Frame Data – What Even Is It?

There has been a lot of controversy lately about AAA game releases, the new consoles, and frame rates. Will it be 30fps or 60fps? Will players notice? Will they care? What new buzzwords can the marketing department come up with to obfuscate the issue? The entire discussion is centred around graphical showcases, and there is one genre that is curiously left out of the conversation: fighting games.

I could digress into the history of big fighting game releases and how they were once the graphical powerhouses of consoles, often because they were being ported from superior arcade hardware, but that’s not what this is supposed to be about. The point is that nobody really cares much about frame rates in fighting games, because for about as long as console fighting games have been worth playing, they have almost all run at 60fps.

It’s so universal that their 60fps frame rates have become a universal measurement within the genre. Actions in fighting games are not seconds or even fractions of seconds, but in animation frames, and since the games reliably move at 60 fps, those measurements are expressed in the form of 1 frame = 1/60th of a second. A move that is i6 (the “i’ standing for impact, meaning that is the first frame in the move where it can make contact with an opponent and do damage) takes 6 frames to come out, or 1/10th of a second. Those numbers are so small that no human can really react to them, which is one reason why frame data ends up looking a lot like math, with players trying to base decision on numbers instead of reactions to images on the screen.

But how important is frame data? Where does it come from? Who uses it, and how? That’s what I’m here to explain.

The Time Before Time

Frame data was not always a thing. I mean, it always existed, it was always there, but players had little access to it. In an arcade, it wasn’t practical (or really possible) to take the laborious steps needed to work out specific frame data, so players opted for the next best thing and generalized.

One of the cornerstones of fighting games–and any other competitive game, but fighting games are perhaps the best at making the principle explicit–is punishment. One player does something that leaves them vulnerable, and the other player capitalizes on that opportunity. The primary use for frame data in fighting games, especially 3D fighting games, is letting a player know when and how to punish (and when and how they can be punished themselves). Every time an attack is blocked or evaded, each player is in a rigid, set position. How they respond makes all the difference in the world, and is the main reason that knowing character match-ups is important.

That’s getting ahead of myself, though. What is important is the fact that punishment has always existed, even before frames were known. This is expressed in the dual concepts of safe and unsafe. A move that is safe (when blocked, usually) is one that cannot be punished with guaranteed damage. A move that is unsafe is one that can be punished. These exist in degrees, because every move will have different recovery frames (The number of animation frames that it takes for the attacking character to return to a neutral state, at which point they are able to defend themselves.) as well as block stun (The number of animation frames it takes for the defending character to return to a neutral state, at which point they are able to move or attack.).

In Soulcalibur a move could be AA-punishable (A standard button setup being A for horizontal attacks, B for vertical attacks, K for kick attacks, and G for the guard button. For the majority of characters their standard AA combo is their fastest guaranteed damage.), BB-punishable, and so on down the speed list, all the way to launch punishable, which means that if that move is blocked the defender will get a full combo. There are other factors to consider, of course; a move can leave the attacker in a crouching state, in which case an AA combo will miss, as they are universally high attacks and go over the heads of crouching characters; a move may leave the defender in a crouching state, in which case they would have to stand up before they could use many moves, and may have to rely on slower or less damaging crouching or rising attacks; a move may leave the attacker too far away for some characters to reach, or the defender may have been too far away to begin with. There are other scenarios as well, but you get the point.

There is math behind all of that, but it came later. It was enough to know that a blocked move gave an opponent enough time to respond with a specific amount of damage, without knowing that move was -12 (frames) on block, and their AA attack was i11.

In fact, some Soulcalibur characters are so specifically built around the concept of safety and punishment that they became part of the standards. Sophitia, the sword and shield using Greek warrior, is the prime example. Her Angel Step and Twin Angel Step B attacks are iconic in the series, and are one of the measures used for a given move’s safety. Consider this: in Soulcalibur 4, Sophitia’s TAS B:4 just frame (the colon in move notations is shorthand for a just frame input that must be timed very precisely) came out in 15 frames and did 80 damage, or could combo into itself for half of her opponent’s life bar at certain distances. Very few other characters could reliably put out that much damage against -15 frame moves, and even those that could didn’t always have the range to pull it off. If you are playing against a competent Sophitia then the moves you use have to be adjusted accordingly. If you are using a particularly slow character against a good Sophitia player, you can expect to see this sort of thing often if you don’t manage bad habits:

The opposite is just as true. Some characters have no reliable punishment when blocking such moves and must depend on the opening for a free mixup instead, that could give them as much damage, but probably won’t.

