Data Kadano's perfect Marth class -- advanced frame data application

[Deleted member]

i wish i had reflexes like you young kids. heh heh heh

 

[Kadano]

What about on YS or DL? I figure you probably you won't land much faster (if at all) so you'll have about the same number of vulnerable frames, but idk how no-impact lands really work, like if there is a window of frames with varying speeds.

No-impact land occurs if your SCDQ’s vertical shift (Δy) is below a certain very low threshold while the SCDQ’s lowest corner collides with a platform. This most commonly happens when a character spins around during his rejump as the SCDQ alternates frequently between short and wide vertical extension here.
Here is an example:

I'm particularly interested in doing it on Yoshi's because of how far out the platforms stick. I can actually do it without DJing forward at all. Does the slant affect how you can or should airdodge at all?

Not for reverse ledgedashes.

A completely flat stage wouldn’t make a difference here.

Can you do the same sort of movement as the no-impact land, but get a fair hitbox out before your invincibility runs out instead?

Easily. Invincibility stays out longer than the hitbox.

Is it harder or easier to no-impact land from Marth's DJ back as opposed to his DJ forward or straight up?

That makes no difference at all.

Also, completely unrelated, can you DJ backwards when ledgestalling and still grab the ledge as fast as possible?

Yes, you can. It’s strangely even possible to dj back but move forward (without using previous momentum). That’s because JumpAerialF and JumpAerialB take their Δx directly from the control stick’s X value on the frame of the jump input. X values in the range of 156-255 (sometimes this range is a bit larger) trigger JAB (assuming facing left). For 156, the actual Δx induced is very little; for 217¹ it’s many times as much.

¹My controller’s octagonal shape prevented me from achieving higher values. Yes, 255 produces more Δx than 217, but the difference is minimal and probably not worth carving out the shape. I’d rather carve out corners for mod90±17° DI to snap into.

 

[Bones0]

I've always wondered if DJs had varying degrees of momentum based on which direction you were holding. Thanks.

 

[TheCrimsonBlur]

This isn't exactly Marth specific, but Kadano is my dude, so I thought I'd ask it here:

Is there any detailed documentation on how getting hit by attacks while grabbed / held work? How is the mash formula checked [my gut tells me its on every frame of no hitstun. Is this true?], why do some moves knock you out and not others, and is the hitstun calculation the same as normal?

I feel like I've read about this years ago, but I can't remember where and my searching is fruitless. Someone asked me a while ago and I couldn't answer satisfactorily. :/

 

[Kadano]

This isn't exactly Marth specific, but Kadano is my dude, so I thought I'd ask it here:

Is there any detailed documentation on how getting hit by attacks while grabbed / held work? How is the mash formula checked [my gut tells me its on every frame of no hitstun. Is this true?], why do some moves knock you out and not others, and is the hitstun calculation the same as normal?

Finding this out would require huge amounts of work for me, so I first searched for Magus420 posts about this. And tadaa, I found one:

When you hit someone being held in a grab it does 1/2 damage. If the hit then does less than 6.00 damage it won't break the grab. The knee does 18% fresh, so it'd need to be staled to 0.66x or weaker to do less than 6 damage after 0.5x grab multiplier and not break the grab.

 

[TheCrimsonBlur]

Finding this out would require huge amounts of work for me, so I first searched for Magus420 posts about this. And tadaa, I found one:

Good stuff. Some of that makes sense intuitively, I guess. I do utilt on a held opponent in 2v1 a lot (late fair too), since I can react if they are hit out, and it covers both the air and ground release. Utilt does like 8-12 damage typically iirc (the lowest hitbox is usually...9? it takes an ultra tipper to hit 12) so it should always fall below the 6.00 threshold.

It usually keeps them in the grab 1-2 times and then pops them out though. Which seems to contradict Magus' statement...why is it that when I do a move from the same positioning, even more stale than before, they pop out? Did they mash out of the grab? It doesn't feel like it, because I don't see the grab release animation, but I guess I could just be misreading the first few frames of the animation.

