Complete Hitbox/Frame Data For Every Character (Now with KO Percents)

[Big O]

Nicely done Thinkaman. The new format is a vast improvement.

I just wanted to quickly mention that you have attacks lasting one frame too long. If the remove hitbox command happens on frame 11 for instance, the last hitting frame is frame 10.

 

[Amazing Ampharos]

@ Kofu Your post was very helpful; I wouldn't have looked at G&W otherwise!

 

[Thinkaman]

Nicely done Thinkaman. The new format is a vast improvement.

I just wanted to quickly mention that you have attacks lasting one frame too long. If the remove hitbox command happens on frame 11 for instance, the last hitting frame is frame 10.

I was assuming it was understood that the latter number was the frame the hitbox disappeared on. I can see how the last active frame might be more intuitive. Updating now.

Edit:
There is a minor discrepancy in the data as currently represented, due to the 4 different ways repeating hitboxes are handled in the code.

1. They might be manually written out, like G&W nair. In these cases, the script is not intelligent enough to collapse them.
2. They might be a loop, like G&W bair. In these cases, the script shows the duration of a single iteration of the hitbox.
3. They might be a single hitbox with a refresh duration flag set, like Palutena uair. In these cases, the script shows the entire duration of the single repeating hitbox.
4. They might be looped via a sub-expression (which we cannot view), like man repeating jabs. In these cases, the script simply has no way of finding the data and it is missing.

 

[Big O]

I was assuming it was understood that the latter number was the frame the hitbox disappeared on. I can see how the last active frame might be more intuitive. Updating now.

I forgot to mention that the partial intangibility/super armor flags also work the same way, so you can also cut those down into one liner statements like super armor frames 1-10 or w/e instead of sentence 1: begins frame 1 sentence 2: ends frame 11.

 

[Thinkaman]

I forgot to mention that the partial intangibility/super armor flags also work the same way, so you can also cut those down into one liner statements like super armor frames 1-10 or w/e instead of sentence 1: begins frame 1 sentence 2: ends frame 11.

It would require some lookahead. It's easier to just have a dumb linear list of all "events", at least for now.

 

[Antonykun]

Huuuuge update. Totally updated the formatting, now accounts for all hitboxes.

Edit:
You'll note that I have taken the liberty of labeling all knockback variables in a newer, vastly more useful convention.

Every hitbox is denoted as either:
Base + Growth

-or-

Fixed + Weight

As we understand it, Base and Fixed knockback does not scale with any other factors except rage. We are not aware of any difference between the two. However, they need to be labeled and regarded differently, because they are stored in a different manner and use different scales. 100 Base knockback is huge, while 100 Fixed knockback is nothing.

Growth knockback scales with hitbox damage, enemy damage, enemy weight, and rage. I have taken the liberty of multiplying all Growth values by their hitboxes damage. There are exactly zero cases where this would not be the preferred way to consider and compare the data.

Weight Knockback scales with enemy weight and rage.

Edit: Updated again to hide redundant hitboxes.

Edit: Filtered out fake 3% hitboxes in throws. (I did a full search to ensure no real hitboxes were affected)

Edit: Added multi-hit loop data. Still missing anything locked in a sub-expression.

I really like how the list looks now. I know just need to really understand this.
Somethings to know for miiswordsman, I think unknown substate 0x227 is Slash Launcher they are both a single hitbox attack that deals 6% damage and it looks like SL sends people at a 55 degree angle too. unknown substate 0x214 might be Blurring Blade but im not too sure on that one

 

[Meru.]

Thanks a lot for doing these, it's incredibly helpful. I cannot imagine how much time it must have taken.

I would like to point out one small error though with Zelda's data:

1. Fair
2. Frame 9-13: 20% 24b/1900g 361° 1.5-Hitlag 0.0-SDI Magic
3. Frame 9-13: 4% 5b/384g 361°
4. Frame 9-13: 4% 5b/320g 361°
5. Enables transition to Fair landing state (landing lag) on real frame 4
6. Cancels transition to Fair landing state (landing lag) on real frame 50
7. 
   
