A3: damage scaling factors?


does anyone know precisely how the damage is scaled in A3? i can’t even tell if it’s linear or a (negative) exponential type, and i have no idea how the rounding works.

i’m guessing it might be in the AAA3 book…


say what?


i think he means like does the damage decrease as the lifebar goes down (a la CVS2)


what I know about damage scaling in A3:

Some attacks (generally just specials and projectile supers) scale by distance traveled

Block damage doesn’t scale, except sometimes due to distance scaling

damage to GC meter doesn’t scale, same exception as above…this is one of the reasons shin akuma is so good…fireballs can be air blocked, but they will chip you down and destroy your GC meter.

Yes, I believe you take less damage as your health decreases, but I assume it’s the same thing as Marvel, where it only takes effect if you have 33% or less life

damage scaling USUALLY starts at a certain number, though I don’t remember what it is. probably 7 hits or something.

All hits in while V-ISM is activaed do less than they would normally. I don’t know if this applies to block damage, but it applies greatly to damage to the GC meter

dizzy also scales, in that once you have dizzied someone, they get harder and harder to dizzy each time. That’s why there are no re-dizzy infinites in A3. That and the flipping.

Air throws may or may not damage scale, depending on the throw.

Weak attacks scale more than mediums, which scale more than hard attacks, which scale more than specials.

That’s all I can think of, off of the top of my head. To answer what I believe was your actual question, I don’t know by what percent the damage continues to drop until it reaches it’s minimum, but I’m pretty sure it’s a consistent percent (or a consistent number for each group…normals, specials, and supers).


Re: Re: A3: damage scaling factors?

i dunno if you are talking about something else, but the gc meter does scale in a poke pattern. in a successful series of uninterruptable pokes, the gc damage gets weaker than what it would be normally if that individual attack was by itself. for example with, say, uhm, x-ryu. do f+fierce(2-hit) and see how much gc damage it does. now do a cps chain with like c.short f+short+fierce, fierce will do less guard damage since you did a successful poke string. i dont know full details on what the rules are for gc scaling but it definitely has to do with poke patterns. and of course, it works the same whe nyou do a gc vc…

if any info is wrong, please correct. this has always been off my own experiences with the game that i learned on my own


isn’t damage kept the same in a VC? i was sure it was. i completely forgot about the damage reduction due to distancing, but i’m sure it’s confined to projectiles (unless Sim limbs prove me wrong. i haven’t tested). i didn’t think damage was lessened as health decreased (ala CvS2).

anyway, i was really just talking about how a fierce after 2 hits does a lot more than a fierce after 99 hits. i’m also guessing it’s a constant % rate. i’m guessing it might be on a jap site somewhere…


Guard Meter damage does scale, and the way it does so is a much simpler than Life Meter.

  • each move does a certain amount of Guard Meter damage
  • doing consecutive moves approximately halves the original Guard Meter damage that that move would do, up to a minimum of 1 slice; the fourth consecutive move, therefore, would only do 1 slice of damage per hit (consequently, multi-hit moves, such as Akuma’s FP red fireball do 3 points of Guard Meter damage)
  • “consecutive” here is the amount of time between hits that the Guard Meter starts to regenerate (around 3 seconds)

example: If you do Akuma’s FP against a blocking opponent, it would do ~ 1.5 GM blocks damage, then if you do it again it would do ~ .75, then ~.3, then 1 slice. If, however, you wait ~ 3 seconds between hits, you can get ~1.5 blocks of damage per hit.

I think damage scaling damage scaling starts after the first move hits; this is most obvious in VCs, where a lot of specials get used. Take, for instance, Akuma’s standard VC: JP DP->fireball->demon flip… as opposed to JP DP->RK->demonflip… The damage on the demon flip in the first combo is really scaled, but not as much as the second combo. Note that both combos have less than 7 hits (as TS suggested). This example also illustrates the scaling brought about by using a lot of specials – the second combo didn’t have a fireball in it, and so does more damage.

