The Role of Damage Scaling

Every time i release a new combo video, especially for SF4 or SSF4, someone comes along to declare that “damage scaling makes it worthless.” I have a number of objections to that claim. First, damage scaling makes sense. Second, damage scaling only matters in combo videos when the author says it matters. Third, combo videos and combos in general are a relatively small piece of the complete fighting game puzzle.

Balancing a dynamic combo system is extremely difficult, partly because players demand substantial character variety. I’d be surprised if even 1% of the damage scaling complainers have ever tried drawing up their own hypothetical combo framework as a thought experiment.

We can all agree that practical touch-of-death combos are generally detrimental to game balance. There need to be fair limits on how badly a single minor mistake can punished. The price of whiffing c.MK should not equal the penalty for whiffing an ultra. Furthermore, characters possessing the best mobility and the most offensive mixups probably shouldn’t have the highest damage potential.

How are such considerations governed without confining players to pre-programmed combos? SF4’s solution is a sharp reduction scale which magnifies the first four or five hits, then blurs out the remainder. This affords designers and testers the luxury of managing a reasonable subset of possible sequences while holding the rest to a combined average of less damage than one hard attack. Thus it becomes much easier to identify which characters can get ahead of the curve, then adjust their damage values and frame data accordingly.

All they have to do is make sure nobody can jump out to massive damage in under five moves, particularly not following a safe light attack. For example, C.Viper can combo multiple fierces, so they should inflict below average damage. She can also shorten her normal move recovery times via Thunder Knuckle cancels, so her frame data should restrict which pokes can combo into hard attacks. Obviously her underlying potential for infinites must be addressed as well.

 
ComboVid.com - Fighting Game Combos, Tutorials, Matches, Screenshots, and Strategy

Assigning additional reduction penalties to Focus Attacks and ultra moves was another critical balance decision. Landing a lvl2/lvl3 Focus Attack produces the same combo opening as causing the opponent to whiff a fierce Dragon Punch. You’re rewarded with basically any ground combo you desire, with ample time to choose.

The difference is, it’s much easier to connect that Focus Attack than it is to bait a fierce uppercut, as evidenced by their relative frequency in top-level matches. There’s no way anyone deserves full damage for comboing lvl2 Focus Attack into ultra, considering how they “earned” that meter by getting beat up. Adding extra hits inbetween doesn’t change how conveniently the opening was created.

Remember, combo skill is only 10% of what it takes to be good at fighting games. I’ll never understand why people keep saying “heavy damage scaling doesn’t reward skill.” That SF4 actively de-emphasizes the role of extended combos is a step in the right direction. Fancy combos are for showing off, but solid fundamentals are (and should be) more valuable.

As for combo videos, damage scaling has no bearing on style. The same is true of most technical achievements, such as comboing two Jaguar Teeth or finding different ways to juggle into cinematic Corkscrew Cross. The only time damage matters in a combo video is when a specific combo is clearly geared toward maximizing damage. Otherwise the author doesn’t care, and neither should you.

29 thoughts on “The Role of Damage Scaling

  1. Maj Post author

    A couple of people asked me to write this, and i was hesitant at first, but it did give me the opportunity to discuss some topics that i’d been meaning to address. Basically, SF4 gets a bad rap for its heavy damage scaling and input leniency, but i don’t see any problem with either of those design decisions. SF4 matches certainly don’t run the longest of the games we’ve played.

    Execution-wise, SF4 isn’t that much more lenient than SF3 was, and it’s still difficult enough for newcomers to have to spend hours practicing special moves in Training Mode. It’s definitely a step in the right direction though, in terms of striking that balance between welcoming new players and placing enough value on clutch execution.

    My general stance on input leniency is that more is better – as long as it doesn’t get in the way of itself. Admittedly, a few SF4 characters have that problem, but it’s not nearly as bad as people make it sound.

  2. BackForwardPunch

    @Maj
    The damage scaling is not a big deal in my opinion, I think it’s a necessity. As it is, full ultra bar usually gives pretty crazy damage if you hit, and I think that’s a pretty steep reward for getting beaten up the entire match. The input leniency bothers me in some instances though, With the SRK motion leniency, there are a few characters that is starts to get annoying for me, in particular, Abel with go for air throw instead of the COD or Ultra on me pretty often, and Makoto will do the uppercut rather than the Dash punch. But like you said, it’s sort of minor.