I'd be interested in knowing if Marth has any true wobble in 2v1 scenarios. Doubt it. Dtilt does like 10 fresh so maybe that but your teammate would have to time the pummel perfectly. idk

----

I'm pretty sure how the mash formula is checked is completely uncharted territory though. That seems very difficult to test.

----

edit: utilt diagram puts me to shame. I didn't know the damages were listed there. Unstaled close utilt does 12 (3e)? They should pop out then....I guess I'm always stale in these scenarios or something because utilting a held opponent has been a staple of mine lol

 

[Bones0]

You can't wobble with multijabs?

 

[AceDudeyeah]

Why is it that sometimes when a person is grabbed, they're instantly released without them mashing out?

 

[Bones0]

Why is it that sometimes when a person is grabbed, they're instantly released without them mashing out?

They probably moved the stick or inputted their movements for a split second after getting grabbed. If you're at 0% or generally low, you can accidentally mash out of grabs quite easily by following through with L-cancel + shine or inputting a second aerial.

 

[Mahie]

Isn't AceDude talking about what happens when you grab people near edges sometimes?

And I'm also interested in the fastest way to mash out of a grab.

 

[Bones0]

Oh yeah, if you grab someone as they slide off that can happen. Idk the specifics on it. Grab releases are just based on your percent and how many inputs you can do (and also getting pummeled or attacked ofc). M2K's Statistics List has each input that counts towards a release.

 

[Kadano]

Kadano, I was wondering how the hitboxes on fair work. I feel like there are like 4-5 different hitboxes, but I am having trouble pinpointing exactly where they are and how they work. I found your utilt diagram immensely helpful and I was wondering if you had anything of the sort for fair? Thank you based frame-data god.

How’s that for starters?

Here is one with altered colors so you can distinguish between the individual frames:

If you don’t like these, I can also make them in the same style as utilt.

 

[TKD+ITA+Mar=]

Thanks a ton Kadano! I am a bit confused though. Doesn't the angle change over time? (Hence the reverse fair hitbox?)

 

[Bones0]

Thanks a ton Kadano! I am a bit confused though. Doesn't the angle change over time? (Hence the reverse fair hitbox?)

Reverses are decided by which side of their hurtbox collides with the hitbox. It doesn't affect the KB angle or strength at all. A reverse sourspot fair will have the same angle as a forward sourspot fair though, for example.

 

[AceDudeyeah]

Isn't AceDude talking about what happens when you grab people near edges sometimes?

And I'm also interested in the fastest way to mash out of a grab.

Exactly.

One more thing that baffles me and makes me SWEAR that it's a mechanic in the game rarely encountered and whenever it is, people blame it on themselves: Up-B'ing facing away from the edge when you just FELT like you were holding towards the edge.
Examples:
a. You were facing away from the stage(referred to as away) as you ran off to edgeguard. You up-b facing towards the stage (referred to as towards) and get hit by an attack/recovery, making you face away again. Holding the same direction the entire time, you input towards up-b again but end up up-b'ing away. This is probably the most common time I have experienced this and remember because I KNOW I was holding up and towards the entire time all the way from before getting hit to the up-b of death.

Scenario to test: Hold that direction as you get hit out of an up-b towards and then before or at the same time you exit a hitlag state with you facing away, try inputting just the b-button to trigger an up-b towards. If this replicates the up-b away, perhaps it's in the fact that you need to fully wait until hitlag is over or that you need to actually input a change in direction on the left stick as well in order to make Marth turn around during his up-B.

b. You were holding NE(towards), and to adjust spacing you move it NNE(a movement to the left but you're still on the right side of your left stick's range). As your thumb was making this movement, you input B and you end up up-b'ing away.
If this replicates it, then it is the act of moving in that direction that causes Marth to turn around, not whether or not you are actually in the right half and then the left side of your stick's field.

c. Unlikely but one other thing that can be ruled out. Up-b'ing towards with in the IASA frames of an aerial facing away. Try holding the direction towards during the aerial as well.

