Brawl+ Hitstun and Throw Physics

[Steel]

So I was recently watching a Brawl+ video located here: http://www.youtube.com/watch?v=QoMFFMjdR58 and noticed something.

Look at :10 (or 1:08) and notice the trajectory and position of Wolf's body when he is fthrown by Marth. That position is usually how Wolf's body should look after 12% or so when more knockback is applied by attacks.

This is how he is supposed to look when fthrown by Marth at 0%:

​

Wolf flips over and safely lands on his feet, we can see this is clearly not the case in Brawl+. This is a problem because it screws up many combos that Marth has at 0% and can most likely relate to other characters as well.

So what's the problem here? My first hypothesis was that there is a certain number of hit stun that is needed for Wolf's body to change into that position (the one shown in the vid). Perhaps the code made that hit stun line shoot over the number needed and Wolf was automatically put into that position. I did a quick frame test and whapped myself on the head for being silly, there is actually more hit stun when Wolf is thrown at 0%.

I'm guessing this is because there is a fixed trajectory similar to Dedede's dthrow when Wolf or another Spacie is thrown at low percent. Perhaps the code is bypassing the fixed trajectory? I don't have the tools to test if this applies to D3's throws as well.

Or this could have already been known and I wasted my time.. but I thought i'd bring it up.

 

[leafgreen386]

The hitstun code actually removes both Falco's and DDD's CTs against most characters.

The point a character goes into the tumble is influenced by hitstun, as you've already seen here. But what confuses me is when you said this:

Perhaps the code made that hit stun line shoot over the number needed and Wolf was automatically put into that position. I did a quick frame test and whapped myself on the head for being silly, there is actually more hit stun when Wolf is thrown at 0%.

Do you mean that there was more stun in regular brawl or brawl+?

If you mean that there was more stun in regular brawl, then I believe you've stumbled upon something interesting here.

See, when I was testing hitstun with kupo, we were using marth's fsmash, which always put pit into a tumble. We ended up coming up with the formula:

h = math.ceiling(l / 165)

Where h is hitstun in frames and l is launch speed. If you're not familiar with it, math.ceiling is simply a notation for rounding up to the nearest whole number (so 3.1, 3.5, and 3.9 all become 4.0). For future reference, math.floor means to round down.

Yet, when the hitstun mod for brawl+ came out, the default of the modifier was .4, not 165. Increasing this modifier increased hitstun, so that means that the actual formula the game had to be using was:

h = math.ceiling(x * l / 66)

Where x is the hitstun variable, which is .4 in regular brawl. So it's very possible that weak hits and strong hits divide by different numbers to determine the hitstun, and the base of x*l determines when a character goes into a tumble, and if it exceeds a certain value, then it's considered a "strong" hit. If it does not exceed that value, it's considered a "weak" hit.

 

[Steel]

Yes, there was more hitstun in regular brawl. I said in the next paragraph that I thought this may have been because there is a fixed trajectory? But i'm not sure.

 

[leafgreen386]

Yeah. That didn't make any sense to me, though. Maybe I'm not understanding what you meant, but why would a hitstun code bypass a fixed trajectory? In all cases, higher hitstun results in a sooner tumble, which means you can DI and tech. The only thing that's changed is it's gone from a weak hit to a strong hit. If the weak hit had more hitstun, then perhaps the value that the game divides the x*l value by changes for weak and strong hits.

Regardless, this just gave me an excellent idea on how to create a "DI weak hits" code (ie. sheik's ftilt or various utilts that combo into themselves because you can't DI them until stupid percents). Remove them (almost) entirely by buffing the hitstun constant, then adjust the dividing value by an appropriate amount so that hitstun stays balanced. Now to get a code that edits that dividing value...

But first, I wanna verify that this is correct. How many frames of hitstun did you measure for each?

edit: I just checked the launch speeds for marth's fthrow in reg brawl and in brawl+ against wolf at 0%. In regular brawl, it had a launch speed of 2487. In brawl+ (hitstun mod at .5175) it had a launch speed of 2719. This is certainly... interesting.

edit2: Wait, I just noticed I have the stale moves mod on... I don't see why that should make a difference, but I'll test it with just the hitstun mod now. I'm not sure if it messes with the fresh bonus or not.

edit3: Nope. No change. It's definitely the hitstun mod that's doing it. This is weird, because it isn't a proportional increase. But I suppose what's even weirder is that hitstun and launch speed are even more related than I originally thought they were.