  3. Maj Post author

    Well, input overlap is kind of a big problem for some characters. I’m not gonna say it doesn’t annoy me when it costs me a match. But that’s the unfortunate catch-22 of simplifying execution in a precise genre. I still think increased input leniency a good thing overall, but certain characters could use some tweaking on a case-by-case basis.

  4. jamheald

    @BackForwardPunch
    That’s not really leniency just the basic memory of inputs. I’m going to presume you were holding forward before the cod then did a qcf so that’s 6236 as you can see there’s a 623 in there, a dp motion, and since dp motions take priority over qcf (if it didn’t I think there would be more problems) you get the dp move. As for ultra it’s probably you pressing punch to soon so you do 236623 then punch, in there there’s another 623 as you can see so since you’ve not completed any other motion you get a dp move. As you can see in both of these there’s no shortcuts, both have a 623, just input memory and overlapping inputs. Without memory you would have to time your button presses perfectly and can wave goodbye to dash charge ultras, and there’s always overlapping inputs in a game.

  5. N00b_Saib0t

    thanks, maj, good read.

    i think jamheald is dead on, btw. if you’re holding 6 then try 236 you get SRK. IMO this makes input requirement MORE precise. if you’re walking forward and get the opportunity to poke with c.mk and cancel to hadouken, you must now put a slight pause when going from 6 to 2 then finishing the command or you will get c.mk->SRK instead.

    when people SRK FADC ultra as ryu they complain about “shortcuts giving SRKs”. its true that the SRK command is very lenient in SF4, but most of them get it from not going back to 5 (neutral) after 66 (dashing) out of the focus attack. i complained about the same thing when i knew NOTHING about the SF4 engine, but realized (read:had it pointed out to me) that you have to return to neutral or the game recognizes 66 236236 as 66236236 and input leniency always turns 6236 into a SRK. the few times i do play ryu and rufus, i dont get nearly as many EX SRKs or snake strikes when trying to FADC to ultra.

    the other thing that bugs me is this mentality of “getting rewarded for being beat up” when it comes to ultras. no one said this about K-Groove, why say it about the revenge meter? both can charge with a form of “parrying” (JD for K-Groove, focus absorption for revenge meter) or by taking damage, and even your super meter in most games has charged to some degree by taking damage (the only obvious exception is ST, but comparing SSF4 or CvS2 to ST is like comparing the most goodest person ever to live to jesus christ and asking which one is more holy). i dont see a problem with the revenge system or how ultra’s work. i did think some dealt out WAY too much damage in vanilla, but SSF4 seems near perfect IMO.

  6. Tarnish

    Maj :
    My general stance on input leniency is that more is better – as long as it doesn’t get in the way of itself. Admittedly, a few SF4 characters have that problem, but it’s not nearly as bad as people make it sound.

    Eh, the input buffer makes the game feel like goop. Players saying “Well if you want this do this” are missing the point of the complaints. I know how to perform a Shoryuken. The game isn’t doing me any favors by making it easier because I already paid those dues… if you haven’t, why am I being told to practice when it wasn’t my issue in the first place? New players, right? Accessibility? Okay, that’s fine.

    I can definitely see SFIV’s changes making things more or less the same, especially with the fact there’s always going to be a training component. The way I’ve come to see it is that it’s also there to accommodate new character types than a game with powerful projectiles and invincible Dragon Punches would allow. That’s a good thing on paper (don’t talk to me about Rufus, dear Capcom this character is dumb). However, reading players talking about Street Fighter in keypad notation talking about practice… I don’t think that works. No one gets anywhere with that argument, it goes both ways. I think we can all agree that the games are different and practice is a necessary component. If it were just an issue of practice, then DPs would still be the way they were in past titles where they were most difficult to perform.

    My issue isn’t even with inputs at this point… it’s with my buttons. I get that this was already stated, but hell, I’ve got time for a sob story. The sliding input frame buffer or Plinking for links causing constant overlap on things like Supers/Ultras/EX Moves gets in the way of things I’ve made intuitive, speed, precision, and reaction. Something I did SEVERAL FRAMES ago can now come back to haunt me?

