Jump to content

Archived

This topic is now archived and is closed to further replies.

Omex

A look at some of BBCP's overheads and how they're animated

Recommended Posts

does posting a discussion on overheads, blockability and etc for discussion automatically count as starting a flamewar? if it does i’ll ban myself after sorry

 

anyway here’s some shit. remember im naturally biased and can’t block anything here do not take my statements seriously, i’m here to present data over say what is reactable or not (barring haku 6b c’mon guys). most complaining is just me attempting to be funny

 

I’m working on a video dissecting some of Blazblue’s overheads and how they’re balanced, not only their speed, range and what gatlings into them, but a few other aspects like video comparison of how similar the startup frames might look to other moves, how smoothly animated they are, when in the animation are you able to identify that they’re doing their overhead, and similar. To do this, I started out by capturing a handful of overheads I thought would be perfect examples on video, and then used frame advance to capture every single ‘key frame’ of the attack. key frames being primary animation frames, or in the case of sprite games, the only frames of animation in the total duration. For example, Ragna’s 24f overhead has 8 or 9 frames of animation, and I screencapped and labeled all of them

 

here’s all of those screenshots for reference/personal use/whatever https://www.mediafire.com/?qwl0k2jc2tc5l91

 

interesting things: ragna’s 6b has 9 frames of animation counting the first and last frame (last frame being ‘on hit’, animates fairly smoothly throughout the duration and the first truly distinct frame that could be identified as 6b rather than anything else is either 11f or 13f, meaning you have ~13f or 11f to react from pure visuals alone.

 

for comparison, bullet’s 24f 6A is 8f total frames of animation, is animated fairly smoothly, and the first frames that could easily be identified visually as ‘this is her overhead’ are 4f, 7f or 10f, depending. that plus less gatling options than Ragna is probably why her overhead is kinda shite?

 

now, hazama’s 6a. 22f startup, and it essentially only has 6 frames of total animation-frames 1, 5, 8, and 20-22. there are frames inbetween 8-20 where his coat buckles flutter in the air somewhat, but the animation is small enough that it’s not worth including in the count here. there’s clearly some sort of balancing going on by making the animation clearly identifyable by frame 8 but then having no sort of animation afterwards until frame 20-i’m not a scientist so i have no idea how certain animation styles affect the human brain reacting to it for a fighting game, but there’s no way this wasn’t intentional

 

similarly, hazama’s 6b is animated in the same way. this is probably to help make it hard distinguish between them, since otherwise people would just react to hazama standing up or crouching to block high/low until he goes into stances (which was a bigger deal back when he couldn’t combo off of them without charging/etc)

 

on the opposite end of things litchi’s 22f [e]6a, the first clearly identifiable frame is actually frame 16(!) and it’s the most well animated overhead i’ve looked at yet with a new frame of animation every 2 frames, giving it a 22f duration with 11 keyframes. most of the frames before frame 16 are all just very slight hand movements however, making it difficult to identify. of course [m]6a has the staff spinning so that is a bigger tell that [e]6a lacks. similar to hazama i have no idea what kind of effect having a lot of keyframes has on making your overhead reactable or not, but litchi has both that and a late identifiable frame going on for her. definitely intentional still

 

hakumen 6b - i’ve said it before but anyone who says they visually react to it is full of shit lol. at 18f startup and 5f total frames (the actual ‘hit’ frame is missing, sorry) the first clearly identifiable frames are at 7f (still mistakable for 5b, also included in the rar) and more definitively 10f (much clearer), which would only give you 8f to react. regardless, i’m 100% sure anyone who blocks this ‘on reaction’ are people who are reacting to hakumen standing/walking forward before doing an attack, and most people who say they can block it to me die when i do walk up 2b confirm.

on top of this, most anyone who’s trying to visually react to it will also die to walk up 5b>renka, which is something film (18-19 dan haku who got top8 at arcrevo) runs on people trying to block him. if haku could actually chain into 6b off of anything but 6a nobody would be able to block it at all lol

 

noel drive overhead and post first hit drive low - first 2-3 keyframes are in fact identical just like it looks in the game, and the next few frames are extremely similar as well regardless. first cleanly identifiable frame for each are probably frame 15 for the overhead (lol.) and frame 15 or 18 for the low, which is made easier by the fact she crouches so you don’t even need to look for the animation so much as her loss of height. lol

 

gauntlet hades - lower keyframe count than most overheads but its’ VERY visually identifiable, so that might be why. something like 5-6 frames for a ~20f overhead? too lazy to check how fast air gauntlet hades hits.