Thank you, Mechanics Majesty.

 

[Bones0]

Hey Kadano, can you explain this?

 

[Kadano]

@Acedudeyeah: Give me an example attack to test. Against another Marth’s fsmash you need to tech so you can ↑B again. Both during walljumptech (holding towards and up always results in that) and afterwards, pressing B on the same held direction causes the ↑B to reverse.

@Bones0: What exactly is there to explain? I didn’t notice anything weird around 5:20.

 

[Bones0]

@Bones0: What exactly is there to explain? I didn’t notice anything weird around 5:20.

Leffen uthrows Shroomed onto the top plat, and when he DJ nairs through the plat to hit him, he immediately lands. The same thing happened to me a few days ago when a Sheik player DJed through a top plat with a bair. I've seen platform cancelled aerials happen as people drop through, and I've seen a few particular rising ones like Mario's uair, but that happens all the time while these seem to only happen because of the opponent getting hit.

 

[AceDudeyeah]

Up-b to the right and hold that direction.
Have any character offstage hit you at 0% from the left with a weak aerial/fox shine so it turns you around without putting you in tumble state so you do not walltech or get stagespiked through out the entire hitstun phase.
Try inputting just the b-button without having to input any direction change since the first up-b. See if you still turn around if you were inputting it right before you exit hitstun. (As, long as it's before you're completely back at aerial neutral state. I don't know if up-b's buffer like double jumps do, but whenever the first frame/frames you're able to input it.)

 

[Kadano]

Leffen uthrows Shroomed onto the top plat, and when he DJ nairs through the plat to hit him, he immediately lands. The same thing happened to me a few days ago when a Sheik player DJed through a top plat with a bair. I've seen platform cancelled aerials happen as people drop through, and I've seen a few particular rising ones like Mario's uair, but that happens all the time while these seem to only happen because of the opponent getting hit.

Alright, I got it. Sorry it took me so long, I’ve been hosting a 64-man tourney this weekend on a local anime convention so I’m still somewhat tired.

SCDQ update routine seems to be the cause. What I have found so far suggests that every airborne action gets its SCDQ updated to the current substate roughly 10 frames afterwards. These 10 frames also count during hitlag. So Fox used his second jump and thus started the SCDQ update timer, naired, and then the substate update occured while his Δy was still halted during hitlag. Thus he went into landing lag while Dr. Mario was still in hitlag.

Up-b to the right and hold that direction.
Have any character offstage hit you at 0% from the left with a weak aerial/fox shine so it turns you around without putting you in tumble state so you do not walltech or get stagespiked through out the entire hitstun phase.
Try inputting just the b-button without having to input any direction change since the first up-b. See if you still turn around if you were inputting it right before you exit hitstun. (As, long as it's before you're completely back at aerial neutral state. I don't know if up-b's buffer like double jumps do, but whenever the first frame/frames you're able to input it.)

I did that. Marth still turns around (reverse Dolphin Slash).

 

[Bones0]

Alright, I got it. Sorry it took me so long, I’ve been hosting a 64-man tourney this weekend on a local anime convention so I’m still somewhat tired.

SCDQ update routine seems to be the cause. What I have found so far suggests that every airborne action gets its SCDQ updated to the current substate roughly 10 frames afterwards. These 10 frames also count during hitlag. So Fox used his second jump and thus started the SCDQ update timer, naired, and then the substate update occured while his Δy was still halted during hitlag. Thus he went into landing lag while Dr. Mario was still in hitlag.

I did that. Marth still turns around (reverse Dolphin Slash).

Awesome. Is this possible/feasible with any character/move? Do certain moves make it easier or harder, or is the whole thing based around being in hitlag at a certain spacing above the plat? I could imagine a lot of crazy stuff being possible if you could do this consistently. Think rising fair onto a plat as a tech chase comboed into fsmash.