The damage (air)throws that contribute to dizziness (Akuma’s, plus another character I can’t remember) scale when used within a VC. For example, if Akuma ends his long combo with an airthrow (assuming his opponent doesn;t flip) it sould only do 1 point of damage. But if Charlie (whose throws do not contribute to dizziness) ends his combo with a backbreaker, the damage is full.

Dizziness scaling: If it takes 8 consecutive FP/RKs to dizzy a character, it would take 9 FP/RKs to dizzy him or 10 FP/RK’s to reddizy. This value increases by 1 hit for every dizzy performed, making dizzies harder and harder to do. Some moves have unusual “dizzy values”, such as Ryu’s hurricane kick (which I think counts as ~ 1.5 hits), or Gief’s up jump+FP (I think counts as 3 to 4 hits).


in the GC scaling example, those aren’t blocked combo strings are they? so scaling would occur if consecutive moves were made within some time interval, regardless of whether they were interruptable or not? sounds like something you could build a strategy around…

it probably looks something like

X = min {D, p}

D = d*[k^(n-1)][hm/q]

D: scaled damage
d: original damage
k: damage scaling factor. 0<k<1
n: number on combo meter. n>1
p: minimum damage an attack can deal.
h: character specific defense rating. 0<h<2
q: damage due to ism (X: 80%, Z: 100%, V: 100%)
m: damage due to mashing. 0<m<1

eg. in a Newb Hadorave (hadoken, hadoken, hadoken…) vs a mashing X-Zangief, IF we assume a hadoken normally does 8 points of damage, the scaling factor is 90%, that Zangief takes 85% of the full damage (because he’s Zangief), and that mashing drops damage taken to 95%, then the 10th hadoken in a combo would do 8*[0.9^(10-1)][0.850.95/0.8]=3.12842 points of damage. after the 19th hadoken, each would be doing p points of damage (p=1 probably). if we knew how much life Zangief had at the start of the VC we could say how close he is to dead by the end of it if we took the meter into account.


jsut changed it to plain old “D”.


That’s the theory.

So yeah, I pretty much hate math.

But anyway, here’s something I typed up the other afternoon, when I couldn’t get online. I think Mondu pretty much covered it all though:

OK, I don’t have my Saturn anymore, but fortunately I have a tape of me experimenting with some stuff and playing around vs the CPU, from about 3 years ago.

Anyway, from what I can tell-

-Damage to the guard crush meter DOES scale during V-ISM for sure, but I don’t think it resets (without a sweep, or something like that). So for example, you’re Ryu, and you do the [fireball/hopick/fireball/c.forward] VC, and it’s blocked. If you stop halfway through, and give the opponent time to breathe,and then start it up again, you’re still going to get minimal damage to the GC meter. I don’t know if it can be reset if the opponent stops blocking, and then starts again. {edit: it won’t be. However, being hit prevents your Guard Crush meter from rising, so mondu may still be correct.}

likewise, if you did the first part of the VC and it connected, and then you screwed up and they started blocking, you would get the normal starting amount of guard crush damage. So it’s by how long they’ve been blocking, and not how long your VC has been active.

-There was definitely evidence that gc damage scales during chains, but it was sort of inconclusive, as it was only done with Gen, and he’s sort of abnormal.

-It looks like you take less damage after flipping from something. Of course, you reset damage scaling when you do this, so it’s still not a good idea.

-I honestly don’t remember if projectiles damage scaling by how long they’ve been on screen affects the GC meter as well. But there are a couple of other 1-hit specials (I know one is Rose’s Soul Spiral, I think Gen’s c.FP in punch style, etc) which scale by how far you are from the person. They don’t scale in actual damage, but in damage to the GC meter I think they do, and maybe in block damage, if applicable.

-There are no situations in which block damage scales like damage will, AFAIK.

Now that you mention it, the 7-hit figure was taken from one of the Marvel games. Brain fart. However, I’d be willing to bet that all damage scaling will start at a certain number with all characters (though I’m sure there moves that are exceptions and which scale early). I’m guessing it’s after 4 or so hits, since combos shorter than 3-4 hits don’t seem to scale.

I’d also like to point out that the combo may have done more damage just because normals tend to do more damage than projectiles. But it’s an interesting thought that the more specials that get used, the more damage scaling increases.