 

also i dug deep into the flamewar necronomicon and recorded tk badmoon and millia 6k too, it was right next to reviving hitler and summoning satan. anyway tk badmoon is RIDICULOUSLY well animated, having 12 frames of unique animation (counting jump startup etc) for an 18f overhead? something like that i don’t play gear. not trying to use millia blocker as hard evidence or anything but i know that tk badmoon feels ‘oddly easier to react to’ compared to 6k even though I don’t react fast enough to actually BLOCK IT, and the smooth animation is my best guess as to why that would be. move is bullshit

 

7 frames of anim for azrael’s 21f 5d. 7f is the first identifiable frame though all of them do a good job of distinguishing it from his 6/3d.

 

hakumen’s TK tsubaki and Zantetsu both have the trait of being very well animated but not having a ‘slash’ of sorts until it actually hits. i think tsubaki starts up in like 21f and it has 13f of smoooooooth animation counting the jump startup. zantetsu is 21f startup and has 8f of animation, smoothly moving, but has a big gap towards the end (1-3-5-7-9-13-17-21)

 

azrael 3/6d - 24f and 27f or so respectively. identical startup until the 10th frame. someone argued that you can tell which of the 2 he’s doing at this point by looking at his knee (not at where he puts his leg) but i’m dubious of it. i’m convinced you have to OS block low > high after visually confirming it’s 3 or 6d startup lol. or, y’know, mash or jump out. 14f for both are still very muddy and his knee is the only solid identifier. 17f on 6D is cleanly identifiable but at this point 3d still looks like he’s raising his leg. 21f of 3d distinguishes cleanly which one it is but you’re already getting hit at this point. this is probably. character’ dum

 

millia’s 6k apparently hits around 20f on sol and as far as i can tell in the video i took ky blocked it on the 21st frame. either way for 20/21f duration there were 10f of keyframes for 21-20f of animation, and it was very, very smoothly animated, you can see every little frame of her flip in it. i’m sticking to my guns that how smoothly animated/how many keyframes an overhead has in a consistent speed somehow affects how we react to them visually but gear is totally a different game from a different era so god knows if both of these moves are different from bb’s for other reasons, but I thought it’d be worth doing and bringing up!

 

that’s all i have a writeup about for the moment but feel free to download/discuss. i am biased and also cannot block anything listed here so yeah

 

http://puu.sh/7SeWO.jpg here’s a graphic i’m gonna use in the video showing where the animation keyframes happen for a few overheads ok bye

Share this post


Link to post
Share on other sites

I did an entire episode of IAD-TV discussing why reactionary defense is more or less impossible to play in ASW games.  There's a lot of data, numbers, and other stuff in the episode that explains why but the long and the short of it is that the animation of an overhead doesn't really matter in the long run because playing reactionary defense in these games is not only the wrong way to go about it, it's flat out just not possible unless you dedicate yourself to only watching one single move or option out of the many at any given point or situation in a match.

Not to say the data here on the number of animation frames isn't interesting, but you can't really take much from it for a multitude of reasons, including stuff like false positives (I.E. moves that animate similar to another move where one may be an overhead and another is something else).  How a move animates has very little impact on it's effectiveness compared to where and how you can use it based on the options of the character and matchup.

 

Share this post


Link to post
Share on other sites

I did an entire episode of IAD-TV discussing why reactionary defense is more or less impossible to play in ASW games.  There's a lot of data, numbers, and other stuff in the episode that explains why but the long and the short of it is that the animation of an overhead doesn't really matter in the long run because playing reactionary defense in these games is not only the wrong way to go about it, it's flat out just not possible unless you dedicate yourself to only watching one single move or option out of the many at any given point or situation in a match.

