kills If a move is noted to not instant kill, it means that Mario died by failing to come back, not by dying off the side of the sceen. Note though, that the CPU level 3 mario DIs and comes back horribly. Thus, the percent for killing with these will probably be much higher against actual people.

UpB(2)This refers to the last hit of PK Thunder2. I had started at the edge of the final destination, so only a few hits or none of the hits other than the last landed. The initial frame hit also seems to have identical strength.

Sweet Spots
DownB: centerish of sphere
DashA: as close as possible
Bair: not sure, but you can get the trajectory just above the horizontal, obviously that is best.

Throws
Stood right next to Mario for the grab. As a result I needed to re-do the Bthrow % as it had an unfair advantage over the strength of the Fthrow without this positioning factor being taken into consideration

Version Info
1.0 First Version, most moves you'd want to use to kill people
1.1 Corrected info on SideB, added some more aerials . Added into the Lucas Guide! Thanks WarMachine
1.2 Now includes every single move. Was missing some throws. Thanks Serpit.
1.3 Added a disclaimer in the intro about DI and weight.

Version Info
2.0 Entirely New Direction in progress - Launch Speed Formulas. Non charged smashes complete.
2.0-2.1 Mini-Updates adding in new data
2.1+ Working on the analysis section
2.2 Added Information linking Launch Speed to Distance Traveled

Bluerockworld · Mar 17, 2008

Yeah, the up smash is so slow, but it pays off

Serpit · Mar 17, 2008

Thansk for the list ^^ But there's something that bothers me...

Unlisted Moves:If a move was not listed it means it doesn't kill at sudden death. Working on the SideB

You didn't list D-Throw and F-Throws, but I'm 100% sure those kill as well.

GofG · Mar 17, 2008

I was under the impression that level 3 computers (and all computers) DI'd randomly, and poorly at that.

I will test these figures with an actual human DIing and see if they hold true.

FightingGameGuy · Mar 17, 2008

Serpit said:
Thansk for the list ^^ But there's something that bothers me...

You didn't list D-Throw and F-Throws, but I'm 100% sure those kill as well.

Arg, you're almost certainly right, at least for the F-throw. Sigh, time to do some more research and update again.

Cerqerati · Mar 17, 2008

Thanks but is the DI factored in there??

Raheelp · Mar 17, 2008

What is basics of DI?

If I get smashed to the left I jam my joystick to the right?

FightingGameGuy · Mar 17, 2008

GofG said:
I was under the impression that level 3 computers (and all computers) DI'd randomly, and poorly at that.

I will test these figures with an actual human DIing and see if they hold true.

They won't.

I never intended these to reflect DIing, just to reflect the relative strengths of the knockback of different moves. I got the idea from famitsu guide, and figured I'd leave as much as possible to "default" settings. The random DI would account for the minor variation I've seen I was hoping he simply didn't DI at all.

This confusion is too common. Should I alter the title more?

Raheelp said:
What is basics of DI?

If I get smashed to the left I jam my joystick to the right?

Not at all.
You jam the joystick perpendicularly to the path you are sent on. Influence anti-parallel to the direction you are launched has no influence. That means that if you are smashed left, you jam up.

There's actually some more info but that's the basics. See this incredible guide for more help. A Guide to DI, Smash DI, C-stick DI, Teching and Crouch Cancelling --- Updated

GofG · Mar 18, 2008

FightingGameGuy, I would advise you do the following:

Redo all of your numbers, but instead of KO percentages, record the percentage at which the move changes from Scream #1 to Scream #2.

When a move, with no DI, would hit a character a certain distance, they change from their "aahh!!" scream to a more "AAAAHHHHHH!!!" scream. This will be a much more accurate measurement (in melee, it was how we determined how much weaker moves were in PAL than NTSC.)

FightingGameGuy · Mar 18, 2008

GofG said:
FightingGameGuy, I would advise you do the following:

Redo all of your numbers, but instead of KO percentages, record the percentage at which the move changes from Scream #1 to Scream #2.

When a move, with no DI, would hit a character a certain distance, they change from their "aahh!!" scream to a more "AAAAHHHHHH!!!" scream. This will be a much more accurate measurement (in melee, it was how we determined how much weaker moves were in PAL than NTSC.)

Wow, that's an excellent idea. I guess I really should have emulated other work on these forums instead of the silly thing I saw translated in famitsu (I would have too had I realized such the screams can be used that way).

I will definitely carry this out, though I'm unfortunately a bit busy over the next couple of days.

GofG · Mar 18, 2008

Err... I just tested it myself, and it turns out... Characters don't scream in brawl. >.<

However, the audience has two distinct cheers. One is about 1/3 as long as the other one. This, unfortunately, only works for moves that have horizontal knockback.

