On impact freeze and hit or blockstun after single attacks or in combos

Classic Fighting Games Super Street Fighter II Turbo
1

Conventions

In this article I will use the following conventions:
[LIST=1]
[]light attack: jab punch or short kick;
[]impact freeze: period of time a character shakes right after being hit or blocking an attack;
[]hitstun: the time it takes to jump or attack after being hit;
[]blockstun: the time it takes to jump or attack after blocking an attack.
[/LIST]

Motivation

So there was a discussion a while ago in the Super Street Fighter II: Turbo HD Remix section about recovery or impact freeze being different somehow during combos or either as the first attack or a combination or isolated attacks. The issue is illustrated by the following fact:

*Factual CPS-2 SF2 information

Ryu can only combo a jab or crouching short into a hadouken if the light attack is the first hit of a combo.*

In other words, say you select Ryu and the round starts. you approach the enemy and do a crouching short canceled into a hadouken. If you were close enough, you will obtain a 2-hit combination: if the enemy did not block the kick, he will not be able to avoid getting hit by the projectile. On the other hand, would you perform a combination of an aerial attack then a crouching jab or short, then special-canceling it into a hadouken, the enemy would always be able to block or reverse the hadouken. It is impossible to obtain a 3-hit combination that way.

Tests

Maj has said that one could test it going frame by frame, which is what I have done. This test can be done with a 60 Hz camera and the actual game, or with an emulator, as suggested.

At first, I have tried it in Super Street Fighter II X: Grand Master Challenge, which is the Japanese version of Super Turbo. This had the same problem Super Turbo would present, which is that the game has all speeds with frame skipping, which makes the results inconsistent. In order to avoid the problem, I have tested it in Super Street Fighter II: The New Challengers (or S2, for short).

In S2, I have selected Ryu as player 2. I have done so by pressing forward kick, which yields a more pleasant color for the gi. As player 1, I have selected Ken. He has no such appropriate colors, so I have chosen the default one, with sad results. Ryu would be the attacking player, while Ken would be the dummy. The test would proceed as follows:[LIST]
[]I press an attack button for Ryu;
[]as the attack makes contact with the enemy, the score increases and I pause the game. This would be frame 0;
[*]I start counting frames and take notes of what happens in the screen.
[/LIST]

The first test consisted of Ryu hitting Ken with a cr.short, with no previous attacks. The data obtained was:

[LIST=1]
[]16th frame: dummy still locked
[]17th frame: dummy gets pushed back
[]18th frame: Ryu retracts his leg
[]22nd frame: Ryu recovers
[*]28th frame: dummy recovers.
[/LIST]

The second test consisted of Ryu hitting Ken with a crouching short kick after aerial roundhouse. The data was:

[LIST=1]
[]15th frame: dummy still locked
[]16th frame: dummy gets pushed back
[]18th frame: Ryu retracts his leg
[]22nd frame: Ryu recovers
[*]27th frame: dummy recovers.
[/LIST]

This shows that impact freeze and recovery do not change for the attacker. However, the character being hit is held locked in impact freeze one extra frame when the crouching short is the first attack of a combo. The remaining recovery does not change, but is delayed by one frame due to the referred fact.

The next test was Ryu hitting Ken with a crouching strong punch, with no previous attacks.

[LIST=1]
[]16th frame: dummy still locked
[]17th frame: dummy gets pushed back
[]18th frame: Ryu retracts his arm
[]21st frame: Ryu gets his arm close to his body
[]24th frame: Ryu recovers
[]33rd frame: dummy recovers
[/LIST]

The following test was Ryu hitting Ken with a crouching strong punch after aerial roundhouse.