Not to say the data here on the number of animation frames isn't interesting, but you can't really take much from it for a multitude of reasons, including stuff like false positives (I.E. moves that animate similar to another move where one may be an overhead and another is something else).  How a move animates has very little impact on it's effectiveness compared to where and how you can use it based on the options of the character and matchup.

 

 

oh that's basically what i'm saying and what i'm moving towards with the info as well, plus it's what I thought before I went into it w it's really just data for the sake of data

 

sorry if that was unclear to anyone~

Share this post


Link to post
Share on other sites

Good writeup. This is the sort of thing people need to consider: when you can SEE the overhead, not how long it takes for the overhead to come out after you press a button. Just out of curiosity, what tools did you use to examine the frame-by-frame startup on these moves? I'd be interested in checking it out myself. Hell, maybe even make a frame viewer out of it, that'd be a cool project.

Share this post


Link to post
Share on other sites

Good writeup. This is the sort of thing people need to consider: when you can SEE the overhead, not how long it takes for the overhead to come out after you press a button. Just out of curiosity, what tools did you use to examine the frame-by-frame startup on these moves? I'd be interested in checking it out myself. Hell, maybe even make a frame viewer out of it, that'd be a cool project.

 

i just captured footage at 60fps and used a video player to frame advance and screencap everything at the appropriate frame. if it helps the replay system has a frame advance button so if there's things you want to look at you can grab a friend, load up a private netplay match, get the replay of you whiffing/hitting random stuff and frame advance the replay www

Share this post


Link to post
Share on other sites

I did an entire episode of IAD-TV discussing why reactionary defense is more or less impossible to play in ASW games.  There's a lot of data, numbers, and other stuff in the episode that explains why but the long and the short of it is that the animation of an overhead doesn't really matter in the long run because playing reactionary defense in these games is not only the wrong way to go about it, it's flat out just not possible unless you dedicate yourself to only watching one single move or option out of the many at any given point or situation in a match.

Not to say the data here on the number of animation frames isn't interesting, but you can't really take much from it for a multitude of reasons, including stuff like false positives (I.E. moves that animate similar to another move where one may be an overhead and another is something else).  How a move animates has very little impact on it's effectiveness compared to where and how you can use it based on the options of the character and matchup.

 

 

How would you recommend blocking mixups?

 

Whenever I play, I always feels like it's a 50/50 chance that I happen to guess right in overhead/low situations.

Share this post


Link to post
Share on other sites

How would you recommend blocking mixups?

 

Whenever I play, I always feels like it's a 50/50 chance that I happen to guess right in overheard/low situations.

 

For a really detailed response, check out the episode of IAD-TV i did on the subject here https://www.youtube.com/watch?v=NREjqcayD-w  even though it's done using guilty gear for all it's examples, the concepts apply to any ASW game.  The basic concept is that playing good defense comes in the same way most things do with playing an ASW game, by seeing more and more setups and matchups and learning how to adapt and anticipate.  The more you play a certain player or certain matchup, the more familiar you become with the options and setups available to that player  and character, the better you will start to anticipate and actually block or escape from different situations where you play defense.  The reason playing raw reactionary defense is so bad in these games is because there are simply too many options for the brain to process and stop using only reaction time.  Instead you have to be able to combine knowledge of the situation so you can eliminate potential offensive options (mixups), and start making educated guesses on how to deal with the position you are in on the defensive. 

I don't wanna derail this guys thread with my own work/thoughts so give that episode a look and if you've still got questions shoot me a pm or feel free to start a thread about playing better defense or what not!

Share this post


Link to post
Share on other sites

Yeah, I agree with Klaige completely, and that's what makes defense in these games a bit tricky to learn.

Being good on offense is about knowing the ins and outs of your character. Being good on defense is about knowing the ins and outs of the opponent's character.

Share this post


Link to post
Share on other sites

Just when I think im getting good enough at blocking and defensive to start telling myself im "good at it" here is yet another article at why Im not -- looks like school aside I have plenty to continue to study with *groan* 

 

ASW games at least keep me from being too bored with their crazyness... Still though I appreciate both sources of information... Its good to be --->constantly<--- reminded that your not as skilled as you'd like to be

Share this post


Link to post
Share on other sites

Learn to play everyone. You'll be fine. I've almost gotten halfway.

 

Jesus SWD...Your signature...