FightingGameGuy · Mar 19, 2008

GofG said:
Err... I just tested it myself, and it turns out... Characters don't scream in brawl. >.<

However, the audience has two distinct cheers. One is about 1/3 as long as the other one. This, unfortunately, only works for moves that have horizontal knockback.

Yeah I just tried it too. They do have an "ouch" sound though, but they sa
it at the end of the knockback, so Mario started dying for whatever move I used before he would change that sound. If he has a second he'll probably never say it outside of an entirely closed off arena.

A new idea popped into my mind. How about using the battle statistics at the end of brawls? My idea would be to beat up a non-controlled human player a bit using Lucas weakest knockback move, his dtilt, then finish Mario off with the move I'm working on. Then I can glean from teh statistics the highest damage of the foe and the corresponding max launcher speed. With a few points, I should be able to use that to determine calculate an actual formula for launch speed of the move. The main problem with this approach is that it will probably quite a bit more time consumming : (.

/Edit:

PERFECT!

My first test
Fsmash Launcher Speed(%) = 2466.6+35.466x
R^2 =1 !!!!!!!!!!

Fsmash Distance(%) = 18.05 +0.9841x
R^2 = .9983

Note to those who don't know stats, R^2 tests degree to which the equation fits the data, R = 1 is perfect.

This data also shows that their is a direct correspondence between distance and launcher speed so it doesn't matter which we chose to describe the move with. I say we pick Launcher speed because the very slight discrepancy in distance is probably caused by all the dtilting.

So yeah, this does give super accurate results, the question will be whether its worth the time invested. I'm actually thinking maybe I should make a subject about this in the general discussion, and try to recruit people to work on it FOR ALL CHARACTERS with me (It would clearly be a herculean task to do all 40 characters on my own). I know I personally adored SuperDoodleMan's frame data for Melee and have always wanted something like this on launch data.

/Edit 2 Current Status

Smashes Done (all R^2=1 unless otherwise noted)

FSmash: 35.466x + 2466.6
UpSmash: 42.864x + 2040.3
DSmash: 41.025x + 1764.5

NESSBOUNDER · Mar 19, 2008

Wow, that's a great idea. Keep up the good work.

GofG · Mar 19, 2008

This is incredible.

FightingGameGuy · Mar 19, 2008

Thanks for all the kind words.

I've completed B-moves (all R^2=1 again)

NeutralB: 19.988x + 1859.8
SideB: 16.011x + 2018.1
DownB: 25.654x + 1812.9
UPB(2): 20.017x + 3260.1

A couple of excellent surprises here.

DownB - as I and many suspected, "surprising but not strong" isn't the best description. Of his B-moves, Lucas' DownB has by far the strongest launch gain with damage. Hence at high percent it will outstrip all his other B moves in terms of launching power. Coupled that with its great trajectory and you have a nice finisher at high percents.

NeutralB - I was shocked to find out how weak this move is in terms of sheer launching power. I bet however, this will form a trend -- the ceilings are closer than the walls so moves that knock your opponent up don't have to have as strong launch. As a test I plugged in my old kill percents and got: Launch Speed at kill by NeutralB at 147 = 4798.036
Launch Speed at kill by Upstamsh at 75 = 5083.644
I'll take the similarity of the speeds (especially given the crapiness of my old technique) to be indication that the NeutralB's launch power is indeed accurate.

UpB(2) - Interesting in that it so far has the largest constant knockback of all Lucas' moves, but its gain with percent is actually much lower than I had thought it would be.

On the Idea of Making this a Project for all Characters
I'll probably post in the general discussion after completing the Lucas data so that i can demonstrate why this project is worthwhile and hopefully convince some people to undertake it with me.

Trozz · Mar 19, 2008

I've accidentally killed people with down-b before. It's a good kill attack when the opponent has over 120+ damage.

Levitas · Mar 19, 2008

It's a great edgeguard. It's like lucario's Dair, but hit's in front of you like melee sheiks fair, and sends them almost purely horizontally. With the Bstick, you can do a min frame count magnet, and it's pretty fast. It's like having another smash in the air.

Edit: I'm talking about his magnet, in case it wasn't clear.

FightingGameGuy · Mar 20, 2008

Done with tilts and starting aerials now.

Check out the nice scaling on the ftilt and fair =D. I'd really love to compare them to those of other characters, but all in good time ; ).

GofG · Mar 20, 2008

This is something that is confusing me: Are you multiplying the unit by some exponent of ten? Those values, if they're feet, seem way off.

FightingGameGuy · Mar 21, 2008

GofG said:
This is something that is confusing me: Are you multiplying the unit by some exponent of ten? Those values, if they're feet, seem way off.