edit4: And get this. Strong hits that were already strong hits aren't affected. In both regular brawl and brawl+ marth's fsmash has a launch speed of 3695 against a 0% pit.

edit5: Re: edit3: Actually, now that I think about it, it probably is proportional, just not to the whole thing. It must only affect one specific part of the launch speed formula. This definitely requires more testing.

edit6: Duh! The hitstun modifier we've been using changes the point at which c*w2 is applied, which is the dividing line between strong and weak hits. It isn't actually changing the launch speed at all, but instead telling something else to change it. Essentially, it's exactly what I originally thought, but I had forgotten about the c*w2 value only getting applied to strong hits which resulted in me going in circles for a bit there.

 

[Steel]

I'll get you those numbers tomorrow.

@all your edits

Interesting, glad to see something is being made of this

 

[leafgreen386]

So then this should mean that a weak hit becomes a strong hit at 1020 divided by the hitstun constant (1020 / .4 = 2550). This would make the barrier 1961.5 (don't know whether it rounds up or down yet) for a hitstun level of +.12. I'll try finding some moves with a launch speed around there tomorrow.

I can already tell you that the brawl+ throw should have had 17 frames of hitstun. What I really need is... well, actually, I need a lot of weak hits and their respective launch speeds if the fthrow has more than 16 frames of hitstun. If they really have a different division constant than strong hits do, then just one entry of data won't be enough to figure it out.

For random guesses as to what that different division constant might be, though... I call 44.

And happy new year smashlab =D

 

[Steel]

Sorry about the delay, been a bit busy. You still want those frame numbers ya?

 

[leafgreen386]

Yeah. Especially for the regular brawl fthrow.

 

[Steel]

Sorry it took so long, had to get new batteries.

Marth's fthrow on wolf:
0%
Frames until DJ: 31
Frames until Aerial: 31
Frames until Air Dodge: 31

20%
Frames until DJ: 33
Frames until Aerial: 33
Frames until Air Dodge: 14

Seems only the air dodge is quicker.

 

[ColinJF]

Can you get it at a few other percentages values as well? Like 30%, 40%, 100%, 150%? One might speculate that the separation of the times increases with higher launch speeds.

I'd be interested in that kind of data for a few moves actually, since what determines the time when you use an aerial/air dodge is still a bit of a mystery.

 

[leafgreen386]

Woah. That's almost double the stun it should have if it were a strong hit. Er... actually, it's exactly double. Rounding. So the division constant is probably 33 for weak hits.

But the 20% throw confuses me. Shouldn't that be a strong hit by then? Unless the fthrow has a REALLY small growth rate. Those stun numbers are way too high for a strong hit, unless it miraculously hits a launch speed of between 5281 and 5445 in only 20%. And then the airdodge comes out that much quicker on that than it does on the 0% hit? That's really bizarre.

 

[Steel]

Your logic confuses me, i need more laymens terms.

But Ill try my best to answer anyway.. Like I said Wolf's flying animation after the fthrow puts him into a 360 kinda thing and he can't do anything until he hits the ground (being we are on-stage), no matter where he DI's. This allows Marth to chain throws on him or tip it to an fsmash.

When we're at 20% however (it probably starts a little lower, i just chose 20% randomly), he goes into a regular hit stun animation. For whatever reason, you can air dodge considerably early at this point.

 

[ColinJF]

Well there doesn't seem to be anything too surprising about most of that. You can't DI weak hits or take any action until the hitstun ends. Taking actions out of hitstun only works for "strong" hits (> 2550 launch speed). That's why you can suddenly air dodge earlier once it's a strong hit.

 

[leafgreen386]

Hm... this is weird. The hitstun shouldn't be lasting that long... Are you saying that you weren't able to act even once you were in the tumble? Because that should be when hitstun ends... unless you were actually in pre-tumble the entire time before that... Your frame numbers are double that of what you should be getting...

From my testing, you go into the tumble at the frame number equal to the attack's launch speed divided by 165, which is also the time you appear to be in hitstun for. Hacks indicate that this should actually be defined as:

h = math.ceiling(.4*l / 66)

Where h is hitstun in frames and l is the launch speed of the move.

The launch speed for marth's fthrow against a 20% wolf is probably somewhere around 2800, which should give 17 frames of hitstun. I'll check the actual value later, but it isn't gonna be getting too much higher than that until much higher percents.