Share this post


Link to post
Share on other sites

Well, at least this isn't Guilty Gear where the overheads are visually lows.  I'm looking at people like Sol BadGuy.

Share this post


Link to post
Share on other sites

Well, at least this isn't Guilty Gear where the overheads are visually lows.  I'm looking at people like Sol BadGuy.

What in the world Sol overhead are you talking about?

Share this post


Link to post
Share on other sites

i think he's probably referring to Sol's Dust, even though that is definitely an overhead by the animation. The question is why it launches with that animation though. Overheads that look like lows would be like, Testament's Dust.

Share this post


Link to post
Share on other sites

i think he's probably referring to Sol's Dust, even though that is definitely an overhead by the animation. The question is why it launches with that animation though. Overheads that look like lows would be like, Testament's Dust.

Thank you.  I was thinking of the wrong Character for some reason.  But yeah, Testament's Dust looks like a low.  Jam's looks like a mid hit, and so on.

Share this post


Link to post
Share on other sites

Mashing YOLO 5/6A is best defence when in doubt.

But out of joke, it's always a 50-50 with lag.

Share this post


Link to post
Share on other sites

You can tell when Hazama's doing his 6a because he lifts his arm over his head, not just because he's standing like 6b

I can block it using the random replays in training mode, but it's too hard doing that in a real match

 

Another thing is i sometimes mix up Ragna's 6b with his 6c, even if their animations aren't that similar

Share this post


Link to post
Share on other sites

No mention of the impossibility of blocking all of Valk's wolf mix up options. Shame.

Share this post


Link to post
Share on other sites

As long as we're talking about overheads and lows that throw people off guard I'd like to give mention to Tsubaki's 6a and 6b and Ragna's 2d

Share this post


Link to post
Share on other sites

so long as we recognize that tsubaki's overhead is one of the shittiest overheads in the game, go ahead.

Share this post


Link to post
Share on other sites

so long as we recognize that tsubaki's overhead is one of the shittiest overheads in the game, go ahead.

It may be shitty but I was always make the awful mistake of expecting and then immediately getting it confused for her 6b animation wise.

Share this post


Link to post
Share on other sites

Wow, this is a great thread and I'm sad I didn't see it until today.  I had actually written up a 5-6 page article about reaction time and blocking quite some time ago, but after "focus testing" with friends and non-players, I decided it was too dense and needed revisions... and then overtime hit at work and I never went back to it.

 

Thank you Klaige for doing an episode on defense.  I've only watched part of one episode before since I'm always so time constrained, but I'm immediately going to go through all the stuff you've put out, and I started watching while posting this.  Also thank you Omex for this great animation breakdown.

 

Anyway, since this is a great topic for this subject, I figured I'd dump some of the neuroscience research I've done.  I encourage everyone to do their own research because you'll learn more and it's good to have people fact-check us.  I figure an info dump is worth it since some of my info differs from what Klaige has stated, and also because of the ability to easily reference raw numbers for people that care about that sort of thing without having to find that part of the video a second time (hint:  it's near the beginning of the video, and the video is worth watching in full a second time anyway).  For the record, I agree with Klaige that it's too hard to consistently defend purely on reaction in this game, and that's part of why this game is well designed.  But it is possible to strengthen your defense with reaction based techniques.

One more reference before I start is a link to shtkn's article about reading Frame Data.

 

 

 

 

Like Klaige said, the average median reaction time for a simple reaction is in the 200-215 millisecond range, or around 12 frames.  Look up tools like (http://www.humanbenchmark.com/tests/reactiontime/) and play with them.  This sort of test is exactly what Klaige was talking about for simple reactions.  Any time you're in a situation where you're expecting one stimulus and you have one response, you can react about this fast.  Block low until you see this specific dust, and then switch to high block.  This is totally possible, though you really need to consider how much time you have to react from the telegraph, not from the startup, and that's why Omex's post is really informative.  Some moves have significantly less telegraph to them and/or ambiguously look like other actions, like Kliff's 5D looking like he's blocking low.