[LIST=1]
[]15th frame: dummy still locked
[]16th frame: dummy gets pushed back
[]18th frame: Ryu retracts his arm
[]21st frame: Ryu gets his arm close to his body
[]24th frame: Ryu recovers
[]32nd frame: enemy recovers.
[/LIST]

This also presents constant impact freeze and recovery for the attacker and constant (delayed) recovery for the character being hit, with the impact freeze being one frame longer when the cr.strong is the first attack. Certainly, the same would happen for fierce or roundhouse, that is, the impact freeze does not depend on the attack itself, but it is longer by one frame for isolated attacks and the first ones in combos.

Finally, I have tested what happens on block. The following two tests are about a blocked crouching strong as a single attack or after a previous attack. In that case, I have used an aerial roundhouse, but any other attacks would give the same results. The difference is that blocked attacks do not increase the score, but it was easy enough to pause when Ryu extends his arm, which is the first active frame of his crouching strong punch.

Blocked crouching strong as a single attack:
[LIST=1]
[]16th frame: character still locked
[]17th frame: character gets pushed back
[]18th frame: Ryu retracts his arm
[]21st frame: Ryu gets his arm close to his body
[]24th frame: Ryu recovers
[]34th frame: enemy recovers.
[/LIST]

Blocked crouching strong right after blocked aerial roundhouse:
[LIST=1]
[]15th frame: character still locked
[]16th frame: character gets pushed back
[]18th frame: Ryu retracts his arm
[]21st frame: Ryu gets his arm close to his body
[]24th frame: Ryu recovers
[]33rd frame: enemy recovers.
[/LIST]

Similarly, the impact freeze is longer if the move is the first attack of a block string. However, the recovery time for blocked attacks is a frame longer with respect to attacks that are not blocked.

Conclusion

In S2, impact freeze is 16 frames for isolated attacks and 15 frames during combos and block strings. The recovery for jab or short is 12 frames after impact freeze ends, while the recovery for strong or forward is 17 frames after impact freeze. These recoveries are one frame longer for blocked attacks.

Extra info

The inconclusive results when testing with Japanese Super Turbo, which has frame skipping, were:

cr.strong
11 frame: character still locked
12 frame: character gets pushed back
14 frame: Ryu retracts his arm
16 frame: Ryu gets his arm close to his body
18-19 frame (varies): Ryu recovers
25-26 frame: enemy recovers

11 frame: character still locked
12 frame: character gets pushed back
14-15 frame: Ryu retracts his arm
16-17 frame: Ryu gets his arm close to his body
19 frame: Ryu recovers
24-26 frame: enemy recovers

cr.short
12 frame: character still locked
13 frame: character gets pushed back
14-15 frame: Ryu retracts his leg
17-18 frame: Ryu recovers
21-22 frame: enemy recovers

11-12 frame: character still locked
13 frame: character gets pushed back
14-15 frame: Ryu retracts his leg
17-18 frame: Ryu recovers
21-22 frame: enemy recovers

This is only what I could see and results would vary, due to frames not being shown.

2

What’s the practical implications of this?

3

There’s lots of stuff to learn from this. One thing is tick throwing. If you tick someone, and they block your attack, the timing for a frame tight throw is different than if they are hit by your attack. If you tick someone with a light attack, the amount of frames you need to delay your throw attempt is different than if you ticked with a medium or heavy strength attack. If you tick someone off of one attack, the timing for a frame tight throw is different than if you tick someone off of a combo or a blockstring.

And because throws in SSFIIX are instant, or zero frame, the difference of one frame of stun (or freeze) while tick throwing, factors into competitive play quite heavily.

Not only that but think about it from the perspective of someone trying to reverse a tick throw. If your opponent ticks you with a medium or heavy attack, you have more frames, a greater input window, to execute the necessary motions for a reversal attack (such as a dragon punch), than if your opponent ticked with a light attack. If your opponent ticks you with a light strength attack, then you have less frames to input your reversal attack motion, so you must be quicker and more precise with your reversal motions in that situation.

