What are Option Selects?

Can you define what option selects are in fighting games, what they do, and why they are useful? I looked them up on Google, but I still can’t seem to grasp the concept of them.

Sorry for answering late. An option select is when you input one thing that can have multiple possible resolutions depending on what your opponent does.

Probably the most common type of option select in fighting games is one where the move will situationally have two different effects depending on what happens. For example, in 1 button fighting games, you can either throw by pressing forward + attack, or with that same input, just get an attack. In KOF in particular, across every single KOF game, it’s always been possible to option select between throwing, and some fast close range attack, because either your C or D (KOF’s two throw buttons) has a close range version of the attack bound to that button which comes out extremely fast. So if the opponent is throwable, you get a throw, if they’re not, you get a fast and powerful attack.

Another common type of option select is inputting additional actions that will get eaten depending on whether your first input comes out or not. For example, in Guilty Gear Xrd, you’re allowed to Yellow Roman Cancel command throws, but not Red Roman Cancel them. This means if you do a command throw, you can input the cancel, and it will cancel if you miss, and not cancel if you hit. So if you can cancel them when they miss, this means you can do something else immediately after they whiff, such as jump and attempt an air throw, meaning you can run into someone’s face, and either get a grounded command throw or an air throw, you’re throwing them on the air and ground in extremely short succession, and whichever one connects will continue. This is effectively an unblockable.

Another version of this is inputting things during hitstop. Hitstop puts your character in a state where you cannot act. By inputting when this time would happen, you’re making it so you’ll get actions if you fail to hit, but won’t get actions if you successfully hit. You can use this to situationally follow up depending on whiff or hit.

Hitstop also has another function, acting as the cancel window for special moves. In some games, hitstop is actually different on block and hit. This is EXTREMELY DANGEROUS! If hitstop is longer on hit than on block, then people can precisely target that interval where hitstop is longer and try to time their cancel for right then, because on hit, they will successfully cancel, and on block, they will be inputting too late to cancel. This means you can commit to risky special moves only on hit, making your moves way more powerful without risking anything. In an earlier version of SFV, Ken had this issue. I heard it also existed in MK9.

In a way, special move canceling by itself can be viewed as an option select, since you’ll cancel on hit/block, but not on whiff. Once you realize this, you can poke with cancellable normals, and time your special move buffer on them so it’s done before the move recovers. Then you can use that special move slightly outside where your opponent is standing to catch them if they move forward and therefore cannot block.

Option selects can occur anywhere where one input or sequence of inputs can be interpreted as 2 possible outcomes. Learning to look for cases like this is very tricky.

These aren’t very common in Super Smash Bros, because there’s almost no cancels in Smash Bros, and moves tend to only have 1 consistent effect regardless of what situation they’re used in. The amount of hitstop on whiff, hit, and block are different, but this doesn’t tend to actually be useful for creating option selects in the smash series. There are cases that are very common however in smash bros where you can position yourself in such a way that regardless of what option your opponent picks, you can react to it and perform the counter option. Smash Bros fans call this option coverage. You sometimes see commentators mistakenly call it an option select, but it’s not the same series of inputs every time, it’s the person consciously changing which input they’ll use based on what they see.

One example of a true option select in Smash Bros is in Project M with G&W. Game and Watch’s throws are all identical in appearance, so you never know which way he’ll throw you. He gets his best followups off down throw, but you can defeat that easily by inputting a tech right before his throw ends. This means that regardless of whether he actually down throws you or not, you will always tech it.

Another one has to do with meteor canceling. There’s actually 2 ways to meteor cancel, jumping, and up B-ing. There is an 18 frame lockout window after being meteor’d and trying to input too soon will lock you out from canceling, but jumping and up B both count independently from each other, so you can try both instead of just one. So if you get meteored, you can input jump first, then up B immediately afterwards, and if you screw up the first one, the second one might work.