 

However, there's more than 1 move in all of Guilty Gear, so now it gets a little more complex.  A complex or recognition reaction is when you have two different stimuli-reaction pairings.  Push the left button when the gray screen turns red, or the right button when the screen turns blue.  Now we're looking at a reaction time of about 384 milliseconds.  This is nearly twice as long to react, and puts most fast overheads in the "unreactable" category.  If instead of doing the "hold downback until you see the overhead" strategy you decide you're going to hit downback against Millia's 5 frame 2K on reaction, or back against her 18 frame 6K, you can't actually block either of them, even though you could react to her 6K if you were prepared and 2K didn't exist.  You know what's worse?  Adding more stuff to watch for in your reactions adds something like 40 milliseconds per stimulus.  If you factor every one of Millia's overheads and lows at all times, you can't block against her on reaction period, and would be better off mashing 1414141414141.

 

So this brings up a good starting point:

 

 

If put yourself into a situation that calls for blocking and you have to do it, try to simplify the stimulus/reaction pairing.

 

 

Klaige talks about this a little bit around the 40 minute mark.  If you know they can't gatling into an overhead out of a specific move and only have access to lows, you don't have to worry as much.  If your opponent is pushed out on block to where 5D/2D is the only remaining mixup option left and 2D is the only thing that'll actually reach, you don't have to worry as much.  If you can dead angle or burst unpredictably to get out of that situation, use it (don't get baited and waste the meter or burst on top of getting hit).  If you can IB > backdash, you might want to do that.  But sometimes you have to block, and when that time comes you want to be prepared.  The more options they have to run a mixup, the harder it is to block on reaction.  Thinking about your opponent's options for running mixups can simplify things for you, like the above examples, but you can potentially take it a step further.

 

One tactic that works for me is watching Millia's legs.  Think about just TK Badmoon and Haircar.  Both of them have her move away from the ground, and Badmoon / Haircar have really similar startup.  However, her legs stay near the ground with Haircar, whereas you wont be able to see them if she did TK Badmoon, so you can actually block this 16/19 frame mixup on reaction by switching to high block if her legs disappear.  Mixing 6K in makes it significantly harder, and you can't do jack-**** after the fact if Millia does Haircar FRC j.K, but at the very least you could deal with one really hard mixup this way.  You also could potentially train yourself to immediately block high on reaction to haircar after a delay of a few frames such that you would actually block it low...  if haircar is blocked low correctly there's enough hit-pause to mechanically switch between blocking low and blocking high and block the FRC j.K.  You're just acting on blind faith in that case though, because Millia could FRC into a low or a throw.  She could have also used any over her other mixup tools besides Badmoon or Haircar in the first place, so you're only really going to be able to use a tactic like this when you're expecting Badmoon or one of her lows.  So like Klaige said, it's just not possible to block reactively against everything.  However, if you have a read that the other player only does lows and TK Badmoon, you could probably block that more than 50% of the time, which is better than making the raw 50/50 decision.

 

 

 

Next up, there's also a psychological refractory period which is sort of cool and you'll want to know about for your offense as well as for learning how to get past being frustrated.  If you receive a stimulus that requires a quick response within 1/3 of a second of another stimulus that requires a quick response, and you weren't expecting that first stimulus, your brain lags.  Like, straight up, you can't even tell that you just lagged and reacted slow.  The slowdown is approximately 174 milliseconds on average.  If you're prepared for a high/low by doing your "I'm going to watch for this specific overhead while blocking low" tactic, you're not actually prepared for a low, or anything else really.  So when they do their 2P that's +2 on block and go for a tick throw, you react slower.  You'll have the move's recover time and hit-pause to react, which is probably going to be about 11 frames of hit pause + a few more for the move recovering, which would actually be enough to backdash or jump if you were expecting a tick-throw.  But you weren't expecting a tick throw.  So you start screaming "dammit I jumped!", but you probably didn't hit the button until maybe 12 frames after you got thrown, because your mind is still processing that the 2P recovered instead of gatling into something else.

 

So on offense, you have this little bit of info to throw in with your conditioning tactics.  Get them to expect a specific mixup and you can cause them to react slower to other things that normally don't work, which might get them to work.  Some gimmicks run on brain farts.

