Optimizing Puff.

[HermitHelmet]

Yeah, discuss.

 

[deadPhoenix]

What do you mean by optimizing Puff?

I feel like HBox pretty much has the ideal Puff. His play is fairly "optimal." He consistently gets rest setups, lands most of his edgegaurds, etc.

 

[HermitHelmet]

What do you mean by optimizing Puff?

I feel like HBox pretty much has the ideal Puff. His play is fairly "optimal." He consistently gets rest setups, lands most of his edgegaurds, etc.

Pretty much the equivalent to a "20XX Puff", but realistic (not TAS) at the same time. It's kind of hard to word it.

 

[_DarkLava]

IMO, optimizing Puff either consists of completely walling opponents and/or perfectly punishing anything with a rest.

 

[Crossie]

completely walling opponents

I'd say this is the optimal defensive play, which isn't necessarily optimal in general.

perfectly punishing anything with a rest

This just doesn't sound realistic.

I think 20XX jokes and TAS videos have misconstrued what optimization is. Moreover, to speak of optimization in a broad sense doesn't make sense unless you're writing a long article about it or something. Discussing optimal tech chases or optimal up throw punishes on spacies if they DI properly or something makes sense, but not optimization in general. I say this not because I don't think this thread should be here, but because I'm very interested in optimizing certain aspects of my Puff and I want this thread to have something concrete to discuss rather than the abstract idea of optimization.

Just as something to get this conversation started, what is the optimal follow-up after you down throw floatier characters like Marth or Ganon and they DI in? Is up air the best you can get? Or am I just slow and the rest is possible? Or maybe they're only partially DI'ing in when I try to rest them and they need to hard DI in. Anyways, just something to discuss.

 

[_DarkLava]

This just doesn't sound realistic.

I think 20XX jokes and TAS videos have misconstrued what optimization is. Moreover, to speak of optimization in a broad sense doesn't make sense unless you're writing a long article about it or something. Discussing optimal tech chases or optimal up throw punishes on spacies if they DI properly or something makes sense, but not optimization in general. I say this not because I don't think this thread should be here, but because I'm very interested in optimizing certain aspects of my Puff and I want this thread to have something concrete to discuss rather than the abstract idea of optimization.

You make a good point, but the way I define optimization is being able to bring your character to the peak of it's playability, and to react to every thing, every attack, movement, shield, recovery in the best way possible, such as a chess computer is able to find the best moves in chess positions, I would akin optimization to this. I think that this "best way possible" is at most times a frame 1 Rest. This is how I would define the term "optimization."

That being said, take for example tech chasing with Sheik. After one Dthrow by Sheik on Fox, Falco, or Falcon, the Sheik should be able to zero to death them with optimized play, in this place, optimized tech chases. The best thing to do, is to keep dthrowing until some X percent where death would be certain. This is what I think a branch of optimization for Sheik is, her tech chasing. I would consider Puff's her edgeguarding and her Rest punishes.

I would argue that being able to punish any whiffed grab with a Rest would be optimal play. Also, say Fox is below the stage and recovers at a high angle. While he's falling, as difficult as the timing may be, a rest on a midair free falling Fox I would consider optimal play, rather than the safer Bair or Fair. This is what I mean when I say perfectly punishing anything with a Rest. Laggy smash attack on shield. Wavedash, Rest. Laggy aerial on shield or CC Puff. Rest.

Just as something to get this conversation started, what is the optimal follow-up after you down throw floatier characters like Marth or Ganon and they DI in? Is up air the best you can get? Or am I just slow and the rest is possible? Or maybe they're only partially DI'ing in when I try to rest them and they need to hard DI in. Anyways, just something to discuss.