Now, infzy brought up the various hit states of hitstun in a brawl tactical discussion thread. Most of my testing was done with marth's fsmash, so if that puts you into a different stun animation than his fthrow does, than perhaps this could help explain this peculiarity. Although I find it odd that the developers would only make two hit classes for launch speed, but make four hitstun classes, which is why I suspected they were dependent on hitstun and not the other way around.

I just don't get why your stun values are so high. One of us probably screwed something up, although I'm not sure who it was yet.

Hm, I guess this is my 1400th post. lol

 

[Steel]

Perhaps the place where i start to count the frames is off? Marth slams Wolf's forehead into the concrete. I start counting the frame Wolf leaves the ground. Maybe that's where we are confused.

 

[leafgreen386]

Hm... that's possible. Do you think you could upload the footage you're analyzing somewhere? I wanna see this for myself.

Although something that came to mind when I was talking this over with kupo is that perhaps throws use a different division constant. Maybe throws use a division constant of 33, while regular hits use a division constant of 66? But even that doesn't make sense, since I didn't think throws had that much hitstun on them.

Could you also try and get the hitstun for marth's tipper fsmash? Try it on a 0% pit. I already have complete data for that, so if you're still getting high values, then the problem will be obvious. It's just figuring out which side that problem was on that will become the issue.

Actually, you should try to get another throw, too. Try marth's uthrow. It'll help us figure out if throws have a different division constant or not.

 

[infomon]

Oh, something I keep forgetting to check.... during Brawl's "hitstun" you can also throw an item, or use a Zair. I suppose you can prolly do these at the same time as you could use an aerial, but it's worth testing that; if you can Zair as early as an airdodge, that'd be sweet.

...... on that note, you can cancel an airdodge with an item throw, which at least gives us one way to attack from hitstun earlier than with an aerial. Now, can you cancel an airdodge with a Zair as well?

Brawl is weird

 

[infomon]

Double-post because this is totally awesome.

"Strong hits" allow you to airdodge even earlier than you can use an aerial, which you can do earlier than hitstun normally ends. You can throw items out of airdodges, and Zairs happen with attack input during an airdodge. So with strong hits, item-throws and Zairs can happen extra early.

To follow with your Marth Fthrow example, Link at 0% hits the ground before he can do anything useful, but Link at 20% can Zair Marth as punishment, or hit him with an item if he's holding one (ex. bomb). Knowing you can do these things earlier than you could do anything else from hitstun... as far as I'm concerned, that's an AT, and probably a new AT.

To extend it further... in my [video on bucket-braking], I show some extreme examples where G&W bucket-brakes by throwing a banana to escape hitstun earlier than even his Nair would allow. Now we know that the impact is even more extreme: I should have airdodged to allow the item-throw to happen even earlier, thus increasing the effectiveness of bucket-braking!

Of course that's of fairly limited use since G&W/DK/Yoshi rarely have items... but almost every character has a momentum-cancelling strategy involving their Jumps or Specials, that helps them avoid the killzones.... so holding an item should give a noticeable increase to everyone's survivability, if you know to item throw cancel an airdodge from hitstun. The benefit will be the number of frames earlier you can airdodge > item-throw, before you could throw an item normally. Which is somehow proportional to knockback. I should probably do another video, including more realistic examples than [this one] (which was just a discovery/question vid).

Furthermore, since Zairs can come out of hitstun quite a bit earlier than aerials, and since Link's aerials are all so slow, Link's optimal momentum-cancelling strategy will almost certainly be to Zair out of hitstun, then jump towards the stage. Unless he's approaching the vertical killzone, in which case he should either omit the jump, or maaaaybe fastfall an aerial instead of the Zair, or both. This can prolly apply to some other characters with Zairs as well. Oh, and of course if Link has a bomb then itemthrow-cancelling an airdodge will be better than Zair.

This all seems really novel and important; I'd really like to publish this outside of the Lab. Of course, this might very well be known by lotsa ppl and I'm just late to the party........ and also, my excitement over momentum-cancelling vastly exceeds that of the general public, so mebbe I'm just really hyped over nothing

Edit: Zairs are all reeeallly slow.... I doubt there are any that are faster than an airdodge, so I doubt any of them are ever useful for momentum-cancelling purposes, unless you're trying to actually tether something as you're flying away. They're still nice for attacking out of weak "strong" hits, though.

In my G&W Bucket Braking video, I show a G&W surviving a DK Fsmash at 195%; that's on the largest possible custom stage, no DI, and G&W is holding a banana. If I throw the banana out of an airdodge, I can survive that hit at 225%, and maybe higher. Rawr!

