Data Zelda: Hitboxes and Frame Data [3.6]

[Vitriform]

Woooooaaaaah, what? I'm very confused now.

So there's a bunch of different scenarios. To make this as clear as possible, what are the frames for each of these scenarios:

1) Ground->ground, with explosion
2) Ground->ground, cancelled
3) Ground->air, explosion
4) Ground->air, cancelled
5) Ground->air, landing during animation (explosion or no explosion different as well?)
6) Air->air, explosion
7) Air->air, cancelled
8) Air->air, landing during animation (explosion or no explosion different as well?)
9) Air->ground, explosion
10) Air->ground, cancelled

These might not all be different, but I'm trying to be as thorough as possible. Thanks!

Format:
Ex) startup -> endlag or components thereof

Frame data:
1) 24 -> 22
2) 24 -> 12
3) and 4) 24 -> 61 frames of endlag -> special fall -> 4 frames of landing lag
5) 24 -> 30 frames of landing lag, not including time spent falling towards the ground while in the ending animation
6) and 7) 32 -> 61 frames of endlag -> special fall -> 4 frames of landing lag
8) 32 -> 30 frames of landing lag, not including time spent falling toward the ground while in the ending animation
9) 32 -> 30
10) 32 -> 30 (no difference!)

In words:
All grounded teleports have 24 frames of startup. All aerial teleports have 32 frames of startup.
All teleports that end in the air, cancelled or not, will have 61 frames of endlag. Landing during the animation will result in 30 frames of landing lag. Landing after the animation (i.e., out of special fall) will give 4 frames of landing lag, the same as a normal impact landing.
Teleports cancelled from ground to ground (platforms included) have 12 frames of endlag.
Ground to ground teleports that are not cancelled will have 22 frames of endlag.
Teleports that land on the ground from the air (cancelled or not) will have 30 frames of endlag.

Someone please correct me if I'm wrong about any of these. I'm not 100% on all of my descriptions, but I think I'm interpreting the frame data correctly. I'll do some testing in debug mode later to validate some of my descriptions.

 

[Sartron]

Format:
Ex) startup -> endlag or components thereof

Frame data:
1) 24 -> 22
2) 24 -> 12
3) and 4) 24 -> 61 frames of endlag -> special fall -> 4 frames of landing lag
5) 24 -> 30 frames of landing lag, not including time spent falling towards the ground while in the ending animation
6) and 7) 32 -> 61 frames of endlag -> special fall -> 4 frames of landing lag
8) 32 -> 30 frames of landing lag, not including time spent falling toward the ground while in the ending animation
9) 32 -> 30
10) 32 -> 30 (no difference!)

In words:
All grounded teleports have 24 frames of startup. All aerial teleports have 32 frames of startup.
All teleports that end in the air, cancelled or not, will have 61 frames of endlag. Landing during the animation will result in 30 frames of landing lag. Landing after the animation (i.e., out of special fall) will give 4 frames of landing lag, the same as a normal impact landing.
Teleports cancelled from ground to ground (platforms included) have 12 frames of endlag.
Ground to ground teleports that are not cancelled will have 22 frames of endlag.
Teleports that land on the ground from the air (cancelled or not) will have 30 frames of endlag.

Someone please correct me if I'm wrong about any of these. I'm not 100% on all of my descriptions, but I think I'm interpreting the frame data correctly. I'll do some testing in debug mode later to validate some of my descriptions.

This is correct, but its 21 frames of endlag on the grounded explosion, not 22. Also air->ground is the 21/12, not 30. Also, I'll add onto the thread about the 4 frames of land lag. Never thought of adding that but it sounds useful.

 

[Vitriform]

This is correct, but its 21 frames of endlag on the grounded explosion, not 22. Also air->ground is the 21/12, not 30. Also, I'll add onto the thread about the 4 frames of land lag. Never thought of adding that but it sounds useful.

Okay, that makes sense. The air -> ground ones in particular are the ones that I wasn't exactly sure about, so it's nice to get some clarification there.