Allow me to draw an analogy back to Chess again. I'm sorry but I love chess, and I think it fits perfectly with the idea of optimization of characters and tactics. Say you play a bad move, it looks like a reasonable move, but ultimately it was a bad move because you lose your queen five turns later because of it. You might ask "well what would be optimal play after I played this bad move that made me lose my queen? How would I not lose my queen in this position?" Assuming the opponent also played optimally to take your queen, then nothing. If you played the best possible moves in those following positions, then your queen is 100% gone. If you screw up your optimal play and play a move that results in a blunder later on then the move you played was simply bad, or rather, not optimal. Likewise, if you are stuck with only a king and the other opponent has only a king and rook, it doesn't matter how optimal and perfect your play is. If your opponent knows what they are doing, you'll lose.

Now allow me to draw that back to smash and Puff. Who's to say that Dthrowing your opponent was the best move possible? There could have been many other variables in deciding. Are they close to the edge? Why not Bthrow (or Fthrow if you were on the other side of the stage) and set up an edgeguarding opportunity? Did they attack your shield close enough for you to punish with an OoS rest? Why didn't you do it? Why were you in a situation where your best option was to Dthrow with no hope of following it up? Did you screw up before that and thus not achieve optimal play? But let's say that Dthrowing the floaty was actually your best move possible in whatever situation you were in, whether it be due to some small prior error or not. If that's the case and you're forced to Dthrow with little hope of following it up with anything significant, then that is, in that specific moment, optimal play. The best possible continuation after a Dthrow might not be a rest because the game's mechanics do not allow for it. The best option would something that would not secure a win for you, this case a Uair or just nothing at all because a Dthrow was all you could have possibly gotten.

So perhaps, I should have worded my original statement better. I believe that as Puff there are two things necessary to play her optimally, perfect your edgeguarding and punishing anything with a frame 1 Rest as long as it is within the game's capabilities and human capability (i.e not being able to punish a Fox who whiffs an Usmash because he is legitimately too far away to reach before the endlag finishes, or trying to react faster than what is humanly possible). And edgeguarding other characters to the best of what is humanly and mechanically possible. Whether it means that the best punish results in the opponent not dying, then so be it.

To add on with Puff's optimization and discussion. Say Fox is recovering back on to Final Destination and begins his Firefox 10 degrees from the stage. Will there be enough time to actually be able to react within human limits if he attempts to sweet spot the ledge? You could try to read it, but would we consider reading movements as optimal play due to their being a chance that it could be wrong? I don't know. Puff doesn't have a lot of skills to actually optimize. Punishes and edgeguarding being the only few I can actually think of. Maybe crouch cancelling too.

 

[Crossie]

You make a good point, but the way I define optimization is being able to bring your character to the peak of it's playability, and to react to every thing, every attack, movement, shield, recovery in the best way possible, such as a chess computer is able to find the best moves in chess positions, I would akin optimization to this. I think that this "best way possible" is at most times a frame 1 Rest. This is how I would define the term "optimization."

I don't know if that definition works for Melee, though, because the game is so amazingly suspended between read-reaction play. If you take a game and look at it frame-by-frame, different people will have different opinions of what options appear optimal at each frame because some people will say a hard read is worth trying, others will want a less committal read, and others will want to wait and try to react. There is no universal agreement on what proportion of reading vs reacting is optimal, but it's clear that it's possible to get huge punishes off reads that aren't possible off reaction.

While he's falling, as difficult as the timing may be, a rest on a midair free falling Fox I would consider optimal play, rather than the safer Bair or Fair.

This seems pretty suboptimal. Why rest and potentially open yourself up to a punish when he comes back instead of just edgeguarding if we're imagining TAS-level perfection anyways?

I won't quote the rest, but in general, I think you weren't responding to what I was trying to ask. I know down throw is suboptimal on floaty characters if it has few/no followups, I was asking if there are followups that I'm just messing up or if there really aren't any. Anyways, down throwing is a decent option near the ledge even on floaties since you force them to either DI off stage or get put in a platform tech chase situation.

 

[Crossie]

Also, while I'm here, I want this board's opinion on Puff using dair OoS. It's technically her fastest aerial and it can certainly lead into things (grabs, mostly), but its hitbox isn't great. It seems a bit risky, but I'm sure it has applications. Maybe as an option for Marths; rather than shield tippers and fair OoS the whole game, like Mango suggests, mix in some dairs so you can get some bigger punishes and some variety. I don't know, just something I was thinking about that I wanted some input on.