 

 

I also want to encourage people to try to use SB more.  If you can prepare yourself to slashback a move, you can slashback it.  Some moves are slow enough that you should somewhat reliably SB them on reaction, like Bandit Bringer.  Yes, that move has an FRC on startup that could bait your SB, but if Sol ever uses BB without 25% meter, it's slow enough that you shouldn't ever block it wrong except for maybe being unexpectedly crossed up.  That said, it's also slow enough that you should be able to SB it on reaction most of the time.  Brain farts can get in the way, but there will be situations where the opponent is committed to an attack that you can slashback, like a multi-hit move when they don't have meter to cancel it.  You also don't necessarily have to do a raw reaction.  Sometimes you can FD to push the other guy out right when the 5/2H starts, and then risk using the SB on the move right before it.  If successful you get the SB benefit and can maybe attack their whiff for knockdown or go for something that'll counter hit into a combo; if you fail the SB because you were late you'll have already blocked the prior hit and the next move whiffs so it doesn't even matter.

 

 

 

A few more quick things before I'm out for the night.  First, you can't just react faster and faster like some DBZ character.  I don't know where Klaige got data on top level reaction times reaching a super fast 100 milliseconds, and if you can link it I'd like to see it because I haven't bumped into that data during my research.  There's a limit to how fast neurons can fire.  You can definitely react faster than that 200 milliseconds, but you're probably only going to be able to shave a few frames off, and you wont be able to do it consistently.  That leads to serious ego depletion (aka you're going to run out of the ability to focus and get tired), and you probably need to be in the proper physical state to react like that in the first place...  Mihály Csíkszentmihályi's book "Flow" says to try to keep your heart rate around 140 for best effect.  A 100 frame reaction is probably possible; I certainly don't want to claim that it isn't.  But I can't see someone reacting that fast consistently for any given amount of time.  Maybe with stimulants, though I wouldn't condone or endorse that (I'll actually think less of you if you - drugs are bad mmmkay).

 

Second, you actually react faster than you can commit to acting, by about 21 milliseconds.  This is about 1 frame, so it's probably not ever going to help in a fighting game that runs 60fps, especially with hardware input delays and HD monitor delays.  Just a fun fact.

 

Also, the listed reaction numbers I gave before were for visual stimuli.  You react to sound in the neighborhood of 140-160 milliseconds.  If you have a move that always uses the same sound cue, it can give away the move depending on when the sound cue plays relative to the move startup.  I recently picked up BB:CP to give it a shot and noticed that with the VA set to English instead of Glorious Nihongo, Izayoi always says the same thing with one of her teleports (maybe even with JP language, but it's harder to distinguish without repetition if you don't speak it).  So you'd always know which teleport she's using within like 10 frames of her using it, and they last for 40+ frames... guess they really want you to die if you use them in a neutral situation.

 

 

 

And here are some references:

 

http://discovermagazine.com/2010/nov/15-the-brain-router-in-our-heads-processing-bottleneck

http://biology.clemson.edu/bpc/bp/Lab/110/reaction.htm

http://www.livescience.com/6041-reactions-faster-actions-study-finds.html

www.cbc.ca/news/technology/story/2010/02/02/tech-brain-action-reaction-speed.html

Share this post


Link to post
Share on other sites

Nice write up Bobman.

As for my numbers i just went through a couple different studies and general averages i could find online as I wasn't worries so much about absolute accuracy in the numbers as I was using them to drive home the point that multi-stimuli reaction based defense was a very poor way to approach the subject.  I remember specifically about the elite response times merely said that reactions as fast as 100ms had been recorded but it did not mention whether they were just freak incidents amongst broader tests or if there were actual people managing to put their median average on response time in that range.  

Basically most of the numbers for response time in my episode are general averages among the data i read about and wrote down, which definitely was not as thorough as you're own research here.  A lot of the stuff you wrote down here is great and some of it i actually wanted to explain in even more detail such as the sound reactions but i felt like i was already being too verbose on the topic when it was just one piece of me trying to explain overall defense in a single episode.

Either way this is good stuff, I feel like it's easy to pick up a lot of myths about blocking and defense in ASW games, it's why i really tried to put a lot of effort into that particular episode.  In my own eyes it's my best episode of IAD-TV in terms of the contents but i realize that defense is far from the most interesting topic for most players.

Share this post


Link to post
Share on other sites

×