True Weight / Survival

[JOE!]

LINK TO SPREADSHEETS​

This thread is archiving the % that each character can live to with (near) perfect DI when hit by the opposing character's "kill moves".

I used Final Destination, Battlefield and Smashville for testing, as they are the most common stages and have uniform terrain/ blastzones (for the most part). All moves have been tested in the center of these stages, with the exception of Throws, which are done at the edges, as they arent considered for kills unless done at an edge in the 1st place.

I used the training room in order to get accurate %, as well as 1/4th time in order to input the correct (smash) DI consistantly. Training mode also keeps moves at a "near" fresh state, only being 05% weaker, but this is more realistic for most characters as their "killers" will be used for other purposes such as damage racking or spacing.

I do not list moves that wil kill over 200%, as it's really kind of silly for an actual match between good players to have a character live that long.

This will be updated as I finish data, but if anyone wants to join it'd be much appreciated

NOTE: When editing the spreadsheets, I use Green to highlight the character's best killer vs that character, Red to highlight moves that'd normally kill, but dont past 200% or so on this character, and Yellow to highlight moves that have a note (like with Bowser's Dair being better vs Mario on SV rather than standard DI procedure).

1/39

Spoiler

DI Method:

1) DI toward top right/left corner.
2) Fair ASAP, then Fast Fall.
3) Double Jump, then Dair.
4) Hold forward to use his Air DI to recover.

​

COMPLETE

Spoiler

Fsmash F: 149%, 141%, 142%* (only on Smashville, using Dair ASAP makes you live 1% more than Fair + FF)

Fsmash U: 137%, 130%, 130%

Fsmash D: 153%, 147%, 147%

Dsmash F: 199%, 186%, 187%

Usmash: 158%, 158%, 159%

Bthrow: 183%, 177%, 198%

AVG: 163%, 157%, 161%

TOTAL: 160%

All other moves kill bowser over 200% on each of the stages, so it's not worth mentioning (aerials off-stage may as well be gimps).

Spoiler

Spoiler

Spoiler

Spoiler

Spoiler

Spoiler

Spoiler

Spoiler

:shiek:

Spoiler

Spoiler

Spoiler

Spoiler

Spoiler

Spoiler

Spoiler

Spoiler

Spoiler

Spoiler

Spoiler

Spoiler

Spoiler

Spoiler

Spoiler

Spoiler

Spoiler

Spoiler

Spoiler

Spoiler

Spoiler

Spoiler

Spoiler

Spoiler

Spoiler

Spoiler

Spoiler

Spoiler

Spoiler

Spoiler

 

[rPSIvysaur]

May I suggest making a google doc for this?

 

[JOE!]

I will, but I also need a thread

 

[Indigo Jeans]

DI Method:

1) DI toward top right/left corner.
2) Fair ASAP, then Fast Fall.
3) Double Jump, then Dair.
4) Hold forward to use his Air DI to recover.

​

FUUUUUUUUU-

Anyhow, nice stuff. I'm excited to see this finished, and I'll try to help.

But yeah, a doc would be nice.

 

[JOE!]

yeah, id need help definatley...

there's gonna be 39 x 39 = 1521 articles....

 

[Luxor]

Hence the need for a more organized way of doing this. I'd say PSA up one character, irrelevant who, that has X BKB and 0% damage on every move with variable, increasing, evenly spaced amounts of KBG. Do that for a few different BKB values, test it all out, and BAM! You have values that match up to numbers, and then you can just compare the test numbers to the PSA values for BKB/KBG of all these different character's kill moves. Veril's formulas are gonna help here.

Or something like this, anyway. Much more efficient and actually... viable. TBH you shouldn't have to test on every character either, just Lucario/Jiggs/Bowser should be enough to get the two extremes and the center. This project has some inexactness built into it to begin with, so doing the few things I said and "inexacting" it a bit more shouldn't be too much of an issue IMO.

 

[JOE!]

this isnt true weight as much as survival, also you cant test DI in PSA...which s what really lets you live long.

According to PSA would be stupidly fragile, seeing as you cant account for the bucket and his Air DI

Edit: added Doc to the OP, only has info for the 1st 4 characters, but you get the idea.

 

[Luxor]

*facepalm*

I meant create a PSA moveset that has moves of controlled KB values ideal for testing, used to test in-game. It's not even possible to test kill% in PSA by any stretch of the imagination. Tomorrow or sometime maybe I'll make a spreadsheet showing you what I mean with a post here as well.