I hope this makes it more clear. Option selects can be really really varied across games. Guilty Gear Xrd used to have one where you could mash roman cancel during combos between 25% and 50% meter, and it would automatically cancel your combo if your opponent bursted out, because they were no longer counted as being in hitstun. This means if they don’t burst, your combo continues, but if they do burst, you instantly YRC, making you safe from their burst. This was later fixed.

Using Advanced AI for Single Player Games

Thoughts on all the people saying they should put OpenAI and Deepmind into single player games to make the enemies smarter? https://www.theverge.com/2017/8/11/16137388/dota-2-dendi-open-ai-elon-musk

This entire line of discussion of is stupid. You can’t compare AI beating humans at multiplayer games with AI used in singleplayer games. The two are trying to accomplish fundamentally different purposes. AI in single player games is a Game Design issue, not a Technology issue. Developing AI that will beat players every single time at a multiplayer game usually isn’t a process of making the AI smarter the same way a human opponent can improve, it’s usually a matter of the AI having better timing, reaction speed, and knowing the best decision to make for each situation. AI doesn’t play the game the same way humans do, with our slow reaction time, our ability to only keep like 9 things in our head at maximum, our limited ability to simulate and predict what the game state will be. If you want to build a perfect AI that always wins at say, street fighter, you could just make it uppercut when you attack, throw when you block, and block when you uppercut, and the game is over. There’s already a number of bots like this at the top of the SFV leaderboards and the only reason they don’t win every match is because of netplay lag. Continue reading

Contact Damage Enemies are Amazing

How come direct contact enemies are so good? Alpha Metroids in AM2R are so great, Konami slapped a sine wave motion to some faces and they are talked about to this day etc, they seem to create much better depth with character movement and abilities than a dedicated “press for iframes” dodge button

Contact damage is nice because it allows the enemy to block your way with their body. In non-contact damage games, you can hug up against enemies to run past them without much issue. It also knocks you back in the process of touching them (having no-damage knockback, accompanied by a shoving animation, might be a smart idea to make enemies better at blocking your way without contact damage) Continue reading

Negative Feedback Makes Me Rage

What do you think of the rage mechanic in smash 4?

It’s irritating. If there weren’t multiple stocks, then I’d outright hate it unconditionally.

Rage is a knockback multiplier, similar to weight. As you gain percentage, you deal more knockback, which also means you deal more hitstun. Damage in Smash Bros increases knockback, so by dealing damage to your opponent, you’re giving them rage, which increases their knockback, much as if you had taken damage. Effectively, by hurting your opponent, you’re hurting yourself, thanks to rage. (Actually, I think rage multiplies knockback more than percentage does, so you’re hurting yourself more than you’re hurting them) Continue reading

Throws in Smash Bros

What do you think of the way throws work in the Smash Bros series?

There’s nothing else like it in any game I’ve seen, except Skullgirls with Beowolf, and I guess King in Tekken.

Lemme just describe how it works, basically, once you get a grab, the other player is held in front of you in a grapple, instead of instantly thrown like in most other fighting games. In this state, there is a timer that counts down until the other guy is released, the timer is longer relative to how high percentage is. (and if the guy is holding up or presses jump, then they’ll jump upwards when released) The player who is grabbed can reduce the duration of this timer by mashing buttons (I hear the best method is to spin the control stick, varies by game since Brawl and Smash 4 handle inputs differently). Once the timer runs out, you’re released as soon as it’s possible. So past a certain point, mashing out of icies wobble is pointless, you’ll escape automatically if they mess up. Continue reading

Fake-out Attacks

What do you think of fakeout attacks? (Both when enemies use them in single plaher action games, or other genre’s, and in multiplayer fighting games)

Enemies using fakeouts in single player games:
I can’t think of a reason to use this off the top of my head. From the player’s perspective, it’s like the enemy randomly attacks or doesn’t. This creates the situation where the enemy uses the fakeout attack, and the player can safely attack them, but randomly sometimes they’ll actually do it, so they’ll trade hits with the player. Beginners might get scared by the fakeout attack, but intermediate will realize it’s basically just an animation where they do nothing important, so they can be attacked.