    Hell, that’s not even the worst part, it’s the way pressing buttons seems to have more execution levied than simple motions. The motions were never really the issue save for extreme circumstances. I recall one match I lost a round because I went for jump MP into Nightmare Booster and I got Knee Press Nightmare instead near the corner. Not only did I get fucked up for saving my meter against Balrog, but he recovers in time to Headbutt xx Ultra 1 me for my troubles, then run away and sit on the life lead.

    I looked at the inputs after the match and the game read them out as:

    KKK
    K

    If I get the following as Guile:

    PPP
    P

    Guile will still do Sonic Hurricane. He will not do EX Boom or regular sonic boom. That seriously blows a fuse in my head and makes me want to spend my time playing something else. Maybe I’d like to just keep my macro’d KKK 7th button in tournament so this shit doesn’t happen, but whatever. Most Viper players actually just say “just don’t have full meter” in her case. I think Capcom can do better than this as a solution. Just want to follow up in agreement.

  7. jamheald

    N00b_Saib0t :
    i complained about the same thing when i knew NOTHING about the SF4 engine, but realized (read:had it pointed out to me) that you have to return to neutral or the game recognizes 66 236236 as 66236236 and input leniency always turns 6236 into a SRK.

    You don’t actually, you’re probably just making yourself time the punches later when you think of returning to neutral which makes sure you don’t hit them at 3 and thus get an srk, ultra inputs take priority over dps.

  8. Maj Post author

    jamheald: Memory is leniency. What makes it easier to perform SRK motions in SF4 is the way the game remembers inputs for longer periods, which allows you to do them slower. But then certain characters with fast dashes simply can not perform QCF motions coming out of a dash. A Makoto player was complaining about that on SRK and there’s not much anyone can say about that problem beyond acknowledging it and telling them to change tactics. If ST had this much input leniency, N.Ken would be absolutely unplayable.

    N00b_Saib0t: Actually everyone did start calling K-Groove “scrub groove” but that didn’t last too long because roll-canceling and A-Groove rose up and made K-Groove’s life miserable. Before that, K-Groove players simply didn’t have to work as hard for wins. Revenge meter has some obvious problems, but thankfully most ultras can’t be comboed directly from normal pokes. That’s really the only saving grace of the whole system in SF4’s case.

    Tarnish: Yeah, SF4 supers cause problems. What’s worse is they made specials cancelable into supers, so you have to be even more precise. In SF4, they probably could’ve solved that (at least partially) by moving supers to a different button than ultras, but then SSF4’s dual-ultra system would’ve brought it up all over again. But there’s still ways of coping with these issues. In your example, you probably rushed Bison’s ultra command and finished it before he landed. If you do it on the ground, KKK~K should never give you super instead of ultra.

  9. N00b_Saib0t

    @maj – interesting. i actually didnt get into CvS2 until after RC was discovered, i got into it right around the beginning of the “should RC be legal” debate actually. i guess i just missed out on K-Groove’s early dominance.

  10. jamheald

    @Maj
    Ah, sorry I always associate it with input leniency in terms of motion, as in not completing the actual listed motion gets you the move.

  11. Tarnish

    @Maj

    I’ve practiced that a great deal in vanilla. But I guess it was the first time I ran into it in a match with full super. I was playing it pretty clean, too. Dictator vs Boxer gets pretty retarded. I got the win, but shit like that doesn’t make me go “Oh, man, I’ll just…” I think what you just told me makes me want to choke Ono.

    Sliding input and link emphasis is not fun game design to deal with. I hit 3 buttons, and trip over someone else’s training wheels. Goodie.

  12. jchensor

    This is completely pointless to any discussion, but I love how you called the plural of the move “Jaguar Teeth” instead of Jaguar Tooths.

  13. Maj Post author

    N00b_Saib0t: CvS2 had a fairly dynamic evolution in that regard. It started out with everyone playing C-Groove to emulate CvS1 or picking N-Groove ’cause running was fun. Then it became apparent that K-Groove was legitimately good despite it’s simplicity. All the while, A-Groove kept getting more and more sophisticated, clawing its way to the top. I remember when we all considered A-Groove to be the worst in the game, or at least tied with S-Groove. It was a very gradual process, which i thought was cool to watch.