So if your opponent ticked you last time with a medium attack, but this time mixed his tick up and went with a light attack, you’ll have to adjust (on the fly) and alter your reversal timing or else you’ll eat the throw. And reacting split-second like that is extraordinarily difficult in SSFIIX due to the incredibly fast pace of the game. Fortunately techniques like piano reversals can help overcome obstacles like this by allowing you to spread your reversal attempts across six frames however some situations and some characters are not afforded that luxury during every reversal chance.

This shows one aspect of how furiously hardcore and demanding SSFIIX really is. Throw timing is so brutal at competitive levels because not only are throws 0-frame, but the timing for frame tight tick throws actually varies based on lots of factors that happen so quickly that they are often too difficult to properly react to. And not just factors of what the tick preceding the throw was, but the timing for frame tight throws and reversals also varies with frame skipping (as shown above), and varies because of the stage speed differences in the arcade version of SSFIIX. So not only must you adjust on the fly with your throw and reversal timing depending on what sequence of moves came before that, but you must alter your timing based on stage speed as well.

A good example of how tick throwing timing is different based on the normal used to tick is Hawk’s 720. Go into training mode, set the CPU to never block, and then try to do a 720 off of Hawk’s crouching short kick and Hawk’s crouching medium punch and see which is easier.

4

Hmmm this does explain a few things like being able to combo Zangief’s cr. lp into cr. mp/hp when the frame data seemed to imply otherwise, but I wonder if this was corrected/altered in ST or HDR. I remember asking Ganelon if the extra hitstun for the first hit of a combo was a myth and IIRC he said it was. I’ll have to see if I can find the thread I asked him that question in.

EDIT: http://shoryuken.com/f280/tick-throw-question-234794/index5.html

5

Thanks VF4!

This would mean that hit stun and block stun are not the same duration in all cases, which has been a cardinal rule for a while now. NKI in particular has been vocal about there being no difference. But what this means is that “take this hit” actually exists, and gives you one less frame of stun (if I’m readying the OP correctly).

This is why I asked about practical implications :wink: It would seem that adjusting timing on the fly based on whether the attack used by the opponent was LK/LP or otherwise unless you can play like a robot…?

Still, all very interesting.

6

It’s known that standing blockstun and hitstun against medium and heavy jumping attacks are different. I though that ‘sac throw’ was a relatively known technique.

7

Folks, the fact that cross-up roundhouse, cr.short xx hadouken never combos in ST or SX proves that, even tough the difference may not always be clearly shown on the screen, the game still shortens impact freeze for the defender so that it does not form a 3-hit combination. The one result that I honestly did not expect was blockstun being longer than hitstun, but this was only proved in S2, not ST/X. If one of you can not test it, I will do it as soon as I get home. This will prove if ST has this difference or not internally, which is what matters, in the end. It goes as follows:[LIST=1]
[]select Ryu on one player;
[]select a dummy character on the other one;
[]have Ryu do a cross-up aerial roundhouse;
[]have the dummy block the aerial attack;
[]have Ryu do a crouching short kick;
[]have the dummy block the kick;
[]have Ryu special-cancel the kick into a hadouken;
[]have the dummy try a reversal or jump.
[/LIST]If the dummy can not jump or reverse due to still being in blockstun, indeed it is one frame longer than hitstun in ST too. Would the first two attacks hit for damage, the character could attempt a jump (even tough still getting hit, but it would never be a 3-hit combo) or reverse every time.

Edit:

Indeed. Newbies and experienced players all knew about it back then. In fact, ST/X’s speed made the technique harder than it used to be, so maybe that is why it may seem obscure or the like by now.

8

No, hitstun and blockstun was always thought to be equal for particular attacks. So yes, you’re correct that the numeric quantity of hitstun and blockstun is different for jumping attacks, but it still being equal for those attacks meant that there was no advantage to “taking the hit”. This new information contradicts that.

9

I’ve known about jumping attacks having difference in frames for hitstun and blockstun well before I registered at SRK. I remember asking GigaMSX why it was easier to execute charge special attacks after a sequence of moves that were part of a blockstring rather than being off of a combo. He said that a jumping attack when blocked yields more stun, enough stun that you are afforded enough extra frames of charge to incorporate moves like a double knee off of a blockstring, where if it was during a combo the double knee wouldn’t come out because you didn’t have enough charge time.

