• Welcome to Smashboards, the world's largest Super Smash Brothers community! Over 250,000 Smash Bros. fans from around the world have come to discuss these great games in over 19 million posts!

    You are currently viewing our boards as a visitor. Click here to sign up right now and start on your path in the Smash community!

A more data-driven look at "safe" and "unsafe"

Zankoku

Never Knows Best
Administrator
BRoomer
Joined
Nov 8, 2006
Messages
22,906
Location
Milpitas, CA
NNID
SSBM_PLAYER
Shieldstun, shieldhitlag, and frame advantage
When an attack hits, there's an amount of time where the person who shielded can't do anything (shieldstun) and the person who attacked will be in freeze-frames (shieldhitlag). To determine advantage, you take when the attack hits, add the shieldstun, and then subtract the shieldhitlag and total framecount. In other words,

HIT + SHIELDSTUN - SHIELDHITLAG - TOTAL = ADVANTAGE

Advantage is typically negative in Brawl, from what I can tell, so don't be too disappointed.

Human Reaction
While people can properly time things and react as quickly as 0.1 second, or 6 frames, after something hits, a normal human will tend to react in 0.2 second, or 12 frames, to something they don't expect. Add these 12 frames to your advantage value to get a "virtual" advantage value. If you end up with a positive number, the attack is pretty much safe.

If you feel like further adding numbers to make you feel better, it takes 4 frames to put away your shield, so add +4 if your opponent wants to do something involving unshielding and attacking.

Unsafe Attacks
So we'll go with the obvious example of something unsafe on block, Marth's fsmash.

TOTAL: 50
HIT: 11
SHIELDSTUN: 13
SHIELDHITLAG: 4
ADVANTAGE: -30
VADVANTAGE: -18

In other words, after shielding the hit the opponent will have 30 frames (half a second) to punish you before you can do anything about it. Even factoring in human reaction gives them an 18 frame (.3 second) window to unshield and do something.

Deceptively Safe Attacks
Now, let's take a look at landing a tipper hit with an fsmash.

TOTAL: 50
HIT: 12
SHIELDSTUN: 14
SHIELDHITLAG: 6
ADVANTAGE: -30
VADVANTAGE: -18

Wait, what? It's the exact same as a non-tipper! However, this is at tipper range and the shieldhit will cause the opponent to slide back, forcing him to waste some time running back up to you. Doesn't really help against projectile users, though.

Safe Attacks
And of course, Marth's amazing dtilt has to come under scrutiny.

TOTAL: 22
HIT: 7
SHIELDSTUN: 12
SHIELDHITLAG: 5
ADVANTAGE: -8
VADVANTAGE: 4

An actual safe move, unless your opponent sees it coming and reacts quickly enough. If he doesn't, you're gaining +4 frames on him with this attack on his shield.

Powershield
Watch out for the powershield, though. The behavior is as follows:
No shield damage
Shieldstun is reduced to equal that of shieldhitlag
Character does not get pushed back by the attack at all
Shield is not actually activated, meaning they don't get the 4-frame loss of unshielding

You can look at it as basically your attack completely missing them. It's impossible to powershield multihit attacks in Brawl, unfortunately (unless you click that shield trigger for each and every hit somehow), but that's fine because powershielding anything with Marth will give you an adequate window to counter with Up+B any time.

Here is what happens to an fsmash tipper that gets powershielded:

TOTAL: 50
HIT: 12
SHIELDSTUN: 6
SHIELDHITLAG: 6
ADVANTAGE: -38
VADVANTAGE: -26

...Yeah, don't spam smash attacks.

Hope this helps.
 

VietGeek

Smash Hero
Joined
Mar 19, 2008
Messages
8,133
Is the Hit the frame it comes out on? Or when the attack can end?

For the most part, it looks good. I just have to sit here and think for a while and I should understand all of this from a frame perspective.

Yay Ankoku.
 

Shaya

   「chase you」 
BRoomer
Joined
Jun 8, 2007
Messages
27,654
Location
/人◕‿‿◕人\ FABULOUS Max!
NNID
ShayaJP
I love this type of stuff with a passion.
Like not exactly the numbers, just the direct conclusion of the numbers.

New avatar? I thought the old one was snazzier ;_;.

Thanks Ankoku.
 

Zankoku

Never Knows Best
Administrator
BRoomer
Joined
Nov 8, 2006
Messages
22,906
Location
Milpitas, CA
NNID
SSBM_PLAYER
HIT is the frame the attack hits on, obviously. I'll think about doing other attacks later, I just did these as an example.
 