 

[infomon]

^^ sorry for triple-posts (are those ok back here? w/e lol)

Just thought I'd say, I'm making a thread about the above, in Brawl Tactical Discussion, since it's of immediate relevance to every character, as far as I'm concerned. And none of this is new, I just put 2+2 together and I doubt most ppl would have thought of this......

Edit: Although it would be nice to know how much earlier you can airdodge rather than use an aerial out of hitstun, that's something we might not solve 100% for a while. The other relevant question is the whole weak hits vs. strong hits, low-percent counterattack-with-airdodgecancelled-itemthrow-or-Zair thing, but w/e.

Edit 2: Sadly I won't have a cheesy video about this (showing off the even-more-broken bucket-braking lolol), and I know ppl don't like reading my walls of text, but oh well

 

[leafgreen386]

Well, the sooner we get some data about the frame you can airdodge on at various launch speeds, the sooner we'll be able to figure it out. It's probably fairly simple, and it may also be tied to the different hitstun animations. In fact, I think I remember someone here saying that the "corkscrew" hitstun animation was the only one that gave "full" hitstun. We'll need to figure out what triggers the different hitstun animations, first, though.

 

[infomon]

That claim about "corkscrew hitstun" was made in my [hitstun animations thread], and turned out to be false (at least in my experiments). There might still be some tangible difference between the animations, but it'd be extremely slight at best.

I was just typing my announcement post, where I would describe
1. using airdodge-cancelled item throws, and Zairs, (and both at once, lol) as a means to retaliate from low-percent (but technically "strong hit") combo attempts from which you could otherwise only airdodge, and
2. airdodge-cancelled item throws for momentum-cancelling / survivability purposes

Would you recommend I hold off on it until we know exactly what makes a hit "weak" or "strong"? It might even help if we have an official name for the distinction....... but I doubt these technical details will even matter to most people, and I'd really like to get this info out.

 

[leafgreen386]

We already know what makes a hit "weak" or "strong." If the launch speed is under 2550, it is a weak hit. Anything over that is a strong hit (where the c value colin mentions is then applied, as well).

 

[ColinJF]

Technically, it's if the intermediate launch speed is above 2550, i.e., the launch speed before w2 * c is added. The application of w2 * c can actually make the launch speed drop below 2550 (since w2 * c can be negative), but it'd still be a strong hit.

 

[infomon]

Oh oops, right I even read that stuff..... I guess I meant, when we know the frame advantage of airdodging vs. aerials, and how it grows. Well, that's hardly relevant at this point in time, although it will be good to know in particular for some characters like Wario, whose airdodge #F rivals that of his fastest aerial, so it's sometimes better for momentum-cancelling if he does one over the other. I guess MK will want to know if he should bother airdodge-cancelled item-throw'ing for momentum cancelling rather than just waiting and using his Uair, lol, but all those sorts of details will depend on our frame data as well as figuring out the airdodge/aerial tradeoff mechanics.

So I'll go ahead and make my thread, I guess?

 

[ColinJF]

Just don't mess it up! Heh.

I'm joking. You know what you're talking about and the thread will be fine.

 

[infomon]

lol ok, I has a thread:
http://www.smashboards.com/showthread.php?t=215117

thanks for the input

 

[leafgreen386]

Mugwhump's animations were both strong hits I'm pretty sure, since they both sent you into the tumble (iirc). It'd be better if you just described what the actual difference between the two are - Weak hits do not cause you to enter the tumble, cannot be DIed (except for SDI), and cannot have their hitstun canceled. Strong hits cause you to enter the tumble after the full hitstun ends, can be DIed, and can have their hitstun canceled by an attack or airdodge.

 

[infomon]

Oh! You're right thanks, fixed. HUARGH/WHARRGARBL are sliding-across-the-ground anims, they're not the same as aerial hitstun anims.

 

[leafgreen386]

I think I finally figured out the discrepancy. Or at least one of the discrepancies if there was more than one. Wolf is a much faster faller than pit is, who I did all my testing on. Thinking back to melee, it was always the floatier characters who got out of hitstun quickly, and I think it's very likely that they kept this mechanic. I'm not sure if they just made various hitstun brackets and stuffed the characters into them or if each of them have a different division constant. I'm gonna have to see if kupo is willing to help me with this again, because clearly there is considerably more research to be done in regard to hitstun.