8. Bair
9. Frame 6- 9: 20% 28b/1920g 361° 1.5-Hitlag 0.0-SDI Magic
10. Frame 6- 9: 4% 5b/384g 361°
11. Frame 6- 9: 4% 5b/320g 361°
12. Enables transition to Bair landing state (landing lag) on real frame 3
13. Cancels transition to Bair landing state (landing lag) on real frame 52

According to this, her sweetspot stays out as long as her sourspot, but this is incorrect. According to Mastercore, the sweetspot only connects on the first frame of the move (9 and 6 respectively). The hitbox gets removed after the first frame.

 

[icraq]

Working on breaking down Luma's things. Will edit as I continue

tico

unknown substate 0x8 jab 1
Frame 3- 4: 2% 100f/50w 70° 1.5-Hitlag

unknown substate 0x9 jab 2
Frame 5- 6: 2% 100f/50w 90° 1.5-Hitlag

unknown substate 0xa jab 3 upper cut
Frame 8-12: 4% 60b/600g 80° 1.5-Hitlag

unknown substate 0xb jab 3 spin
Frame 0- 1: 1.2% 15b/120g 50°

unknown substate 0xc jab 3 spin punch
Frame 6-10: 3% 60b/390g 80° 2.0-Hitlag

unknown substate 0xd dash attack
Frame 5- 7: 3% 45b/150g 45° 1.5-Hitlag

unknown substate 0xe nair (that means nair is a frame 2 move OOS!)
Frame 2- 5: 4% 40b/680g 361° 1.7-Hitlag
Frame 25-28: 3% 40b/510g 361° 1.7-Hitlag

unknown substate 0xf dair
Frame 10-13: 5% 60b/650g 80° 0.3-Hitlag

unknown substate 0x10 uair
Frame 9-14: 4% 30b/600g 28° 1.5-Hitlag

unknown substate 0x11 bair? (has two hitboxes?)
Frame 2- 9: 4% 110b/240g 92° 0.5-Hitlag
Frame 2- 9: 4% 110b/240g 108° 0.5-Hitlag

unknown substate 0x12 fair
Frame 7-11: 5% 90b/550g 45° 1.5-Hitlag

unknown substate 0x13 ftilt
Frame 7-10: 4.3% 40b/537g 361° 1.5-Hitlag

unknown substate 0x14 utilt (frame 2!)
Frame 2- 2: 8% 120b/480g 88° 2.0-Hitlag Ground-Target-Only
Frame 3- 4: 4% 85b/240g 98° 0.5-Hitlag
Frame 3- 9: 4% 80b/240g 93° 0.5-Hitlag
Frame 5- 9: 3% 80b/180g 90° 0.5-Hitlag

unknown substate 0x15 dtilt
Frame 7-12: 3.5% 35b/420g 30° 1.5-Hitlag

unknown substate 0x17 fsmash
Frame 16-17: 7% 43b/1155g 361° 1.7-Hitlag
Frame 16-17: 5% 43b/825g 70° 1.7-Hitlag (sour spot inner body hit)

unknown substate 0x1a usmash
Frame 9-11: 6% 60b/852g 85° 1.5-Hitlag
Frame 12-16: 5% 60b/710g 85° 1.5-Hitlag

unknown substate 0x1c dsmash
Frame 6- 7: 4% 40b/620g 33° 1.5-Hitlag
Frame 18-19: 5% 40b/775g 33° 1.5-Hitlag

unknown substate 0x5a ??
Frame 0- 1: 1% 50b/104g 52°

unknown substate 0x5b ??
Begins partial invincibility on real frame 0

unknown substate 0x5c ??
Frame 0- 1: 1% 70b/100g 45° 1.4-Hitlag

unknown substate 0x5f (partial invincibility on luma eh? only does 1% dmg eh?)
Frame 14-14: 1% 35b/35g 361°
Begins partial invincibility on real frame 0
End partial invincibility on real frame 13

unknown substate 0x62 ??
Frame 12-13: 2.8% 30b/140g 55° 0.5-Hitlag Magic

unknown substate 0x74 ??
Frame 5-12: 3% 55b/270g 55° 1.25-Hitlag
Frame 13-32: 5% 65b/450g 55° 1.5-Hitlag

 

[Thinkaman]

According to Mastercore, the sweetspot only connects on the first frame of the move (9 and 6 respectively). The hitbox gets removed after the first frame.