 

[Kadano]

Awesome. Is this possible/feasible with any character/move? Do certain moves make it easier or harder, or is the whole thing based around being in hitlag at a certain spacing above the plat? I could imagine a lot of crazy stuff being possible if you could do this consistently. Think rising fair onto a plat as a tech chase comboed into fsmash.

No, that phenomenon is limited to certain (probably very few) scenarios. The aerial needs to have a longer downward SCDQ extension than the neutral air state. For Marth, this is only the case at the very end of his second jump. But at that point it hardly takes any longer for him to land regularly, so the benefit is negligible.

 

[AceDudeyeah]

http://www.youtube.com/watch?v=WuCCmymFpMU&t=46s

The flicker pivot. You dash to the right as normal, then flick the stick left and let go immediately. Is it a 1-frame window for letting go of left to avoid dash dancing?
Can you merely tilt left as if doing an ftilt rather than fsmash?
Is the timing less strict if you were to tilt rather than smash left?

 

[Bones0]

What determines grabbing the ledge from an airdodge? Does your airdodge animation have to be completely over or something? I've always wondered why it seems so impossible to grab the ledge with Falco/Fox, but even a FFer like Falcon seems to have a much easeir time grabbing it so I didn't want to chalk it up to fall speed.

 

[AustinRC]

I

http://www.youtube.com/watch?v=WuCCmymFpMU&t=46s

The flicker pivot. You dash to the right as normal, then flick the stick left and let go immediately. Is it a 1-frame window for letting go of left to avoid dash dancing?
Can you merely tilt left as if doing an ftilt rather than fsmash?
Is the timing less strict if you were to tilt rather than smash left?

It's called a flicker because you actually flick the control stick and let it return to neutral. I think it's three frames because of the turn around? Basically you are just pivoting into the standing animation instead of a move. The timing to tilt rathern than smash feels tight just because of the difficulty of how gently you have to press the joy stick right after the pivot I believe.

 

[Kadano]

http://www.youtube.com/watch?v=WuCCmymFpMU&t=46s

The flicker pivot. You dash to the right as normal, then flick the stick left and let go immediately. Is it a 1-frame window for letting go of left to avoid dash dancing?
Can you merely tilt left as if doing an ftilt rather than fsmash?
Is the timing less strict if you were to tilt rather than smash left?

Yes, no, no.

To give a more meaningful answer to your questions, I first need to explain how pivoting is triggered.
Dashes are caused by moving the control stick to an x value of <65 (left) or >191 (right) while y is within 128±~48. During the dash, the Turn animation can be triggered by fulfilling the dash criterium of the opposite direction. If this criterium is met during the first Turn frame as well, the character proceeds to his Dash animation. If, however, x increases from, say, 64 to 65, he will not dash, but stay in the Turn animation.
Ftilt to the left can be triggered with fresh x values from 65 to 105.

The hard part about this is that no matter how gently you push the control stick to cause the pivot, it will only be so far that pressing A on the same position will cause an fsmash and not an ftilt. So you have two options:
1. Like AustinRC wrote and demonstrated in his video, let the control stick return to neutral after the pivot motion and then press it slightly forward again to do the ftilt.
2. Press A just after the pivot frame and simultaneously only lower the pressure on the stick, but don’t let it return all the way to neutral (or more specifically, the [106-150,106-150] area). In my experience, moving the stick a bit down too makes this easier.

What determines grabbing the ledge from an airdodge? Does your airdodge animation have to be completely over or something? I've always wondered why it seems so impossible to grab the ledge with Falco/Fox, but even a FFer like Falcon seems to have a much easeir time grabbing it so I didn't want to chalk it up to fall speed.

Yes, grabbing the ledge is possible after the last airdodge (EscapeAir) frame. Fox has 0,23 gravity, Falco has 0,17 and Falcon only 0,13, so Falcon accelerates considerably slower than the spacies.