Emblem Lord

The Legendary Lord
Joined
Aug 11, 2005
Messages
9,720
Location
Scotch Plains, NJ
NNID
ShinEmblemLord
3DS FC
3926-6895-0574
Switch FC
SW-0793-4091-6136
And this is one of the reasons why I hate smash.

Because no attack gives "true" frame advantage on block, like in other fighters.
 

¯\_S.(ツ).L.I.D._/¯

Smash Legend
Joined
Apr 27, 2008
Messages
12,115
Location
Chicago, IL
I agree with ^. I would like to definitely like to see at least like Fair and Bair and some others. But full moveset would be very helpful. Thanks Ankoku.
 

Nibbity

Smash Journeyman
Joined
May 20, 2008
Messages
368
Location
Connecticut
i'm truly impressed, this gives me more insight of safe moves, etc. Thank you very much. It's scary seeing what a PS can do to...well, almost anything Marth has in his moveset.
 

GPEternity

Smash Journeyman
Joined
Jul 31, 2008
Messages
271
Location
Bay Area, CA
unfortunately i train on computers that'll shieldgrab everything and an old ladies panties. but its at least good to know there are some attacks that can be considered safe against real people.
 

VietGeek

Smash Hero
Joined
Mar 19, 2008
Messages
8,133
unfortunately i train on computers that'll shieldgrab everything and an old ladies panties. but its at least good to know there are some attacks that can be considered safe against real people.
Which is what the Virtual Advantage is for. To show the advantage you have on normal people.

Also, does Ankoku like change avatars every 4 days? Why Ankoku, why? <_<
 

3xSwords

Smash Lord
Joined
Mar 14, 2008
Messages
1,193
Location
Bergen County
So is there any move for Marth that would register as safe even without human reaction being considered? I know EL said there was none but there should be at least one move that is completely safe right? I'm thinking along the lines of FF fair landing in the middle of attack animation. But is that not safe either?
 

Zankoku

Never Knows Best
Administrator
BRoomer
Joined
Nov 8, 2006
Messages
22,906
Location
Milpitas, CA
NNID
SSBM_PLAYER
I'll test out fair and uair maybe. Also, Shieldbreaker is weird because its shieldstun is equivalent to the shieldhitlag.

Does anyone know the timing for autocanceling a fair?
 

3GOD

Smash Ace
Joined
Jul 23, 2006
Messages
745
Location
Athens, GA
I've been investigating this with a few characters (mainly ones I play/want information about). Here is the data I've gathered on Marth's aerials:

Nair:
First hits on frame 6
Landing lag: 9
Shield hit lag: 12
Shield stun: 13
Advantage: -8

Dair:
First hits on frame 6
Landing lag: 32
Shield hit lag: 11
Shield stun: 12
Advantage: -31

Fair:
First hits on frame 4
Landing lag: 9
Shield hit lag: 11
Shield stun: 12
Advantage: -8

Bair:
First hits on frame 7
Landing lag: 21
Shield hit lag: 13
Shield stun: 15
Advantage: -19

Uair:
First hits on frame 5
Landing lag: 9
Shield hit lag: 12
Shield stun: 14
Advantage: -7

For each of these, the advantage assumes that Marth lands on the next frame after the shield hit lag ends. Also, for each of these, the attacks were tipped - for the Nair, the hit lag/shield stun data is for the 2nd hit.

Regarding your original post, I do not think it's realistic to expect the "virtual advantage" in Brawl especially for a couple of reasons. The opponent obviously expected an attack (hence why they shielded), and they also get a few frames of shield hit lag during which they can react. Not only this, but they can buffer jumps, grabs, rolls, dodges during the hit lag. Honestly, I think the true "advantage" numbers should be used when determining the safety of an attack.

Someone may want to double check these for me because I noticed a discrepancy when checking out the Dair. I think it might be the case that diminishing returns also reduces hit lag and shield stun (assuming you connect with the Dair on a non-shielding opponent a couple of times for instance).

Nevertheless, I think the data indicates that Fair is decently safe. Nair is a little riskier probably since you have to be closer to your opponent to get the tip, and Uair could be decent if the opponent has a slow grab or you're behind the opponent. Personally, I'd avoid using Bair and Dair (obviously) as shield pressure.

I'll test out fair and uair maybe. Also, Shieldbreaker is weird because its shieldstun is equivalent to the shieldhitlag.