 

[Massive]

Against characters without a shine, dair is remarkably safe OoS and on shield. It doesn't have a huge amount of utility, but it's not a bad escape option.

 

[_DarkLava]

I don't know if that definition works for Melee, though, because the game is so amazingly suspended between read-reaction play. If you take a game and look at it frame-by-frame, different people will have different opinions of what options appear optimal at each frame because some people will say a hard read is worth trying, others will want a less committal read, and others will want to wait and try to react. There is no universal agreement on what proportion of reading vs reacting is optimal, but it's clear that it's possible to get huge punishes off reads that aren't possible off reaction.

Which is why I doubt we will ever reach the general consensus to what "optimal play" actually means. Meaning perfect in all scenarios. This is because Smash is an unsolvable game, not even partially solvable. The nature of Smash doesn't allow to play perfectly, because to do so would mean to know what you're opponent is thinking. It would be akin to two omniscient beings playing a game against each other. It seems rather far-fetched and far away that we could even think about playing perfectly in this hypothetical year 20XX. Look at the game of Chess. Arguably more time has been put in to Chess by its community than any other modern game, and while different openings are better than others, if you play that opening vs them it doesn't necessarily mean that you will win that game. Smash just has so many variables that makes the game seem so impossible to play perfectly. But I suppose I'm preaching to a choir.

While he's falling, as difficult as the timing may be, a rest on a midair free falling Fox I would consider optimal play, rather than the safer Bair or Fair.

You were right in your rebuttal. Why rest when you have perfect edgeguarding? Which furthers my statement of just optimal edgeguarding and punishes with Rests is fine enough with Jigglypuff. It just so happens that the situation I used affirmed the other one instead of the one that I said it did. I suppose Resting would only be used at a last stock situation, but then again, maybe optimal play means to just Rest and accept the fact you'll get a few percentage done on you out of the punish, but know that you'll be able to secure a kill. But can we really say that because you're playing optimally and shouldn't drop the edgeguard anyway. It confuses me.

Anyways, down throwing is a decent option near the ledge even on floaties since you force them to either DI off stage or get put in a platform tech chase situation.

This was a mistake on my part. I had assumed that this situation would be played on Final Destination with no platforms. But yes, down throw on to a platform and going into a tech chase is good because it has the possibility to result in a Rest if played well enough. And honestly tech chase opportunities on platforms IMO should be ended with Rest, or . If you can't be killed out of a Rest punish, then try to Rest in a tech chase on platforms. You might end up taking a few percents but that is obviously a trade worth taking.

In reference to Dair OoS, I don't think it's a very good option. I've never seen from what I remember Hungrybox or Mango perform a Dair OoS, they normally jump away, wavedash to grab, or nair. I just haven't seen Dair OoS. Mainly because the benefits of Dair comes whenever you hit the ground while the move is still in use, so that the character who is receiving the Dair does that weird standing flip. I don't think that Dair lasts long enough immediately OoS to put it into the ground to get that weird flip or the appropriate hitstun to actually get a follow up. Like falling Dair you can get a good follow up, but Dair OoS, I'm not too sure. I'll test this at some point or someone else can. If you can do this, I can't think of a reason not to do it. As in escape option, I think you'd be better to just jump away if you have the ability to Dair, so you can reset.

 

[Crossie]

Against characters without a shine, dair is remarkably safe OoS and on shield. It doesn't have a huge amount of utility, but it's not a bad escape option.

Cool, good to know. What about IC? Is it safe on them or are you likely to miss one of them with some of the hits and then get shield grabbed?

Which is why I doubt we will ever reach the general consensus to what "optimal play" actually means.

Alright, as long as you're willing to admit that, I'm more okay with your definition. I'm not sure it's the definition I'd use, but it's internally consistent.