 

[MookieRah]

In Marth dittos at very low percent is it possible to pivot fsmash someone if they DI away?

 

[Kadano]

No, that’s not possible. At 0% before fthrow, the fastest way to pivot is to dash for 4 frames. If the fsmash is executed perfectly afterwards, the thrown Marth gets his shield up just at the frame fsmash would hit him.
Small step fsmash would come out in time, but it doesn’t have the necessary range.

 

[AceDudeyeah]

I see. Okay thanks.
What is the maximum/minimum X value? I'm assuming 0 is the leftmost input possible.

 

[AprilShaw]

How’s that for starters?

What do BK/KG stand for?

 

[Kadano]

I see. Okay thanks.
What is the maximum/minimum X value? I'm assuming 0 is the leftmost input possible.

You’re welcome. Both x and y go from 0-255, but the hardware cuts this range down to approximately 23-232, with the exact borders varying from controller to controller.

What do BK/KG stand for?

BK = Base Knockback, KG = Knockback Growth.

 

[Bones0]

Kadano, could you explain M2K countering this Sheik's up-B? Does the counter use up her invincibility somehow? She was angling her up-B down (so the counter wasn't hitting her during lag). It looks like Sheik had already reached the invincible part of her up-B before the counter went off, but I'm not very familiar with Sheik's up-B invincibility. I always thought she was invincible starting at the explosion.

Clip in question is at 1:52.

 

[Kadano]

I looked at the individual frames and to me, it looks like the Sheik went for a 180° angle and thus lost her intangibility before she grabbed the ledge.
I reproduced the situation and when Sheik uses a downward angle, she always grabs the ledge while intangibility even if Marth’s counter activates from her ↑B.

 

[Bones0]

I looked at the individual frames and to me, it looks like the Sheik went for a 180° angle and thus lost her intangibility before she grabbed the ledge.
I reproduced the situation and when Sheik uses a downward angle, she always grabs the ledge while intangibility even if Marth’s counter activates from her ↑B.

Oh whoops, I should have realized Sheik was going sideways by her tag's movement anyway.

Thanks.

 

[strawhats]

wasn't even aware that m2k used marth in teams to complete a 2 v.1 comeback. The only time I ever seen the dude go marth in teams was back in 2k7 at MLG Long Island with Isai. Holy **** at that counter though.

 

[Mahie]

So when exactly can you counter her upB?

 

[Kadano]

When she moves to the ledge with a straight or upward angle, her intangibility disappears for 8 frames before she starts falling and thus is able to grab the ledge.
The teleport part of Sheik’s ↑B is only one frame (311 36 as it is called internally) that repeats for up to 20 times, less if she collides with a wall or ceiling during any of these. If she collides after [311 35], she skips straight to her vulnerable 8 frames and her hitbox doesn’t even come out.
Now assuming Marth’s counter connects on the first frame Sheik’s hitbox comes out ([311 36], first iteration), his counter will hit from frame 10 to 16, counting from the moment [311 36] starts.
This means that in case Sheik uses her ↑B so far away from the stage that she will go through all 20 iterations of her [311 36], she will be intangible the entire time your counter’s hitbox is out. Also, because her hitbox is considered a projectile and disconnected from her body, counter’s pseudo-hitlag will increase the duration this hitbox stays out. Now if your counter also hits your opponent, this is not a problem because the hitlag lengthens your effective intangibility so that the [311 36] hitbox is gone once your intangibility has ended. If, however, your counter whiffs, your intangibility runs out while the [311 36] hitbox is still there, so it will hit you.