EDIT: @ Sartron , you say air -> ground should have 21/12 frames (no shorten/shorten) of endlag, but in my testing in debug mode, it seems to be 21 frames of endlag regardless of whether the teleport is shortened or not. Is this intentional behavior? If it is (and I assume it is), then you might want to edit the original post to reflect that.

 

[4tlas]

There still seems to be some confusion. What I'm hearing so far is all grounded startup is 24 frames and all aerial startup is 32 frames. Ground->ground endlag is 21/12 when not cancelled/cancelled. Ending a teleport in the air has 61 frames of endlag.

Other than that, everything is still confusing.

 

[Sartron]

EDIT: @ Sartron , you say air -> ground should have 21/12 frames (no shorten/shorten) of endlag, but in my testing in debug mode, it seems to be 21 frames of endlag regardless of whether the teleport is shortened or not. Is this intentional behavior? If it is (and I assume it is), then you might want to edit the original post to reflect that.

I was not aware of a certain quirk with air->ground up special cancel, there's an animation I didn't know that's being triggered. I've updated the thread to account for this. All teleporting scenarios should now be accounted for.

There still seems to be some confusion. What I'm hearing so far is all grounded startup is 24 frames and all aerial startup is 32 frames. Ground->ground endlag is 21/12 when not cancelled/cancelled. Ending a teleport in the air has 61 frames of endlag.

Other than that, everything is still confusing.

Ending a teleport in the air puts you in an animation which lasts 61 frames. Landing during this animation gives you 30 frames of landing lag. Landing after this 61 frame animation gives you 4 frames of landing lag.

Ending a teleport on the ground without a cancel gives you 21 frames of lag. Ending a teleport on the ground from the ground with a cancel gives you 12 frames of lag. Teleporting from the air to the ground and canceling it gives you 23 frames of lag.

That's all.
To even further simplify it:

Air = 30/4
Ground = 21/12/23

case closed

 

[Magus420]

When you cancel there's a 2 frame transition where she's still invisible/intangible before going into the ending which allows snapping onto floors and getting the grounded ending. Might make more sense by understanding each of the components of the move rather than remembering the combined frame totals of various situations. There is Startup -> Travel -> Cancel Transition (if canceled) -> Ending -> Landing Lag (if you went into Ending Air).

Startup
Ground - Hit: 6-7, Total: 24
Air - Hit: 7-9, Total: 32

Travel
Total: 19, Invisible/Intangible: 1-19

Cancel Transition (if Cancel)
Total: 2, Invisible/Intangible: 1-2

Ending
Ground (ends into Wait) - Hit: 2-3 (if No Cancel), Total: 21 | 10 (if Startup Ground+Cancel)
Air (ends into Special Fall)- Hit: 2-3 (if No Cancel), Total: 61

Landing Lag (if Ending Air)
During Ending Air: 30
During Special Fall: 4

 

[4tlas]

So if you go absolute max distance AND cancel, cancelling is slower than not (air to ground)? That seems...unintuitive.

 

[Jamwa]

what is the best direction to quarter-circle sdi usmash? and which direction to asdi? (pls quote me so i get the notification, thanks for any help)

 

[WhiteCrow]

what is the best direction to quarter-circle sdi usmash? and which direction to asdi? (pls quote me so i get the notification, thanks for any help)

If you are a fast faller below 50% you can SDI down and out so you can crouch cancel and punish between hits. If you are a fast faller above 50% you should SDI up and out or just focus on DIing to make lightning kicks hard to space or straight up so we can only up air or nair.

If you are a floaty or mid weight I think SDIing up and out will almost always be the best option. Don't forget to mash fair if you're at lower percent so you can clip Zelda if she jumps early. Smaller characters (Olimar, Kirby, Jigglypuff) can SDI down and out below 50% and CC much like how space animals can.

 

[Boiko]

If you are a fast faller below 50% you can SDI down and out so you can crouch cancel and punish between hits. If you are a fast faller above 50% you should SDI up and out or just focus on DIing to make lightning kicks hard to space or straight up so we can only up air or nair.