Fixed! Minor other updates and new special subaction labels as well.

 

[icraq]

hey thanks for updating the luma section! I don't know if you saw but I updated my list, unknown substate 0xd is Luma's dash attack. not sure how I missed that

 

[Lavani]

hey thanks for updating the luma section! I don't know if you saw but I updated my list, unknown substate 0xd is Luma's dash attack. not sure how I missed that

I already posted all of Luma's stuff you know, lol

 

[icraq]

I already posted all of Luma's stuff you know, lol

LOL damnit
i didn't think to open your quote box... alright well, i guess we double confirmed it!

 

[DanGR]

What is hitlag "magic?"

Rosalina's normal and low-angled fsmash have "1.2-hitlag magic", but up-angled fsmash only has "magic.".

Is that 1.2 just extra hitlag on top of normal hitlag?

 

[Lavani]

I think that's referring to the element of the attack.

Magic, fire, electric, etc.

 

[icraq]

What is hitlag "magic?"

Rosalina's normal and low-angled fsmash have "1.2-hitlag magic", but up-angled fsmash only has "magic.".

Is that 1.2 just extra hitlag on top of normal hitlag?

Most moves tell you their hitlag. Magic is just the element of the attack, it isn't related to hitlag at all, it's just formatted close together.

 

[DanGR]

Ah, so up-angled fsmash's hitlag just isn't shown, and 1.2 is the hitlag value of normal/low fsmash. Thank you.

 

[icraq]

Ah, so up-angled fsmash's hitlag just isn't shown, and 1.2 is the hitlag value of normal/low fsmash. Thank you.

yeah and I believe it's safe to say if the hitlag isn't mentioned, it's at 1.0.

 

[DanGR]

Looking through more of Rosalina's data...


D-air 80 degree angle seems strange. That's inconsistent with my experiences. May be worth looking into.

Confirming these are arm and head hitboxes:

Luma's Fsmash
Arm hitbox
Frame 16-17:    7%       43b/1155g 361° 1.7-Hitlag

Head hitbox
Frame 16-17:    5%        43b/825g  70° 1.7-Hitlag

Last but not least:

unknown substate 0xd
Frame  5- 7:    3%        45b/150g  45° 1.5-Hitlag

This seems to be Luma Dash attack. It's all in line with frame data of soloRosalina's DA, and behaves consistently with traits of Luma's dash attack, in my experience.

edit: ALSO, for anyone interested in viewing R&L's data from the pastebin file, I have it available in an easy to read format here: https://docs.google.com/spreadsheets/d/1q75HZookwyqWAWE_UsgPtvRIoiYXTAK2T9n2VkBM5do/edit?usp=sharing
I'll be updating it as more is known.

 

[Thinkaman]

Yeah, I only show the following values if they deviate from the default:
Tripping (0.0)
SDI (1.0)
Hitlag (1.0)
Element (0)
Aerial-Target-Only
Ground-Target-Only
Unblockable

Magic is an element, like Fire or Electric.

 

[icraq]

something to note is the tripping variable is definitely wrong, or something weird is up. I know rosalina's nair can trip but it's not flagged for tripping, think her ftilt can too. rosa's dsmash is listed as 0.3 for tripping but i've never seen that move trip.

also i was thinking, since luma's dair apparently has 0.3 hitlag... does that mean it's harder to DI? less time to react to it and such

 

[Amazing Ampharos]

I'm going to fal back on Brawl mechanics and just assume they're the same because I've never seen it be different from this. Random tripping is mostly about hitting into the ground weak enough so the move doesn't launch, and the trip rate on moves just increases the odds if that situation happens. If the situation happens, it always has at least a chance, and if it doesn't, it can never happen no matter what the trip rate is set to. Rosalina's dsmash I think always launches even 0% opponents (I dunno, maybe like 0% Bowser isn't?), but you'd probably see it trip a lot against a metal foe or something.