The example sequence for this will be Dictator’s:

  • crossup j.rh, st.lk, st.lk, cr.mk, double knee (scissor kicks)

Dictator’s j.rh causes 11 frames of hitstun and 20 frames of blockstun on a standing opponent. So if you try for that above sequence as a combo you have 9 less frames to try to get enough charge for the double knee. And Dictator’s double knee has a charge time of 71 frames which is the longest in the game. However if your opponent blocks your j.rh and you continue with the blockstring, you have nine extra frames during that sequence to acquire a charge of 71 frames after the crossup j.rh, which is why it’s easier to perform a double knee at the end of a blockstring rather than a combo.

This is why you never see Dictator players go for scissor kicks with his big ToD combos. They go for psycho crusher instead because psycho crusher has a charge time of 55 frames. Go into training mode and try to to the above sequence as a combo and see how difficult it is. In order to time the sequence as a combo you have nine less frames of leniency to build up a charge, the margin for error for that string as a combo is quite high, which is why it isn’t practical. However then set the CPU to block and try that sequence as a blockstring, you should find that it is easier to get a double knee at the end of the sequence when it is a blockstring.

  • crossup j.rh, st.lk, st.lk, cr.mk, double knee <- needs 71 frames of charge

  • crossup j.rh, st.lk, st.lk, cr.mk, psycho crusher <- needs 55 frames of charge

So a sequence with a psycho crusher needs 16 less frames of charging time than a sequence with a double knee on the end of it.

Lastly, the hitstun from a jumping heavy attack on a crouching opponent is 21 frames, which is one frame higher than if they blocked the attack standing (because you have to be standing to block most aerial attacks). That is why in combo videos you see lots of characters [media=youtube]ZaD5twbdSNw&feature=related"]ducking and eating a crossup to start some huge ridiculous combo. They duck for two reasons, one being that they get put into hitstun for a long time allowing for longer charge, and two because hitting a ducking opponent with a crossup pushes them directly into you allowing for extra hits in the combo. In NKI’s touch of death video he [URL=“http://zachd.com/nki/NKI-Vol4.Super.Turbo.txt”[/media]:

In Super Turbo, ToD’s are present (though not nearly as abundant as they were in
WW). The requirements for ToD’s are as follows:
-Must not require special set-up, like the opponent to be ducking.
-Must not require a super (though it can use a super, provided that the meter is
built up entirely within the combo itself).
-Must not be character-specific. This means that it must work on more than one
character. Ken, for example, has a ToD that only works on Zangief. This is
not a true ToD.

10

Fair enough, but there have been a lot of threads where people have said something like “sac throwing is a myth”.

11

A long time ago I was told, or I read somewhere that the 2nd hit of a combo putting the opponent in 1 frame less stun than if it were the first hit. Good to see it confirmed. I’m wondering though is this a continued effect? What is the data for a third hit?

12

Nice thread old school BR, thanks for taking the time to do the tests. Hitstun being longer for the initial attack does explain why cr Short, Hadoken will combo but not cr Short x2, Hadoken.

One thing I’ve always wondered, and hasn’t yet been answered in this thread or any other is how come Ryu can combo cr Short x2 or cr Short, cr RH but not cr Short, cr Forward or cr Short, cr Strong. All of Ryu’s crouching normals (and Ken and Akuma too) have a startup of 4 frames or less, so it doesn’t make sense that cr Short, cr Forward will not combo even though it has the same startup time as cr RH or cr Short.

13

One thing that has always bothered me is: isn’t the amount of hitstun/blockstun generated by jumping attacks largely irrelevant? If you hit the opponent’s head with j.HK when you’re still high in your jump arc, more time has to elapse as your character lands on the ground and you can perform your next move. Whereas if you hit your jumping attack late, you’re closer to the ground so have less time between attacks. My point is that there will always be some time when your character is still descending and that it’s likely variable each time unless you play (again) like a robot.