In order to hit her out of her ↑B, all of these criteria must be met:
1. Sheik uses an angle between 0° and 180°. (Angles between 197° and 343° will make her grab the ledge directly from her intangibility.)
2. Sheik collides with a wall during iterations 3-16 of her [311 36]. (“During” means that this iteration will be canceled out. If she collides on the first iteration, [311 36] is skipped so her hitbox doesn’t even come out and she goes directly into her 8 frames vulnerable lag. If she collides on the second iteration, she will grab the ledge on frame 10, which is the first frame your counter would hit. If she collides on iteration 17, your counter hitbox is already gone and she can safely grab the ledge.)
3. Sheik is within your counter’s range.
4. You are behind Sheik, in other words “more offstage” than she is, when counter is triggered by her ↑B. (If you are on the other side of her, you will counter in the wrong direction.)

 

[Kadano]

I just noticed some posts that were written when I was replying to an earlier post, so I hadn’t seen them yet. Sorry. I’ll quote all of them so you get an alert.

why is it that when I do a move from the same positioning, even more stale than before, they pop out? Did they mash out of the grab? It doesn't feel like it, because I don't see the grab release animation, but I guess I could just be misreading the first few frames of the animation.

I'd be interested in knowing if Marth has any true wobble in 2v1 scenarios. Doubt it. Dtilt does like 10 fresh so maybe that but your teammate would have to time the pummel perfectly. idk

I'm pretty sure how the mash formula is checked is completely uncharted territory though. That seems very difficult to test.

You can't wobble with multijabs?

Why is it that sometimes when a person is grabbed, they're instantly released without them mashing out?

I'm also interested in the fastest way to mash out of a grab.

Issue 1: Utilt breaking a grab after repeated use – You probably hit the opponent when CaptureCut started. During CaptureCut, attacks hit just like normal without the grab-specific 6% threshold, damage halving and the 20 frame CaptureDamage animation.
Fresh utilt always breaks the grab (to get a non-tipper, you would have to stand so close that you’d also hit the grabbing player), so I don’t really see a different explanation than what you guessed.

Issue 2: Can Marth wobble with jabs? – Yes, he can. The grabbing player doesn’t even need to pummel. Marth can also wobble with dtilt, but it’s only a 2 frame window for every dtilt input if the grabbing player doesn’t pummel, so I think jabs are clearly better. (Every attack that can be repeated after 20 frames or less can wobble. If the grabbing player uses pummel and both players act perfectly synched, the attack may last as long as 40 frames and still maintains the wobble)

Issue 3: Explain instant grab breaks – This is usually caused by Falco’s laser. Grab connects, but on the same frame, his laser hits you and ends the grab. To trigger this phenomenon, you need to grab at a certain moment (1 frame window) before Falco’s second laser of his ledgehop double laser comes out. If you stand very close to the edge when doing this, Falco won’t land on the stage and experience ridiculous lag, preventing him from using his second jump and / or airdodge to get to the ledge. He has to use his ↑B, so if you manage to get the instant grab break, his stock is gone if you don’t mess up.
Download 60fps example video
Youtube upload of the same file
.gif of the most relevant part:

Issue 4: How to mash out of grabs quickly – I posted about that just a few days ago on the German Smash Boards. Here is the Magus420 reference that explains everything well. Basically, the perfect mashing consists of a different cardinal direction on the control stick every frame, and additionally a different button (A, B, X, Y, L, R, Z) every frame. Here are examples of Marth mashing out of a grab at 0% with 50% perfection (2 inputs every 2 frames), once with pummel and once without:

100% perfect mashing would have made him escape the pummel before its hitsbox would have come out.

 

[Mahie]

ELI5 : Never ever try it ?

(Regarding Countering Sheik's upB)

 

[Kadano]

ELI5 : Never ever try it ?

I wouldn’t say that. Umbreon certainly would, though.

Counter activates on frame 5. Sheik’s ↑B direction is determined two frames before her hitbox comes out. Thus, if you time it tightly, Sheik is not able to react to your counter by choosing an upward angle. Also, if Sheik moves away to make your counter whiff, you can simply edgehog her, so she probably won’t do that.
If you call the Sheik choosing a straight angle, go for it, I say.