Does anyone know the timing for autocanceling a fair?
As you can see from my data, a lot of moves don't have much shield stun (one reason why shielding is so awesome in Brawl, and shield pressure is more risky compared to Melee). Generally, the shield stun is only a couple of frames more than shield hit lag for aerial attacks (from all the characters I've tested thus far).

As for the auto cancel window, I highly doubt that the difference between hitting the shield and the auto cancel window (along with the standard 2 frames of landing lag) would be less than the 9 frames of lag on Fair, Nair, and Uair; but it might be the case for Nair.
 

Zankoku

Never Knows Best
Administrator
BRoomer
Joined
Nov 8, 2006
Messages
22,906
Location
Milpitas, CA
NNID
SSBM_PLAYER
Diminishing returns doesn't happen on shield hits.

In my tests with tipper hits, I've found that shieldhitstun is always disappointing compared to the shieldhitlag. You might be better off going for the max-range nontipper.
 

3GOD

Smash Ace
Joined
Jul 23, 2006
Messages
745
Location
Athens, GA
Diminishing returns doesn't happen on shield hits.

In my tests with tipper hits, I've found that shieldhitstun is always disappointing compared to the shieldhitlag. You might be better off going for the max-range nontipper.
What I had in parentheses was important there...I know that hitting a shield doesn't decay your moves, but what I was saying was that a move that is fresh might have more hit lag/shield stun than a move that is already decayed.

I'm not sure of this, but in my tests, the Dair had a hit lag of 12 frames a couple of times. Then I actually hit the opponent (not a shield) with Dair a couple of times. Then the hit lag was 11 frames. It's possible that it was the difference of a tip/non-tip, but I've seen this kind of slight variance in hit lag from other characters too. I've just assumed up to this point that my frame-by-frame analysis of the video was a little off (sometimes it's difficult to tell exactly when hit lag ends), but I'm beginning to wonder if decayed moves might affect shields a little differently than their fresh equivalents.
 

3xSwords

Smash Lord
Joined
Mar 14, 2008
Messages
1,193
Location
Bergen County
I've been investigating this with a few characters (mainly ones I play/want information about). Here is the data I've gathered on Marth's aerials:

Nair:
First hits on frame 6
Landing lag: 9
Shield hit lag: 12
Shield stun: 13
Advantage: -8

Dair:
First hits on frame 6
Landing lag: 32
Shield hit lag: 11
Shield stun: 12
Advantage: -31

Fair:
First hits on frame 4
Landing lag: 9
Shield hit lag: 11
Shield stun: 12
Advantage: -8

Bair:
First hits on frame 7
Landing lag: 21
Shield hit lag: 13
Shield stun: 15
Advantage: -19

Uair:
First hits on frame 5
Landing lag: 9
Shield hit lag: 12
Shield stun: 14
Advantage: -7
Can't you perform an action from an auto cancellable aerial at frame 4? So would Uair have an advantage of -2? And fair an advantage of -3?
 

3GOD

Smash Ace
Joined
Jul 23, 2006
Messages
745
Location
Athens, GA
Can't you perform an action from an auto cancellable aerial at frame 4? So would Uair have an advantage of -2? And fair an advantage of -3?
Actually, auto-canceled lag is 2 frames (meaning you can perform an action starting on frame 3), but the issue is that often an attack's hit box disappears a while before it is auto-cancelable.

For example, character X might have a Fair that attacks on frames 4 - 8 with a landing lag of 10 frames, but it may not auto-cancel until frame 24.

If you were to hit the opponent's shield on frame 4 and land on frame 5, there would be 10 frames of landing lag plus whatever shield hit lag the attack has and minus the shield stun.

If you were to hit the opponent's shield on frame 8 and not land until frame 24 (when you get the auto-cancel), you've got 16 frames between the hit and the landing plus 2 frames of landing lag plus the shield hit lag and minus the shield stun.

In this example, you actually are better off going for the standard lag as opposed to the auto-canceled lag. It's attack dependent as to which is better...for instance, I know that Peach's Fair benefits greatly from an auto-cancel (actually the only attack I've found thus far that doesn't have a negative advantage). I haven't really checked out Marth's auto-cancel data, but I'm fairly sure he's not going to gain anything by auto-canceling.
 