A move that is in the -15 range on block is pretty obvious, though. They are generally your big-ticket moves, powerful-looking attacks that have an obvious recovery time, or low attacks. Most experienced players can intuit the type of move that would qualify for TAS B:4 punishment just by looking at it and what it’s supposed to do. That sort of thing is also easily tested in matches and in training mode. For that reason, the frames involved aren’t even that important. A TAS B:4 punish would be found whether the exact frames of any given move were known or not.

Where actual frame data becomes important is less intuitive.

Small Numbers, Big Impact

As I’ve said, it’s possible to play fighting games without knowing frame data. Many good players have done so, though it can take more work on their part, or they may simply be the type of player who gets by on good execution and reads. However, frame data is still important and valuable, only not in some binary safe or unsafe paradigm.

This is where the math comes in. Don’t worry, I am terrible at math, so this will not be complicated.

Let’s take a classic example: the frame trap. What is a frame trap? It is a situation where an attacking player creates an illusion of vulnerability in order to lure the defender into attacking when they shouldn’t. How is that accomplished? Purely through minute frame differences that are not obvious to the naked eye–if they were, it wouldn’t be much of a frame trap.

A player who falls for a frame trap can end up looking pretty stupid, which is why I’m going to use myself as an example.

Rewind the clock 5 years. I bring my camera to the recreation area of a downtown apartment building. Today we will by playing the Soulcalibur 4 Regionals, a semi-major tournament that will see players travelling from across the border to participate. This is back before streams and laptops with capture cards were a regular feature of all but the biggest and most well established tournaments. At this point in time I am pretty much my community’s archivist, using a Canon PowerShot digital camera to take low-resolution and often hand-held videos of old, worn-out CRT TVs that I will spend hours transcoding and uploading to YouTube after each tournament and gathering, all in an effort to keep a record of one of the more competitive Soulcalibur scenes in North America.

That’s my excuse for how terrible the quality of these videos will be. Feel free to ignore them if your eyes are offended by 480p and scratchy AV wiring.

I am playing one of the worst characters in the game, Rock. There are many reasons why he is a terrible character, and most of them will be on display in this match, but there is one way that is particularly important. (Being honest, that’s also an excuse. The truth is that I am also a poor player that didn’t put in the work, though in my defence, I owned an Xbox while all tournaments were played on PS3s, so it wasn’t very easy for me to practice.) He has really bad frames. Just awful. The worst in the game. He actually has a move that is unsafe (can be punished with guaranteed damage) when it hits.

Anyway, this Siegfried player shows up. Siegfried is the guy with really big sword. There is nobody I play regularly who uses Siegfried, so I have very little experience against the character. Certainly, I have very little experience playing against competent Siegfrieds. I am an aggressive (read: mindless) and tenacious (read: stubborn) player, and Siegfried is a character that depends a lot of ambiguous and odd setups that prey on players with bad character knowledge. And who play characters with bad frames. As should be expected, this does not go well. Pay particular attention to Siegfried’s a:g:A attack, the high, sweeping horizontal slash that is signalled by the character flashing white. It is an i18 high attack (I should note that i18, about 1/3rd of a second, is within the range of moves that can be reacted to, if the player is paying attention.) that does moderate damage and knocks down. Nothing about the attack  is particularly remarkable, except that it is +2 on block for Siegfried. A whole +2 frames may not seem like very much, and it isn’t. That’s exactly my point.

If you don’t want to sit through all of that, and I wouldn’t blame you, here are some relevant time stamps: 2:01, 2:22, 3:28, 3:50, and 5:02.