If you are a floaty or mid weight I think SDIing up and out will almost always be the best option. Don't forget to mash fair if you're at lower percent so you can clip Zelda if she jumps early. Smaller characters (Olimar, Kirby, Jigglypuff) can SDI down and out below 50% and CC much like how space animals can.

Percentage doesn't matter in instances like this. Each hit has 0 KBG so the knockback will always be the same regardless of the opponent's damage. You can only SDI during hit lag frames which are calculated using the formula:

floor(c * floor(e * floor(3+d/3)))
Where
c= 2/3 if CCing, otherwise 1
e= 1.5 if electric, otherwise 1 (always 1 for attacker)
d= damage

So each hit of Zelda's up smash, doing 1 percent (aside from the final hit), has 4 frames of hit lag, regardless of the damage that the opponent has, and the knockback will never change, so you won't go higher/lower depending on your damage.

 

[WhiteCrow]

good information

I was only bringing up percent in regards to SDI>CC since that's one of the best ways to escape the move between hits. Thanks for the info though, I didn't know some of that.

 

[SuperUltraHyperOrochiPika]

Can you add the Weakest and strongest hitbox images on Zelda's timed Din's explosions. Or if someone has an image can they post it?

 

[Mentor]

Sartron I've gathered a huge amount of frame data relating to 3.6 dins & am planning on making a post/guide describing all the frame data & mechanics for each state of dins. Since this is the frame data thread, I'm not quite sure how the frame data I've gathered should coincide w/ this thread or if my frame data should be imported to this thread. Can anyone give some guidance on this? I don't mind the outcome as long as all the data is posted somewhere.

 

[Sartron]

Sartron I've gathered a huge amount of frame data relating to 3.6 dins & am planning on making a post/guide describing all the frame data & mechanics for each state of dins. Since this is the frame data thread, I'm not quite sure how the frame data I've gathered should coincide w/ this thread or if my frame data should be imported to this thread. Can anyone give some guidance on this? I don't mind the outcome as long as all the data is posted somewhere.

It's fine to make your own thread. You have much more in-depth information that would go past mine, and I'm sure you would appreciate having control over your own content in its own thread.

 

[BO~AT]

I was looking at the dtilt frame data and it claims that it has shorter IASA frames (30->25) if the previous move was another dtilt.

I went into debug and didn't seem to find this to be the case. Is this an error or am I just bad at this?

 

[Sartron]

I was looking at the dtilt frame data and it claims that it has shorter IASA frames (30->25) if the previous move was another dtilt.

I went into debug and didn't seem to find this to be the case. Is this an error or am I just bad at this?

I actually had it written backwards, IASA should be triggered on frame 30 if the previous move was down tilt, not 25.

 

[Miryafa]

How do you get -5 shield advantage for jab? The way I read it:
Duration: 29
Active hitbox: 15
So the lag is 29-15=14
Shield stun: 3
So 14-3=11 frames where the opponent can move and Zelda can't, which sounds like -11 shield advantage.

 

[Kaeldiar]

How do you get -5 shield advantage for jab? The way I read it:
Duration: 29
Active hitbox: 15
So the lag is 29-15=14
Shield stun: 3
So 14-3=11 frames where the opponent can move and Zelda can't, which sounds like -11 shield advantage.

IASA is frame 24, though, which means you can act on frame 24.

1-15) Jab animation
16-18) Enemy is in shieldstun
19) Enemy can act
24) Zelda can act

 

[Miryafa]

That explains the difference. Thank you.

Why is it that some attacks don't show IASA? E.g. Fsmash. Surely you don't wait for the end of the animation with that attack, because it shows shield advantage -8, but 24-39 = -15 != -8.

 

[Kaeldiar]

That explains the difference. Thank you.

Why is it that some attacks don't show IASA? E.g. Fsmash. Surely you don't wait for the end of the animation with that attack, because it shows shield advantage -8, but 24-39 = -15 != -8.

Not every move has IASA frames. You're also forgetting the 7 frames of shieldstun caused by f-smash's last hit

 

[Miryafa]

Not every move has IASA frames. You're also forgetting the 7 frames of shieldstun caused by f-smash's last hit

Aha, of course. Thank you!