3xSwords

Smash Lord
Joined
Mar 14, 2008
Messages
1,193
Location
Bergen County
Srry I meant auto cancel as in landing during the attack animation. Yeah how would you describe that? Srry about my noob moment ><. But if you time the uair right you can do it so that right after the soonest possible moment the uair will attack, you land and cancel the animation. So if you did that would you be able to get a positive advantage?
 

Emblem Lord

The Legendary Lord
Joined
Aug 11, 2005
Messages
9,720
Location
Scotch Plains, NJ
NNID
ShinEmblemLord
3DS FC
3926-6895-0574
Switch FC
SW-0793-4091-6136
Marth has no auto cancel's on landing. That Uair lag is the same no matter when u land form it as far as I can tell.

Auto cancel lag is usually a few frames.
 

3GOD

Smash Ace
Joined
Jul 23, 2006
Messages
745
Location
Athens, GA
I think there is some confusion about what "auto-canceling" is.

Basically, all aerial attacks have a certain duration for their entire animation. If you land after the animation is finished, you get 2 frames of lag (just as if you landed from a jump with no aerial attack).

Some (maybe all?) aerials also have a certain frame after which they are "auto-canceled" upon landing. If you land after this frame, the attack auto-cancels and gives 2 frames of landing lag. This is different than just landing after the end of the animation; that is, the auto-cancel frame often occurs a few frames before the animation actually ends.

Peach's Fair is the easiest way to see this. If you float Peach's Fair and continue floating, you'll notice how much time the entire Fair animation takes. You can keep Fairing while floating to get a feel for how long the attack animation lasts. If you float a Fair and land right as the attack comes out, you will get a significant amount of landing lag (22 frames). If you land just a few frames later (like 4 or 5 frames later), you will get virtually no landing lag at all (2 frames) because of the auto-cancel. Notice that this attack auto-cancels well before its entire animation is completed.

In general, the auto-cancel frame is pretty close to the end of the attack animation. Many people have confused the low landing lag of some aerials with auto-canceling. For instance, Marth's Fair, Nair, and Uair all have 9 frames of landing lag - not too bad, but this is not auto-canceling. When one factors in generally low landing lag with auto-cancel frames being near the end of attack animations, it's usually not beneficial to get the true auto-cancel.

A general comment about shield pressure in Brawl:

Brawl is NOT Melee. Shield pressure cannot be achieved safely without good spacing or crossing over shields. For every attack I've investigated thus far, the trade-off between lag and shield stun is never advantageous. We are probably going to have to give up on the idea of pounding a shield with an aerial and then doing a quick ground attack to prevent a shield grab.

As this is the Marth forum, I seriously doubt that many of you think this way to begin with (since Marth has always been about good spacing), but some Melee Fox/Falco players would be well advised to focus more on spacing (and crossing over) and forget about using pure speed as a pressure tactic. Meta Knight seems to be the exception to this rule since basically everything he does is safe.

For the question about frame data, I capture the video using a lossless codec and then deinterlace to get 60 frames per second. Then analyze the video frame by frame to get the data. All this is done in VirtualDub.
 

Zankoku

Never Knows Best
Administrator
BRoomer
Joined
Nov 8, 2006
Messages
22,906
Location
Milpitas, CA
NNID
SSBM_PLAYER
2 frames? No, it's 1 frame/4 frame, depending on how hard of a landing/how far you fell. The actual landing animations are much longer (like 15 frames and 32 frames), but they're IASA-able as early as frame 2 and 5.
 

3GOD

Smash Ace
Joined
Jul 23, 2006
Messages
745
Location
Athens, GA
2 frames? No, it's 1 frame/4 frame, depending on how hard of a landing/how far you fell. The actual landing animations are much longer (like 15 frames and 32 frames), but they're IASA-able as early as frame 2 and 5.
When I said 2 frames, I meant that frame 1 is the first frame on the ground, frame 2 is another grounded frame (still considered lag in my opinion). On frame 3, any buffered move (buffered during the previous 2 frames) will begin - shield, jab, etc. all start on the 3rd frame of being on the ground from a short hop. If you mean that the lag is 1 frame in that you can input a command on frame 2, then you are correct that you can input a command on frame 2 (I'm pretty sure you can input one on frame 1 as well). But the actual movement of your character in response to your input will not begin until frame 3.

I didn't go into the additional lag from higher falls just to keep it simple. Since the context was auto-canceling, I figured that it was reasonably safe to assume that people wouldn't be interested in non-aerial attack lag from long distance falls. By the way, the lag on these longer distance falls is 5 frames (any input will result in the character starting that action on frame 6).
 
Top Bottom