Maybe it would make sense in a game where enemies attacking requires an action that costs something from the player, and there’s a tell between the real and fakeout attack. Like in Furi, fakeouts would mean you can’t parry the next incoming attack, so if you made a system similar to this you could have the mixup between fakeout or not, then maybe a small reaction period for you to realize you’ve been faked out to let you dodge or something. Similar was used in the Mario and Luigi RPGs, where enemies would sometimes fake you out, so you’d jump over incoming attacks that weren’t coming, setting you up to be hit by the followup.
Continue reading

Smash Bros Move-Staling is Pointless

Any thoughts on move-staling in Smash?

I don’t think it serves a real design purpose. It weakens repeated attacks, which can make the effect of attacks subtly inconsistent, changing the amount of damage, knockback, and shieldstun. The thing is, there’s really no need to make repeated attacks weaker. Making repeated attacks weaker doesn’t prevent any type of degenerate play, it doesn’t encourage any specific tactical plays that are beneficial for the game overall, or add a significant situational factor that can be taken advantage of in the moment like stun.

It mildly discourages using the same move a lot, a tactic that many people would call spamming, but the thing is, there’s nothing wrong with spamming. If using the same move works versus your opponent, then you should keep doing it, not be forced by the system to use other moves to keep your useful moves powerful.

And stale moves can interfere with a lot of things, like it changes the knockback threshold on moves that will cause knockdown versus not, it can change shield stun, making safe on shield moves unsafe.

Thankfully the effect of stale moves in Melee is so small that it can largely be ignored, and PM had the good sense to remove the knockback component of stale moves completely. In Brawl however stale moves had a more extreme effect on knockback, enough that if you played a character like fox, it was recommended you only hit with the second hit of up air to kill, because the first hit would invoke scaling, reducing kill potential. Smash 4 has reduced the effect of stale moves, sitting it somewhere between Melee and Brawl, so it’s probably more tolerable in that game, but in general I don’t think it’s something that has a place in Smash Bros.

In a good fighting game, there doesn’t need to be a regulatory system preventing you from using duplicate moves, because in a good fighting game, using the same move repeatedly is a bad idea because it opens you up to be countered by your opponent.

Notably, Skullgirls has a mildly similar system in its game, the IPS, preventing you from using the same move to start a combo more than once, but of course this doesn’t mean that any of the moves in that game are situationally weaker in the neutral game, it just prevents you from doing infinites and practically nothing else. Using systems like this makes a lot more sense for limiting the length/strength of combos in traditional fighters than anything in Smash Bros, which doesn’t have issues with combo length.

Stale moves just feels like a design loose end trying to fix a problem that didn’t need fixing.

RNG Fighting Game Moves

What do you think of fighting characters that have special moves that change based on RNG (Like Luigi’s missfire, Peaches turnip pull, or anything else you can think of)?

I’m not really a fan, because sometimes the opponent just gets a lucky draw and you lose and oh well. Random stuff basically always does this unless there’s fore-knowledge of when it’s gonna come up ahead of time, which doesn’t completely negate the random effect, but mitigates it somewhat.

There’s a few other fighting game characters that have random effects like that, I can think of Faust, Teddy, and Zappa, from Guilty Gear and Persona 4 Arena. Phoenix Wright too, but the randomness screws him over more than his opponent. Game and Watch has a random effect in smash too, on his judgment hammer.

I mean, I just think all instances of randomness should be replaced by something that is deterministic and can be kept track of relatively easily, but which still looks random to the untrained eye, and has about the same level of frequency in outcome.

Barring that, there’s ways to build more consistent randomized systems, some of which you can see in Project M, some of which you can see in Dota 2.