But can we really say that because you're playing optimally and shouldn't drop the edgeguard anyway. It confuses me.

Exactly. Since it isn't necessarily possible to cover 100% of the options it's hard to say whether "optimal" play will necessarily get the edgeguard or not. Especially with characters like Fox who have so many different options when recovering. So at some point, if we built a computer to play optimally, it would have to calculate how confident it was in its edgeguard and if it was confident enough it would do that, if it wasn't confident enough it would have to go for the rest and take the punish.

In reference to Dair OoS, I don't think it's a very good option. I've never seen from what I remember Hungrybox or Mango perform a Dair OoS, they normally jump away, wavedash to grab, or nair. I just haven't seen Dair OoS. Mainly because the benefits of Dair comes whenever you hit the ground while the move is still in use, so that the character who is receiving the Dair does that weird standing flip. I don't think that Dair lasts long enough immediately OoS to put it into the ground to get that weird flip or the appropriate hitstun to actually get a follow up. Like falling Dair you can get a good follow up, but Dair OoS, I'm not too sure.

Well, Mango never does it because he never uses dair at all. He's said he can't consistently get the l-cancel. But regarding Hungrybox, I agree that he doesn't really use it, but I don't think we should literally be emulating him 100%. Learning from him, yes. Trying to be able to do the things he does, yes. But not direct emulation. Hungrybox isn't the perfect Puff, he just looks like it because there aren't any other Puffs at his level to challenge him. He has his own idiosyncrasies and his own not-always-optimal style. Short hop dair definitely can put them in the flippy animation you're referring to. Puff's short hop is pretty low, the animation definitely doesn't end before she hits the ground. Even if you don't fast fall the attack doesn't finish before she hits the ground. Anyways, the point is I'm pretty sure it has applications of some sort.

 

[Massive]

Cool, good to know. What about IC? Is it safe on them or are you likely to miss one of them with some of the hits and then get shield grabbed?

It's best to not get caught in shield vs. ICs or if you are, just to get away as fast as possible. You don't want to risk one of them CCing and punishing your attack.

 

[_DarkLava]

Even if you don't fast fall the attack doesn't finish before she hits the ground. Anyways, the point is I'm pretty sure it has applications of some sort.

Yeah I took the time to finally test that. It definitely works without a doubt. From what I've noticed, the flippy animation occurs when the character is at a high enough percent, which depends on the character itself. So I think that if you try to do it before that percent it would be a bad option because I think they can act before you can. I also can't confirm this but I think they can unless their in that animation. It's hard to test this because I by myself at the moment.

 

[-LzR-]

I don't think the flip is necessary for the drillgrab or drill usmash combo. At least I get these all the time even when it rarely flips. I think the flip is necessary for a drillrest though. Does anyone have the framedata on the advantage when the flip occurs? There might be even better followups like Fsmash available.

 

[Crossie]

I don't think the flip is necessary for the drillgrab or drill usmash combo. At least I get these all the time even when it rarely flips. I think the flip is necessary for a drillrest though. Does anyone have the framedata on the advantage when the flip occurs? There might be even better followups like Fsmash available.

I talked to someone today who said you have a seven frame window if you get them in the flippy animation. That didn't sound right, though, because that would mean you'd need to do a frame perfect up smash or else they could shield or buffer a roll.

 

[Massive]

Does anyone have the framedata on the advantage when the flip occurs? There might be even better followups like Fsmash available.

As far as I know the actual frame data for this has never been examined.

However, there is some ancient knowledge I can transfer to you, knowledge penned by the legendary Phanna.

 

[Crossie]

As far as I know the actual frame data for this has never been examined.

However, there is some ancient knowledge I can transfer to you, knowledge penned by the legendary Phanna.

Guaranteed to work at 0% on Fox? Is this info trustworthy? I have tons of respect for Phanna, but that just sounds so wrong.

EDIT: Nevermind, Phanna's a god, definitely going to try to work this into my list of ways to style on people I know I can beat