He uses the a:g:A to set up a frame trap and I fall for it every time. Why does it create a frame trap? Because the +2 frames that Siegfried has after an opponent blocks the move means he can knock 2 frames off the start-up of any move he does next. Siegfried isn’t exactly the fastest character in the game, so that still limits his options, but consider this: the moves I am hit by after blocking the a:g:A are a 3[B] (the upward sword strike that launches me into the air at 2:01) and 66KA, the headbutt and hilt smash that hits me every other time. Both of those attacks are i17, which again, is not that fast. On the other hand, Rock’s fastest attack is i14, and also has no range, so it’s useless as a response. In general, Rock’s fastest attacks are in the i15 to i17 range. That’s where the math comes in. His i17 move would lose to my i15 move and trade with my i17 move if we were both in a neutral state, but with his +2 after the a:g:A, he will always come out ahead. In fact, you can see that I keep eating Counter Hit attacks (the red flash). A Counter Hit (CH) happens when a move interrupts the start-up of another move, and Counter Hits do more damage, as well as having other properties like stuns, knock downs, and launches. It’s been a few years, but I am pretty sure that I was trying to respond to the blocked a:g:A with a 6k (The 6 here is a directional input. Look at your keypad and you can see how the notations work. The player is always assumed to be facing right.), a knee attack which is i17, making it at least 2 frames too slow.

Why is it a frame trap again? Because it’s not always easy to recognize +2 frames, but it is always possible to suffer from the deficit. That is one major reason why character knowledge is so important. If I had known before hand that Siegfried’s a:g:A was +2, I would have known to not make the mistakes I did.

Another player at the tournament, who is both more canny than I am, and is also using a much better character, witnesses my defeat, and then has a match against the same Siegfried.

Notice his response to the a:g:A. He does nothing. Why? Because he knows there is a trap, but not exactly what it is. Unlike Rock, Sophitia has moves fast enough to beat an i17 attack, even with that 2 frame disadvantage, but that would require knowing that it’s exactly a 2 frame disadvantage. At the end of the day, frame traps are traps, and must be fallen into in order to work, which is why many of them are on the level of gimmicks, and the best are only used by players who have set them up within the match and have worn their opponents down mentally. If the defending player chooses not to do anything, the attacker will lose out if they try to press their advantage too far. Turning an i17 move into an i15 move doesn’t change that they will still be negative on block. If Siegfried does a:g:A and then 66KA while his opponent does nothing, he is now at -8 himself, or potentially even worse off since 66KA ends with a high attack that can be crouched under and counter-attacked for free. Abusing frame traps is something that is done with careful consideration.

That’s not to say that it’s impossible to work out such a frame trap without knowing the frames of every move involved. It can be done with the right training mode settings and a lot of free time. But it’s not nearly as practical as seeing a move with a lot of recovery time and then throwing something out to try to punish. As I said already, if a frame trap is that easy to work out on the fly, then it’s not a very good frame traps, or will only work on players who had no idea what is going on. Cough.

There are other uses for frames in fighting games. All sorts. I just picked a very prominent one that demonstrates how small frame advantages and disadvantages matter. As the frame gaps widen, their uses become even more important, but also more obvious.

Say the Siegfried goes for the 66KA after the a:g:A is blocked. Now he’s at -8, and his opponent gets a free mix-up. It’s only a mix-up because -8 is not enough for most characters to punish. Even very fast attacks are mostly in the i10-i11 range in Soulcalibur. Sophitia definitely doesn’t have anything that fast. In 3D fighting games nearly every move will be negative on block, or can be ducked under, jumped over, or sidestepped, and there are few (if any) projectiles and ways to cancel moves and hold onto an advantage, so it’s not really possible to have block strings like many 2D fighting games have. Players are not eager to simply trade blocked attacks, because they will end up coming out behind, not ahead. What does Sophitia do with that +8 frames when she can’t do anything fast enough that Siegfried won’t be able to block or dodge out of the way? She could try for a throw, which is i17. No attack Siegfried has could interrupt the throw, and if he chooses to stand and block it will connect. However, he still has enough time to duck underneath it, or possibly even back-step or sidestep. So maybe Sophitia will try to anticipate that and throw out a BB attack, which will hit Siegfried if he crouches or back-steps, or a low attack that will hit if he stands and blocks. This is your typical mid/low/throw mix-up, and is a cornerstone of 3D fighting games. Less important than the mix-up itself is that at -8 Siegfried has no choice but to react to the next thing Sophitia does, and unlike in a frame trap situation, he knows it.