 

[Mr.Random]

Huuuuge update. Totally updated the formatting, now accounts for all hitboxes.

Edit:
You'll note that I have taken the liberty of labeling all knockback variables in a newer, vastly more useful convention.

Every hitbox is denoted as either:
Base + Growth

-or-

Fixed + Weight

As we understand it, Base and Fixed knockback does not scale with any other factors except rage. We are not aware of any difference between the two. However, they need to be labeled and regarded differently, because they are stored in a different manner and use different scales. 100 Base knockback is huge, while 100 Fixed knockback is nothing.

Growth knockback scales with hitbox damage, enemy damage, enemy weight, and rage. I have taken the liberty of multiplying all Growth values by their hitboxes damage. There are exactly zero cases where this would not be the preferred way to consider and compare the data.

Weight Knockback scales with enemy weight and rage.

Edit: Updated again to hide redundant hitboxes.

Edit: Filtered out fake 3% hitboxes in throws. (I did a full search to ensure no real hitboxes were affected)

Edit: Added multi-hit loop data. Still missing anything locked in a sub-expression.

The old format made it easier to compare buffs and nerfs. Now it's impossible. For example I was comparing Fox's Upsmash to its Brawl strength. In Brawl: 17% 30 BKB 96 KBG. In Smash 4 16% 30 BKB 94 KBG. I can't compare anymore

 

[Thinkaman]

The knockback formula has a small offset to knockback growth's scaling with damage, which varies very slightly based on enemy damage and weight. It is usually about 4. This is why my adjusted data previous seemed off for extremely low damage moves with high KO power.

I have updated the data to include the raw value, and the damage-adjusted value, which is now multiplied by (damage + 4). This should be a far more accurate model for comparing KO power.

 

[dragontamer]

Edit: Filtered out fake 3% hitboxes in throws. (I did a full search to ensure no real hitboxes were affected)

Those 3% hitboxes probably have something to do with the throwing animation.

In 2v2 or other multiplayer formats (1v1v1 or whatever), you can strike an opponent while throwing another opponent. It doesn't affect 1v1 play, but its something worth noting IMO.

 

[Lavani]

Actually, isn't that the damage that happens if you hit your opponent and force a grab break? You don't get off damage-free from that.

 

[dragontamer]

Actually, isn't that the damage that happens if you hit your opponent and force a grab break? You don't get off damage-free from that.

Why would that be part of the "throw" data?

I'd imagine that'd be part of the "grab" data, but not when things are broken out into "up-throw", vs "down throw"

 

[Lavani]

Why would that be part of the "throw" data?

I'd imagine that'd be part of the "grab" data, but not when things are broken out into "up-throw", vs "down throw"

I have no idea, but those breaks do 3% so it seems likely to me considering it's consistent across everyone.

 

[Mr.Random]

The knockback formula has a small offset to knockback growth's scaling with damage, which varies very slightly based on enemy damage and weight. It is usually about 4. This is why my adjusted data previous seemed off for extremely low damage moves with high KO power.

I have updated the data to include the raw value, and the damage-adjusted value, which is now multiplied by (damage + 4). This should be a far more accurate model for comparing KO power.

So how do I compare moves kill power from previous games to this one?

 

[Thinkaman]

You can compare the raw values directly. There was an identical (or extremely close) offset in Brawl as well.

 

[Mr.Random]

You can compare the raw values directly. There was an identical (or extremely close) offset in Brawl as well.