    Tarnish: You’re right, tripping over training wheels is exactly what happened. There are places where SF4 lies about whether you’re being accurate or not. When it comes to landing recovery, there’s a huge buffer that didn’t exist in prior games. SF4 makes you think you’re pressing the buttons as early as possible upon landing, but in reality that would’ve given you nothing in older games. The buffer also doesn’t make the move come out any sooner, so you can just learn to do the command later/cleaner without any penalty. In other words it’s not taking anything away from you. It’s just giving you a safety net that you can get tangled up in. But if you’d been practicing this with full meter in Training Mode, i don’t think you would have too much trouble adapting to it. (Not that i wouldn’t make custom adjustments if it was up to me, knowing that Bison’s super is practically useless anyhow.)

    jchensor: Haha yeah that was a factor in choosing that example. I was like, “When am i ever gonna get to say Jaguar Teeth again?” Btw i like how that’s what finally got you to register. On an unrelated note, Go Lakers!

  14. ShadowXSnake

    Capcom need someone like you on their payroll as a special advisor to combo potential or something, you really have a great way of dissecting and analyzing the technical aspects of combos, but you never seem to forget the importance of keeping combos fun.

    That said, I always thought it would be nice if combo scaling followed a bell-curve so that single attacks do very little damage, damage goes up for subsequent hits, and then dives back off like the current system. That way, consistency in combo execution would be rewarded, but slow poking games and extended combos would be somewhat discouraged.

  15. Tarnish

    @Maj

    Training mode shall henceforth be referred to as “The Salt Mine.” The sad thing is that I’d probably use Psycho Punisher before Nightmare Booster against Rog, I was playing casuals that match I’m referring to. I think those critical execution flops are so haunting, and they’re still happening to even the best of players, so I can at least say I’m not the only one.

    Still, I didn’t need to FADC anything, I kind of asked the question at the end of the match “How did I have all this meter in the first place?” The guy just wasn’t that good.

  16. platinum

    Damage scaling is really dependent on the game engine, imo. It’s fairly easy to do combos in SSF4, so the damage scaling used complements SSF4’s pace. I just wish that there was more of a “swing” in SSF4’s pace, meaning that comebacks can happen more often. I really don’t feel that with SSF4 as much I do with 3s. Probably, its because of “parrying”, but damage scaling plays a huge role in swing states.

  17. mr. newbie

    i don’t know if its a scaling issue but kens target combo is basically useless because it does no more dmg/stun than a regular fierce and only serves to scale the next hit.

  18. BackForwardPunch

    Thanks for the responses, guys. But I think as it goes on it sort of explains my point of view in the comments, I don’t think it’s as much practice as the way the game is made, since I never have this problem on other Street Fighter games. Of course, none of this is discouraging against SSFIV, it’s still hella fun, just would have been nice to have those few little tweaks. Like Maj suggested and a few others, by now I’ve learn to do the motions the way the game wants them.

  19. shaku

    ShadowXSnake :
    That said, I always thought it would be nice if combo scaling followed a bell-curve so that single attacks do very little damage, damage goes up for subsequent hits, and then dives back off like the current system. That way, consistency in combo execution would be rewarded, but slow poking games and extended combos would be somewhat discouraged.

    No offense, but in a game like street fighter, that would be a terrible idea. It would totally cripple poking (footsies in general) and grappling characters and greatly reduce character variety (all good characters would be very similar in playstyle).

    I personally think the current damage scaling system is great. Without it, you could easily do half damage combos and games would end too quickly, though maybe that’s just my preference. But it also helps to balance the cast.

  20. Pokey86

    Personally i am amazed such a simple damage scaling system has workes so well in SSFIV Here’s all the damage scaling system in SSFIV boils down to

    – Scaling is based on moves not hits

    – scaling reduces as

    %

    100
    100
    80
    deducts 10% for each hit from here on.

    – Focus attack scales twice. Example if you hit ultra as the second hit of a combo you get 80% Focus, where as most normal attacks would give 100%

    – Ultra has the same effect as focus.