Frame data is sometimes useful in combos, but less so in 3D fighting games than in 2D fighting games. Most combos in 3D fighting games will involve launches and stuns, and strings of attacks that cannot be separated (AA is a natural combo, but doing a single A will leave the attacker at such a disadvantage that they will not be able to connect it with anything else, and will probably be open to a mix-up.) so the frames involved aren’t very important. It’s more a matter of going into training mode and seeing which moves and strings can connect with other moves and strings before a stun ends or the other character hits the ground, and not specifically what their frames are.

There are other ways that fighting games can use frame data, but they’re also designed to be obvious even without the exact numbers, though the numbers behind them can be revealing. A good example of this is the Guard Break or Crush that was first added to the Soulcalibur series in Soulcalibur 2. These moves are always accompanied by flashes of blue lightning, and have block animations that are exaggerated to make it obvious that the defending player is at a disadvantage. When they were first introduced, Guard Breaks had huge frame advantages on block, something comparable to a Guard Impact, which is in the ~+20 range. Landing one against a blocking opponent meant the next attack was going to hit, unless they defend themselves with a risky Guard Impact. Other 3D fighting games have similar moves, such as Akira’s 6P+K+G guard break in Virtua Fighter, which has long been one of his best moves. In an odd twist, careful examination of frame data revealed that the Guard Breaks in Soulcalibur became much, much worse over time. They still had the blue lightning, and there was still that exaggerated block animation, so it was easy to think otherwise, but almost nothing is guaranteed in Soulcalibur 4 after a Guard Break. Some of them are in the +7-9 range, while many others are as low as +1-3, and there are even a couple that are negative on block, despite how they appear on the screen.

Going back to generalizations, players will change how they categorize a move depending on the circumstances. A move can be safe or unsafe when they’re worried about it being blocked; it can be disadvantage, neutral, or advantage on block when they’re concerned with what their next action should be. There is also the matter of whiffing moves, or moves that miss completely. Earlier I said that when a move is blocked there is both a recovery time for the attacker and block stun for the defender. If the attack doesn’t connect, then there will be no block stun. That is why positioning and movement are key elements of 3D fighting games. There is a tactic, often called zoning, that is specifically built around moving in and out of an opponent’s effective attack range in order to bait them into attacks that will miss, and then punishing them for it. Manoeuvring around attacks that would be safe if blocked can create opportunities that wouldn’t exist otherwise.

As a final example, I’m going to mention the common 2P, 2A, etc. low punch move that most characters in 3D fighting games have. It is noticeable for being designed to fill a hole that is specifically built around frame data. A 2P (I’m just calling it that because that’s what they are in Virtua Fighter, the game where I first encountered them.) does almost no damage, doesn’t combo into anything, and has low range. It’s easy for the casual player to dismiss them for that. But they still fill an important niche: the +frame panic button move. Their damage and range are not what makes them useful, their speed and frames are. When a player is feeling pressured, under constant assault by their opponent, sometimes they just need a move they can throw out that will give them enough +frames to take away that momentum. It will have enough +frames to stop an aggressor cold, so that they are forced to stop attacking or will lose out to the defender’s next move. These attacks became so prevalent and integral to how some 3D fighting games were played, that over the years they have been heavily nerfed, and entire strategy discussions centre around simply how to deal with players who rely on 2P moves to frustrate opponents. Virtua Fighter in particular has  been very harsh, to the point where they give characters moves specifically designed to beat 2P spam.

I could go on for much longer on the minutia of frame data, but I fear it will become progressively less interesting. I do hope that I’ve made clearer what frame data is and how it is often used. My next subject will be how frame data is discovered, which is becoming a lost art now that most new fighting games publish their frame data in strategy guides before the games have been officially released.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s