I always thought that v-ism combos will start to scale when the meter reaches 50% of what you originally started your vc with.
I’ll do some research.


the math up there basically says that as the number of hits rises, the net damage per hit (compared to original damage) falls. it assumes that scaling takes place as soon as the combo meter shows up. in such a case, a j.FP s.FP 2-hit combo would do less than if they hadn’t comboed, theoretically (1+1 < 2). but i’m sure the engine has a rounding system as well, and if the decrement in damage is “insignificant” it would be rounded up to the full sum (1+1 < 2 ~ 2. so damage counter displays “2”).

i also figured that “VC JP.DP > hadoken > demon flip” does less than “VC JP.DP > RK > demon flip” because hadokens did very little damage in this game relative to normals, and this one is even a far one. so the RK would deal good beefy damage and cause the flip to scale, while the hadoken would cause the same amount of scaling but with less than half the damage of an RK. end result, both combos deal same number of hits but different damage because the hadoken did very little damage, not necessarily because it was a special.

random question; anyone remember that part on one of Jmmy’s vids where V-Ryu hadoraves X-Zangief and GCs him? the guard meter depletes rapidly at first, then more steadily, then it’s almost stagnant once 2/3 of it’s gone, barely decreasing. then suddenly somehow it’s ALL gone and i have no idea why. i’m not aware if it was doctored like the Z-Guy 100% so i’m not sure what’s going on. anyway, i’m thinking maybe it was a GC reset type and Ryu left a gap to blow that remaining 1/3 out of the way, but everyone seems to agree that GC damage remains scaled even if a string doesn’t lock your opponent in blockstun.



Actually, it does matter if it was a special. In the example I gave, the demon flip does less damage if it were preceded by a fireball as opposed to being preceded by a normal.

I haven’t seen this vid, but I do know that when you try to GC the CPU (not the training dummy) this phenomena does happen (first few hits take out a lot of meter, then slowly depletes, then suddenly depletes by a huge margin). For some reason, attempting this on a training dummy, or a human being just blocking, doesn’t cause this phenomenon to happen. I’d like to know if that X-Gief in the video was human controlled or not, and if it were human if it was mashing or something.

My guess is that attempting to move out of actually causes a reset of sorts.


I’m a bit surprised no one has hacked the game data for stuff like this. There are probably a lot of tech savvy people at SRK.

Until then, just manually plug in the numbers and use interpolation or Lagrange polynomials. I think if you are careful about starting life and character defense, this will work well. Guard gauge would be a lot harder to do since as far as I know the game doesn’t list numbers for that.


Aa Sfa3

I’ve a page from the All About SFZ3 guide, about Damage Scaling but it’s all in japanese. I can give the link if someone is interested.


Uh… I’d be interested in that, my mate speaks japanese


Ok, I’ve uploaded this page here : http://ftproductetard.free.fr/veja/23.jpg
If your mate can translate some infos from this scan, please post them.

If you’re interested, I’ve also other scans from the AA SFZ3.


No personnal offense , but that’s maybe one of the most useless thread I have been able to read past in the week .
Seriously , do any of you make an analysis about damage scalling like a uncertain situation in a chess game ? …
For your information,because I am not a monster , damage is reduced each time by a part of the real damage it has to do clear isn it ? :wink:


You need to brush up on your reading comprehension buddy.
I know you’re from France, but don’t talk if you don’t really understand the subject matter.
You seem to have a habit of responding to dead topics wayy too late, and contributing nothing in the process.


I have contributed to help people in the ryu questions since it’s my main character ,and also to a rolento thread which is very nicely written .
Also , don’t critisize somebody who is stronger , and know the game better . Always stay humble . It’s not kind if I said damage scale was useless to know , and for that I feel sorry , but don’t speak about me personnaly .
You must be very jealous to answer something in that way .
Anyway , see you :wink:


Wow. Someone’s pretty high on themself.
Aren’t you the same dude who said that IQ is the only thing that matters for being good at fighters?
And you’re assuming a lot when you say that you know the game better than me.

Wonderful advice you’re giving there.