    – Super doesn’t double scale

    With such a simple system they’ve made confirmable combos naturally weaker… Punish combos naturally stronger, focus combos instantly scaled (as well as 70% ultra after focus) & lastly it’s simple enough that with characters like Sakura or Seth you can judge how much stun you’ve caused & adjust your mix ups accordingly.

    Only problems with the scaling system is that focuses mid combo absolutely destroy damage output, making the mechanic of fireball -> Focus attack (hit) -> Ultra a nion useless trick.

    Personally the only “fix” i would do is make EX Focus attack not double scale.

  21. jamheald

    Pokey86 :Personally the only “fix” i would do is make EX Focus attack not double scale.

    I’d like that. On the topic of leniency it makes you have to have real tight inputs on sj kunai with Ibuki when you have super.

  22. ShadowXSnake

    @Shaku, I’m not asking to make it game breakingly weak to drop in pokes without follow-up, just slightly weaker overall. To be honest, as incredibly skillful as it is to run a good footsies game, I always find that it can get very boring to watch when a match devolves into a series of single normal attacks and the occasional shoryuken. I find it MUCH more impressive when someone uses their good footsies game to land a poke, and then continue punishing with a combo, which would greatly benefit from a system that scales up THEN down.

    Grapplers already kinda fall outside the realm of damage scaling, and I don’t see how a system like this would do anything but make them stronger, provided you learn to grab after a focus attack or something. Also, I don’t get how this would reduce character variety, there’s already damage scaling, this would just be a different scale.

    If SSFIV goes 100, 100, 80, 70 etc., What I’m saying is something like 90, 95, 100, 80, 70 etc.

    Pokey’s got a point though, perhaps it is better to just keep it simple. I may be over complicating things here, it probably isn’t practical.

  23. Pokey86

    start at 90?

    Time paradox or what? wouldn’t 90 be 100 :P (sorry, gotta lol for real there)

    Grapplers are tuned with damage scalingin mind, just like, as Maj pointed out, characters with linkable fierces have a tendency to have nerf’d variations.

    Take Sakura for example, she can do 4 fierces in a combo with no meter from a jump in, her jump attacks do the typical 100/200 (damage/stun) but her fierces do 90/120… she loses 80 stun at 100% when compared to nearly every character. Now if they didn’t do this i would have a guestimate that she could stun ALL of the cast in one combo.

    Going by my typical, non heavily calculated guesses as a combo video maker.

    the difference between footsies & combos is that, footsies adjust constantly… you can practice for years on it but it’ll still vary… You practice some preset strings of combos for years & they don’t change, the second you’ve confirmed (Be it 1/2 or even 3 hits) the combo is all wrote from practice. you can never EVER say that about good footsies.

  24. Maj Post author

    Bell curve damage is an interesting idea though. Maybe it’s worth rewarding players for having the presence of mind to combo an extra move or two whenever they spot an opening, as opposed to impulsively hitting fierce or sweep or whatever. I’d be curious to see how that would affect expert gameplay.

  25. shaku

    ShadowXSnake :
    If SSFIV goes 100, 100, 80, 70 etc., What I’m saying is something like 90, 95, 100, 80, 70 etc.

    Oh, okay, that’s better than I had originally envisioned when you said bell curve. What I mean by reducing character variety is that characters who have strong poking games will become less effective (grapplers too, in your example, zangief’s spd would only be able to do 95% max damage after FA, unless your opponent was dizzied), and most good characters will have short, simple combos that fit the scaling curve, whatever it may be. Don’t know if that makes any sense, however, I do think such a bell scaling curve could be interesting in some games, I just wouldn’t really want to see it in a street fighter game. Like I said before, I think the current system fits the game perfectly, and would be happy to see it in future SF installments.

  26. Maj Post author

    Unfortunately, those exact formulas really weren’t documented before CvS2. In fact i don’t even remember CvS1 having a numerical damage display in Training Mode, so you’d have a hard time deriving it yourself.

    SFA3, MvC2, and 3S did display damage numbers in Training Mode, so you could figure it out for those, but i don’t remember seeing the formulas anywhere. I’m sure they’re posted on some obscure blog somewhere though. Maybe try asking on those specific forums on SRK?

Leave a Reply