 

[JOE!]

I get what your saying, but I just dont trust PSA to test DI specifics and the like...

Unfortunatley this leaves my grind-fest as a more accurate method IMO

 

[rPSIvysaur]

What he meant was to make an entire testing moveset that mimics everyone's kill moves. Most likely this would be best done for moves with the angle 169. Other moves I'd assume would have to be tested via Joe's method.

 

[Luxor]

We could play with angles in PSA too, but that would be more for later (darn extra variable ). And JOE, we don't have permission to access the spreadsheet
It might help if you shared it with the Smash Lab members.

 

[JOE!]

derp, link shoudl work now.


And how do the angles predict DI?

 

[rPSIvysaur]

Link doesn't work still. Make sure it is part of the Smash Lab folder.

Angles are the direction that the attack sends you in. If it's a semi-spike, you're going to DI differently than being U-tilt'd by Snake.

Most kill moves use 169. Based on that, we can create different standard (a lot of them use 100KBG, some odd BKB and use the damage as the determinate) kill percents. We can even place them on frame one to make it easier for us. There are a lot of different things we can do with this.

 

[JOE!]

adjusted the link again, all of the smashlab should be able to see it, also added character charts

 

[Thinkaman]

http://www.smashboards.com/showthread.php?t=156908

 

[JOE!]

Neat, did you account for DI there?

 

[Yink]

Do you still need some help JOE? I'd like to help out.

 

[JOE!]

Do you still need some help JOE? I'd like to help out.

Any help is greatly appreciated

atm Bowser is the test character, so pick a character (thats not Mario or Weegee) and use the chart to find which attacks to use, and his described DI method on FD, BF and SV.

 

[Luxor]

I'd recommed figuring out a more efficient method now, before you waste time on 42^2*5-7 kill moves * 5 trials... yeah. I'd recommend just testing Luke/Jiggs/Boozer as sandbags and you've got your spread. That's 14x easier and just as good.

 

[JOE!]

not exactly....

I mean, GW is Jiggs-tier in weight, but his air speed and bucket brake, as well as other DI like the key/etc could propell him past lucario....

Or we know Bowser is technically the heaviest, but DK has better air speed + momentum break, and DDD falls like a rock with a better recovery....

The point of this is to rank every character's survival to find out which characters are truly "heavy".

Speaking of GW...and DDD/Peach, are there any codes to get a certain hammer/Minion/Sports Equipment each time?

 

[Luxor]

Recovery honestly isn't even a factor in this. With halfway decent DI and a double jump, anybody can make it back onstage from a typical kill move; it's really only an issue against spikes. As far as bucket brakers go, test them if you must, but at least find some way to simplify this project because it's logistically impossible as it stands. The point is not to have exact kill% for every character; that would be ideal but overwhelming. Just get a representative spread and let users do a little mental interpolation.

 

[T-block]

I think there's a lot of worth to doing something like this, but this project seems to be overdoing it.

I started something like this on the PT boards (and I intend to resume it soon) http://www.smashboards.com/showthread.php?t=261281. It explains what I've done so far and my motivation behind it.

There's just too many variables for most of this stuff to be overly useful. Stage boundaries, stage positioning, move staleness, opponent's DI, etc. That's why I limited my trials to vertical KO moves at zero staleness, because the KO percents of these moves vary much less with opponent's DI and stage positioning. What are you gonna do about characters like Squirtle or Jigglypuff, who get a significant number of kills with a f-air far off stage? And the 200% assumption is silly... I've had plenty of matches see percentages higher than that. Just a few weeks ago I played a tournament match where my Diddy survived until 250% against a Pikachu, dying to u-tilt. Dedede and Snake get past 200% against my Diddy all the time. I've seen Squirtle's d-throw fatigued not kill Snake at 200% before, but d-throw is definitely a KO move.

So all in all, I think it's a good effort, but maybe the idea needs a bit of refinement?

 

[Yink]

Any help is greatly appreciated

atm Bowser is the test character, so pick a character (thats not Mario or Weegee) and use the chart to find which attacks to use, and his described DI method on FD, BF and SV.

Well obviously I'm going to do Ness first <3.