One thing Project M did with Peach was make her Fsmash totally deterministic, it rotates the 3 possible items each time it’s used. Another thing is making it so Luigi’s next misfire is guaranteed to show up within the next 6 times it’s used, and allowing you to hang onto it. And with Game and Watch, they made it so the judgment hammer cannot repeat the last 2 numbers, and it has a light over it indicating whether the next number is even or odd, so you can stack the odds in your favor

Zappa in Guilty Gear has an ability to summon a ghost, and this is actually dependent on what second the timer is on. Some seconds still produce a random result however.

DotA 2, and a few other games, use what they call a PRNG (Pseudorandom number generator). Technically all games use a PRNG if you want to get really technical, because computers can’t generate random numbers, but that’s beside the point. Basically, when something says it has a certain odds of success, like 25%, instead of it actually having those odds each hit, it starts at a much lower chance of success and that chance increases drastically for each time it doesn’t succeed. This is set up so an ability with a 25% chance will occur roughly 1 in 4 times every time. By doing it this way, they reduce “floods” and “droughts” of a result coming up or failing to come up. People don’t actually have a very good mechanism for understanding random chance, we tend to fixate a little on the odds in the short term, and expect more uniformly distributed results than are actually likely to occur. This system makes it so a 25% chance effect will only very rarely occur twice in a row, and will only very rarely fail to occur after 6 attempts.
http://dota2.gamepedia.com/Random_distribution

My general recommendation is, don’t include random effects if you can avoid it, it usually doesn’t add a lot, even if it’s funny some times. If you feel it’s necessary, find an RNG algorithm that works for you instead of flat % chance.

The Majesty of Dark Souls 3’s Backstab

Can you explain how the backstab mechanic is different in the Souls games and why you think 3 has it best?

Easy.

Backstab works basically the same in Demon’s Souls and Dark Souls. If you’re behind the enemy, standing within a certain area, and press R1, then at that moment both you and the enemy will play a joint animation where you backstab him. During this animation you’re both invincible. It’s sort of like a 0 frame throw in old versions of street fighter. Oh, and you can’t backstab if your shield is up.

This meant that backstab, no matter what weapon you had, was essentially instant and inescapable. Your only defense was preventing the enemy from getting behind your back at all. Backstabs were more powerful than regular attacks, so a lot of PVP was just about backstabbing and avoiding backstabbing.

In Dark Souls 2, they decided to mess with backstabs a little. Instead of instantly going into the combined animation where you’re stabbing them, they gave you a little punch animation first. This punch was unblockable and would cancel on hit into the full backstab. If it missed then nothing else would happen. This is a bit more like a modern fighting game throw. You have a chance to roll out or walk away before getting hit by it. This significantly reduced the role of backstabs in PVP, while keeping them viable. Backstabs were still really powerful in PvE however. Also from a thematic standpoint it’s kind of weird that you punch the guy then stab them, didn’t really match the prior games.

In Bloodborne it seems like they REALLY didn’t want anyone to fuck with backstabs at all. In order to get a backstab, you needed to charge your R2 attack all the way up and hit the enemy in the back with it, much like the positioning for backstabs in the other games. Then they’d become vulnerable and you’d get to do the 0frame throw thing for a ton of bonus damage. So overall you were getting a lot more damage out of successful backstabs in bloodborne, but it was a lot harder to set them up mid-combat, and they were now basically completely irrelevant to PVP.

In Dark Souls 3, they clearly wanted to have backstabs match demon’s souls and dark souls 1, and to do this they assembled a mechanic that’s practically a work of art. Basically, when you press R1 within the activation range, instead of directly entering a joint animation with the enemy, you’ll instead play a specific backstabbing animation. If the enemy stays close to you, then at a certain point in this animation, there’s an invisible unblockable hitbox that comes out that forces them into a joint animation with you, canceling whatever they were doing, and canceling your lone backstabbing animation into the joint backstab (using animation blending). If they aren’t in range of this hitbox, then you’ll just continue to play the lone backstab animation, which functions as a regular (slower) attack basically. So you smoothly either get a backstab or just stab without needing to punch them first like dark souls 2, and without an instantaneous startup like DaS1 and DeS.