Can you show me how? Compare Luigi's Up smash in Brawl to it in Smash 4

 

[Thinkaman]

Brawl ground Up-B: 19%, 70 base, 73 growth (1679 adjusted growth)
Brawl aerial Up-B: 14%, 58 base, 72 growth (1296 adjusted growth)

Smash 4 ground Up-B: 25%, 40 base, 77 growth (2233 adjusted growth)
Smash 4 aerial Up-B: 20%, 40 base, 66 growth (1584 adjusted growth)

Note that in Brawl, this move hit on frame 6. In Smash 4, it is frame 8.

 

[dragontamer]

I suddenly realized that I forgot to "like" this post.

Great job on the scripts and everything man. Parsers are a b**** to program, but it looks like whatever program you made is doing the job well.

 

[Waymas]

Thanks for all this information man!

 

[ChronoPenguin]

Is WFT's grab actually 6, or multiple characters for that matter
It seems when you press your standing grab on a character they will start to enter the
-ive been grabbed- animation on frame 3, and then on frame 6 they are fully in her arm. I assume from frame 3 you can not actually do anything unless you had a lingering hitbox that will knock you out of the grab.

I kept looking going frame by frame,, but when I checked
Is what I see as frame 3. Megaman is being affected

is frame 4 and its clear now that I've got him.


Our Frame 6


Ill admit to being ignorant.

Why is Frame 6 "the grab" but not Frame 3/4 or 5?

I do not believe WFT is the only one like this.

Now I ask because if Im looking for my quick options to utilize. Should I be taking WFT's grab as a frame 3 or 4 option? Or Frame 6?

 

[Thinkaman]

Where are you getting these numbers/footage? Standing grabs in smash bros have generally been frame 6 (or 7) since Melee. It's very easy to compare them to other faster attacks by doing both at the same time. Jabs beating grabs from a neutral state happens all the time in gameplay.

 

[ChronoPenguin]

I'm taking them myself.
Since grab armor doesn't exist Its expected for the jab to win since grabs don't clang.

I took them myself and it's fine if you are critical of them. The root of what I want to really know is if in this situation if the initial image as I labeled frame 3 acts as the grab. If at that point they can't enter any new commands and commands in process are interrupted.

Partially out of curiosity if this particular grab will beat out some rolls if used at the same time (despite the unlikelihood of the circumstance) and to confirm my process.

What appears to be frame 6 had a notable jump from frame 5 which I neglected to upload.
this being it.

As a reference I used Shayas shield data and getting the expected 18 frames, leading me to believe I am not 3 frames early. Though I can put what I presume to be Frames 1 and 2 as well.
1.

2.

Frame 6 seems to correlate to Mastercore2 and your pastebin didn't have any ammendments to conflict with what I could read off Mc2 other then the transition state at Frame 8, so everything seems good except for Megaman going through animation change as early as frame 3 and it being clearer at frame 4 and 5.

 

[ZHMT]

Purin:

Rest

Hitbox out frame 2-4

I am reading this right? So this is the first game where rest has a active hitbox over one frame?

Also, are hitlag modifiers ignored by shields in this game as well? I mean it makes sense if that's generally a reason Marth is better than Lucina, his 0.75x hitlag multiplier.

Would this make moves with a hitlag multiplier higher than 1 less safe than usual?

 

[B!squick]

So, which of Bowser's Fire Breaths is which and why do they hit on frame 0-6 or 0-14?

 

[Kofu]

So, which of Bowser's Fire Breaths is which and why do they hit on frame 0-6 or 0-14?

The first is Fire Breath, the second is Fireball (or w/e it's called, don't remember off the top of my head), and the third is Fire Roar. I believe the hitboxes listed are the active frames of the flames themselves and do not include the startup animation that actually involves Bowser. You can see how this works by looking at Villager's (murabito) FSmash. On his main frame data it says that the charge release takes 20 frames. Under the data for his bowling ball it says the hitbox appears on frame 5. You have to add those two together, meaning the hitbox functionally appears on frame 25. Unfortunately I don't know where to find the frames for Bowser's various Fire Breath startups.

 

[B!squick]

I see. Thanks for the info. Sorry if that was answered already.