14

Interesting thing here: for short attacks, the reversal frame in S2 (no frame skipping) is the frame 25 after blocking or hitting. But if you try to jump that frame, it does not work. I have tried reversals at that frame and held the button to be sure there was no negative-edging involved, and it works every time. I have also tried jumping at frame 25 on hit, and it works. But if you block, it does not.

On frame 25 after a jab or short, if a single character attempts a throw, it gets it, whatever character it is, on hit or block. But if both attempt the throw, the one defending always wins, which is the same result Ganelon had obtained.

More testing coming…

Edit: to make it sure, the button is pressed at frame 24 and must remain pressed for the transition to frame 25. If you do a special or a throw, it works. Jumping or normal attacks do not come out that way, on block.

Edit 2: as expected, for short attacks as second hits of combos or block strings, the reversal must take place from frame 23 to 24, that is, one frame earlier, to due the reduced impact freeze.

15

Odd thing about the game is that the hang time after an air attack at the same height seems to be variable depending on the direction of jump, character attacking, attack used.

For instance if Honda knocks Dictator down in the corner, Honda can do a veritcal jump and hit Dictator with a Short attack at the top of his jump arc, and then Honda seemingly teleports immediately to the ground.
Honda can immediately Oicho.

I would have thought that blocking or getting hit by that attack very high up would give Dictator the ability to absorb the light attack and hit Honda with his own attack before he could land. Nope.

It would be interesting to see which jump moves have which properties.

16

Thanks 'spacer, that’s exactly my point: that knowing the amount of blockstun/hitstun for jumping attacks is interesting on a theoretical basis but it doesn’t seem like you could use that knowledge in a practical (i.e. repeatable) way because of the variability in those situations.

17

A specific example is that when Ryu hits a crouching opponent with a (deep) jumping roundhouse, he can link to towards + FP, but not if the opponent is standing.

18

Well, it gives him the ability to throw if he recovers after Honda touches the ground. That way he will have the reversal on his side.

I do not agree. Every time you jump at a character who is recovering from a projectile and you hit him early, you know because he was hit that he will recover fast, so you often want to try a throw as soon as you land or an invulnerable move. It is a very common situation.

19

Fair points, I can see how those examples are valid.

20

I was actually assuming you were asking about the first hit after knockdown since that topic was where everyone was talking about whether the first frame upon wakeup was special or not. There’s no extra stun from a reversal or meaty. After some tests just now, I can confirm that oldschool_BR is certainly correct about the first move of an attempted combo having 1 frame more stun than all subsequent moves. And it makes sense for gaming properties as well.

I tested again and stand by hitstun and blockstun having the same frame duration in ST. Either SSF2 is different or you may be missing something. Like I’ve mentioned in the past, HSF2 T0 is the most accurate way to test ST in no-frameskip. Note also that at no-frameskip speed, there are often, if not always, 2 reversal frames; make sure you’re not accidentally doing the later reversal and confusing that for 1 frame longer stun. For the most straightforward tests, try having the defensive character execute a reversal that’s very easy to perform and see on the first frame, such as Chun Li’s lightning legs.

As for the sac throw, it has use but only within a very limited window. Against fast jumps and deep jumping attacks, your opponent will just combo you. Against jumping light attacks, it’s always a waste of life to sac throw when you can block with the same stun. Against slow, very early medium or hard jumping attacks, you can defend and close the longer distance resulting from the knockback to throw without taking damage. But against medium or hard jumping attacks that aren’t too deep or too early, that’s when sac throws have their use by trading damage.

Still, even then, it’s often not worth going for since you almost always come up short on damage with a teched throw. The situations I can think of where a sac throw would be useful are if:

-You’re playing a character with a command throw
-You’re playing a character with a hold
-They’re playing an old character
-You can afford to trade damage
-You think they’ll forget to tech the throw