I decided to use launch speed as the basis of measuring knockback strength, NOT distance traveled.

I chose this statistic for one main reason: it is directly given in the match statistics. In contrast, only TOTAL distance traveled is given. Lucas has no move that racks up percent easily and has no knockback at all, so the knockback from his dtilt, although small, provided a systemic interference in using the TOTAL distance traveled as "distance traveled by move in question".

The conversion from Launch speed to Distance traveled is:
Distance Traveled = 1x10-5*(Launch Speed)^1.8432

(I calculated this by racking up damage with Fox's sideB, the only move in the game to my knowledge with absolutely no launch of its own, then getting the distance and launch speed statistics from his FSmash. R^2= 0.9997)

While I am 100% certain that using the Launch Speed data is wholly superior as it avoids the systemic flaw previously mentioned, I had mistakenly thought there was a proportional relationship between launch speed and distance traveled not a power relationship. Distance traveled is far more relevant for smashers, as it is what ultimately determines death. However, unfortunately the equations for distance traveled cannot be simplified at all as the exponent cannot be distributed across the sum of the constant launch speed and the scaling launch speed : (.

Levitas · Mar 21, 2008

FightingGameGuy, I think you meant Fox's neutral b, his laser.

This is the same process that was used for melee. It's reliable. The only thing that it doesn't show is the angle of launch, which also affects a move's killing power.

I'm thinking that if you wanted to do launch distance of whatever move, you could use fox lasers and x move on the home run contest for a decent effect.

GofG · Mar 21, 2008

Levitas, the problem is that Sandbag's weight is unknown, but very heavy. Also, the fact that some character's moves work differently on the sandbag (pikmin, samus's dsmash, etc), and also the need to calculate the exact distance from the center of the platform to the point where it begins counting distance...

I think the next goal should be to figure out, in whatever unit brawl uses, what the dimensions of BF/FD/SV are (BF/FD/SV all have the exact same dimensions, as Foxy proved a few days ago).

Levitas · Mar 21, 2008

Ok, if some moves work differently, that's out. However, calculating the distance of the actual platform is easy. The hard part is the grass between the platform and 0ft distance.

FightingGameGuy · Mar 21, 2008

Levitas said:
Ok, if some moves work differently, that's out. However, calculating the distance of the actual platform is easy. The hard part is the grass between the platform and 0ft distance.

Calculating the ceiling for stages should be very easy but I'm a bit more confused about calculating the horizontal dimensions. Don't we need to have at least one completely accurate trajectory/angle before we can decompose the launch vector into the proper component vectors and do the calculations? If we estimate the launch angle then our figure for the horizontal dimensions will also pretty much be an estimate too.

If you know another method, or an accurate way to find the trajectories other than the HomeRunBat contest please tell me.

One idea I'm cobbling together would be to measure the ceiling of FD, then get the dummy Mario to 999 and hope to find an move with a 45ish degree angle. The move should kill them off the top first. Hopefully then we should be able to use the total flight distance again as the hypotenuse and some basic trigonometry to get the exact angle. We could then use that move to determine the horizontal dimensions. I'm going to try this plan with Fox's b-throw.

Yeah, you're totally right, I meant fox's neutralB not his SideB.

Jihnsius · Mar 21, 2008

And idea to figure distance in feet or meters (horizontally) instead of generic vector units:

Find a character that has a certain 'technique' that has no animation (crouching, lying on the ground, something of the sort) and have them do it on the turf on the homerun stadium. Take a snapshot and measure the distance from head to toe in pixels, then measure the distance from one foot/meter marker to the next in pixels. Divide character size by marker distance to get a relative distance in feet/meters to actual distance. Do the same for each stage from a certain set distance (for example, two characters rolled all the way against the edges) and then multiply the first number by the second.

EDIT: Although the "actual" distance isn't really relevant, as all units in the math are determined by vector units anyway.

Hardcorenesser · Mar 22, 2008

MATH!!!!! NO!!!! IT BURNS!!!!!!! why would you put math? just put what kills em and at what percent i am not a guy that likes to go through all that trouble...

Lucas Kill Move %s, a quantitative measure of knockback

Smash Apprentice

Smash Apprentice

Smash Journeyman

Smash Champion

Smash Apprentice

Smash Cadet

Smash Rookie

Smash Apprentice

Smash Champion

Smash Apprentice

Smash Champion

Smash Apprentice

Smash Master

Smash Champion

Smash Apprentice

Smash Ace

the moon

Smash Apprentice

Smash Champion

Smash Apprentice

the moon

Smash Champion

the moon

Smash Apprentice

Smash Lord

Smash Journeyman

Information

Network