I'd recommed figuring out a more efficient method now, before you waste time on 42^2*5-7 kill moves * 5 trials... yeah. I'd recommend just testing Luke/Jiggs/Boozer as sandbags and you've got your spread. That's 14x easier and just as good.

Yeah um...I mean I'm willing to do the work like you say JOE, but Luxor does give a good point.

Do you want me to proceed this way (as shown in Luxor's post or, just do it the way you wish)?

 

[rPSIvysaur]

I have to agree with Luxor. Seeing as kill percents is an imprecise science, I believe that collecting further arbitrary data (which changes as you move across the stage), you would probably expedite the process of collecting this data.

 

[JOE!]

well, it's only ludicrous when you look at it as a small-scale thing...

it becomes much more managable when there's a group of people doing like 3 characters a week or such?

 

[Luxor]

JOE, running some quick calcs, even working at unrealistically fast speeds, this is a 400 man-hour project. 800+ is a much more conservative estimate, I'd say 1000-1200. Even if it only took 800 hours and every single last Smash Lab member, from the 'honorary' ones like Dant tothe less active ones worked on it, it would take over 20 hours of work each. When you could divide the entire scale by 14 and have a 60 man-hour project that's no less accurate in the long haul. And I didn't even account for different stages and locations being tested; that'd EASILY push it over 4000. Not gonna happen. This is volunteer work.

 

[JOE!]

allright, but how can we make a formula that also depicts DI?

as of now, I do this from center stage as that is the max % you can be hit by said move and live if they hit you towards the side of the stage you're on...can that also be accounted for?

 

[Veril]

I mean, GW is Jiggs-tier in weight

That is not even a little true...


DI allows for ~20-25° variance from the default without accounting for SDI and reversibility. So to formulate the effect of DI you need to look at hitbox flags (esp special collision), SDI modifier, and angle. Also knockback and damage values in relation to character since pre-tumble hits are not DIable... and whether the move has a discrete multihit... and... Its not possible, don't bother trying.

Sorry, even simple accounting for kbg, bkb, ∂kb, damage, character % and weight vs angle (just change the angle of a move ±20° to test) is extremely daunting as a task. Accounting for complex effects like SDI and flag stuff is not something you should even attempt if you're not aware of actual character weight values (and even then, its not worth it).

 

[Luxor]

Joe, I never meant use some arcane formulae to calculate kill%. What I'm saying is only test each character's kill moves on Bowser/Jiggs/Lucario. As Veril said, a formula-based approach is worthless/impossible and what's going on now is just madness.

 

[JOE!]

whats going on now is just bowser <.<

 

[T-block]

Don't take it personally man... we're just trying to save you time =P

 

[JOE!]

I know, just getting mixed messages on what should be done with this is all

 

[phi1ny3]

That is not even a little true...


DI allows for ~20-25° variance from the default without accounting for SDI and reversibility. So to formulate the effect of DI you need to look at hitbox flags (esp special collision), SDI modifier, and angle. Also knockback and damage values in relation to character since pre-tumble hits are not DIable... and whether the move has a discrete multihit... and... Its not possible, don't bother trying.

Sorry, even simple accounting for kbg, bkb, ∂kb, damage, character % and weight vs angle (just change the angle of a move ±20° to test) is extremely daunting as a task. Accounting for complex effects like SDI and flag stuff is not something you should even attempt if you're not aware of actual character weight values (and even then, its not worth it).

I asked about marth tipper fsmash to eliminate some of these more trivial things such as SDI for a possible idea of consistency, but if we're looking into every move, this won't work . Although this project will probably be veeeeeeeery time consuming. We should probably definitely just do jiggs/luc/bowser for now. This is a highly desirable project that I'm sure a lot of people would like to know about though, so maaaybe more work in the future.

 

[Yink]

^

Well because Bowser's done already, shall I take Jiggs or Lucario?

 

[JOE!]

bowser isnt "done", he'd "being done"

due to school/other johns, i only have data for him vs Mario atm, however in a few I'll post the other character "kill move" charts to test, and just repaste them onto a sheet for Lucario and Jiggs...

 

[JOE!]

Updated charts to have all charatcer's kill moves for Bowser, Lucario and Jiggs

So they should gather a decent spread....however in the future I may want to expand this to Snake (as he is known to live to ******** %'s), GW (testing the bucket) and MK (because we'd all like to know the Kill %s for him)