 

[Kaeldiar]

As I'm going over this, I see an error, likely due to the way things are calculated. Nayru's is -4 on block when land canceled. It has effectively 7 frames of endlag, because you have to account for the 3 frames (25-27) that she is in the air before she lands. Then she goes through 4 frames of landlag (28-31), and she is actionable on frame (32). The opponent is in shieldstun for 3 frames (25-27), and is actionable on frame (28). #Math #ItCantBeTHATGoodOnBlock

 

[Kaeldiar]

As I'm going over this, I see an error, likely due to the way things are calculated. Nayru's is -4 on block when land canceled. It has effectively 7 frames of endlag, because you have to account for the 3 frames (25-27) that she is in the air before she lands. Then she goes through 4 frames of landlag (28-31), and she is actionable on frame (32). The opponent is in shieldstun for 3 frames (25-27), and is actionable on frame (28). #Math #ItCantBeTHATGoodOnBlock

Sartron

 

[zeldaplaya666]

Something I've noticed with up b is that you can act sooner if you do the inputs after waiting a little bit as opposed to mashing them. For instance if you do a canceled teleport where you mash Nayru's and a cancled teleport where you don't but only press b once when it is first available, when you press b once you act significantly faster. while I only used cancled teleport as an example it does the same thing with all variations of teleport ending on ground. Does anyone know why this is?

 

[luxingo]

That's strange. I can't think of a reason the game mechanics would work like that, and I also just tested it. Mashing Naryu's after a telecancel was just as fast as pressing B once after a telecancel for me.

My explanations for why it might seem like it's faster by pressing B once are:

• Mashing B might feel like it takes longer since you're doing more in the same amount of time, or
• Maybe you're mashing B too slowly and you don't press B on the first available frame.

 

[Kaeldiar]

Something I've noticed with up b is that you can act sooner if you do the inputs after waiting a little bit as opposed to mashing them. For instance if you do a canceled teleport where you mash Nayru's and a cancled teleport where you don't but only press b once when it is first available, when you press b once you act significantly faster. while I only used cancled teleport as an example it does the same thing with all variations of teleport ending on ground. Does anyone know why this is?

Lux is right. We can't mash at 30 fps, which would be "perfect" mashing, so when you mash, you don't actually usually the first possible frame. You end up hitting B later. That's why it's important to have good timing in Smash Bros. Understanding how long it takes for hitstun to wear off or for a move to end means that you can act as fast as possible.

 

[Miryafa]

I'm having am extremely hard time reading and understanding this all. I'm trying to parse dtilt right now.

I understand the move has 4 hitboxes (though not which hitbox goes with which number). They appear on the listed frames, deal the listed damage, and have the listed IASA frames.

Is there someone who understands it and could explain to me everything else?

So I hit my opponent with e.g. The toe on frame 8. Strongbad said (I think) that hitstun is 0.4 times knockback. So they take (30 + 0.55 * %) * 0.4 frames of hitstun, i.e. 12 at 0%. I have to wait 17 frames (til frame 25) to act out of it, so I'm -5 frame advantage on hit (and also on shield, because it has 6 frames of shieldstun). There's 6 frames of hitstun, but it doesn't matter to me because I'm waiting 17 frames to act anyway, so the 6 frames in this case just mean my opponent has 6 frames to SDI.

And if I wanted to follow up with dtilt, I'd be waiting another 5 frames to connect, (assuming my opponent was back on the ground), so I'd be -10 frame advantage to hit with the second dtilt.

 

[Kaeldiar]

I'm having am extremely hard time reading and understanding this all. I'm trying to parse dtilt right now.

Okay, so I finally took the time to figure out all this crap...D-tilt is a rather complicated move to break down, because it serves several purposes. It was even more complicated in 3.0, but in either case, here's what you need to know from a "how to use this in combat" standpoint:
1. meteor on grounded opponents, which makes it a great combo starter
2. toe sends aerial opponents up
3. hip is a meteor on grounded opponents
4. Using multiple d-tilts in a row is bad, because you can't act as fast

I understand the move has 4 hitboxes (though not which hitbox goes with which number). They appear on the listed frames, deal the listed damage, and have the listed IASA frames.

Is there someone who understands it and could explain to me everything else?

D-tilt has effectively 2 "hitbox states." The first being frames 5-7, and the second being frame 8.

Spoiler: Frame 5

Spoiler: Frame 6

Spoiler: Frame 7

Spoiler: Frame 8

I'm going to break this down frame by frame. On frames 5-7, there is effectively one, single long hitbox that stretches across the bottom. This hitbox has Hitbox ID 0, which means it has highest priority. That actually doesn't mean too much, though, because it only hits grounded characters, and the others are aerial-only. SO, if you hit a grounded character with d-tilt on frames 5-7, they get meteored by that hitbox. It is also important to note that this hitbox is very low to the ground, which makes it the best choice for shieldpoking.

Against aerial foes on frames 5-7, one of two things happens. They can get hit by the higher priority toe hitbox (ID 1) and get popped up, or they can get hit by the lower priority hip/thigh hitboxes (2/3) and get meteored.

Frame 8 works only slightly differently. The hitbox that normally was very low to the ground to shieldpoke grounded foes reaches out a bit more, for the sake of coverage. It is a much larger hitbox with much more range, and it covers the same space as the previous toe hitbox. Additionally, hitboxes 2/3 that previously only hit aerial foes will now ALSO hit grounded foes (yay! coverage!)

The big toe hitbox that sends aerial opponents up works the same on frame 8.

So I hit my opponent with e.g. The toe on frame 8. Strongbad said (I think) that hitstun is 0.4 times knockback. So they take (30 + 0.55 * %) * 0.4 frames of hitstun, i.e. 12 at 0%. I have to wait 17 frames (til frame 25) to act out of it, so I'm -5 frame advantage on hit (and also on shield, because it has 6 frames of shieldstun). There's 6 frames of hitstun, but it doesn't matter to me because I'm waiting 17 frames to act anyway, so the 6 frames in this case just mean my opponent has 6 frames to SDI.

Although the hitstun formula is Knockback*0.4, the knockback formula is much more complicated than that. I would recommend using Strong Bad's Knockback Calculator. Most importantly, what you did does not factor in the amount of damage the move does. At 0%, most characters would take 17 frames of hitstun, which makes it +1 on hit (assuming we hit on frame 8), as you are actionable on frame 25, whereas they are still in hitstun. Additionally, it would be -13 on block (because let's be real, how on earth are you going to whiff their shield for the first 3 frames of the attack?). Both of you suffer hitstun, then your opponent suffers 6 frames of endlag, they are actionable on frame 5+6+1 (12) and you are actionable on frame 25.

And if I wanted to follow up with dtilt, I'd be waiting another 5 frames to connect, (assuming my opponent was back on the ground), so I'd be -10 frame advantage to hit with the second dtilt.

-4, for reasons discussed.

AND NOW THE FUN PART

So you would never want to follow up d-tilt with another d-tilt. You always have a better option. D-smash, grab (to beat a buffered shield), and up-tilt are the BnB. F-smash and Fair are also possible, depending on circumstances.

D-tilt's lowest hitbox doesn't like to show up in debug mode

Use the va-jay-jay hitbox to meteor Marth when he tries to recover. At low%, you can CC > d-tilt meteor him.

D-tilt is a god at the ledge even if you don't get the meteor, because you can get things like d-tilt > run off > fair, or d-tilt > d-smash. It's also one of her lowest reaching moves

D-tilt meteors grounded opponents. This is significant, because it means they get "stuck" on the ground. It beats CC, and they have to go through that dumb "stuck in ground" animation, which means mucho follow-ups

 

[Miryafa]

I just learned it takes 4 frames to start dashing forward, but 2 frames to pivot dash.

Also that Zelda can only dash from standing/crouching (or pivot dash from dash/run). Any move or landing makes her walk. Important because it means you can't easily dash attack out of another attack.