Got any Questions about SF3:NG, 2I, 3S, or 3SOE?

[Maj]

That does suck. Well, at least you can reset the "orange" gauge with one of those attacks before a combo to create a baseline you can reproduce later.

Why would they want to make randomness so prevalent in charge partitioning though? Seems rather cruel.

[Pokey86]

i have 5 trials left.

Yun with 12% health VS Alex - I get rhwon & die, i suck with Yun

Ryu Vs Akuma 0 Health - I haven't even tried it

Evo moment 37 - I haven't even beaten the first rep of legs yet

Dahlsim wannabe's 5th trial - 6 of those goddam arm things in the corner, i got to 4 once

Yun 5 - Kill me now, it's probably for the best.

Shoutouts to Ibuki trial 5 for breaking my fingers, i almost cried when i got a success on that, which is funny cause my inputs felt way off. why, for the love of god WHY are jump ins to standing moves so much harder in this game!?

[Maj]

People have found a couple of AI exploits for Handicap Trial 5.

http://www.youtube.com/watch?v=CA_snKen1z
http://www.youtube.com/watch?v=PRYF2L96ZPM

As for jump-in followup timing, SF4 is the exception there. Jump attacks never used to produce so much hit stun in Street Fighter. Not sure why SF4 turned the whole mechanic upside-down, but suddenly you can jump into the corner, crossups create a massive gap, and you can combo after kicking Zangief's mohawk. It took a while to get used to.

[Pokey86]

you're probably right, i think i have to learn in reverse. Before SF the other SF game i combo'd relentlessy on was SF EX series, if i recall the jumps ins were pretty lenient in that to.

I've always seen the jump in part of a combo as the "tacked-on" part, not exactly hard, just annoying to reposition yourself when you fuck up. Now that i actually have to time the hit after the jump in, it;s beginning to irk me. something i'm just going to have to work around i guess.

& yeah, it seems comboing a cross up now is very difficult, i've yet to actually do it in a proper match. I've still been hitting the low attack direclty after, however i just noticed it's not comboing, then again cause i'm in England i'm no doubt playing pretty crappy people who arn't blocking propperly. God knows i scrub out alot & try to mash SRK on wakeup... which doesn't really work in this game. (Reversals seem tighter, SFIV must've been hella-lenient)

[onReload]

i have 5 trials left.

Yun with 12% health VS Alex - I get rhwon & die, i suck with Yun

I just did this one; back off and build meter with low strong and palms; alex will usually screw up and eat a palm or two for damage and you gain some meter; he'll do that stupid superman dive (d+HP) and miss you, rolling right in front of you... it has horrendous recovery; when you see that, go for a combo that gives you space again (low strong mp dash punch, target combos, whatever). when you get the chance, go for a simple combo into super, then juggle with MP shoulder, HP dash punch...then go back to palm spam. only took a few tries.

for the yun gej combo, yeah, well, i'm no good at those, but remember that in GEJ mode, you can cancel his palm into another special, so if you have trouble doing the HK Nishoukyaku after the Kobukushi, just try inputting it like a cancel.

also, some characters can get crossed-up in the corner, but most can't. feels so nice to have my back to the wall~

[CPS2]

(Reversals seem tighter, SFIV must've been hella-lenient)

Yeah I tested reversal windows in a bunch of games a while back. IIRC 3rd Strike is 2 frames, for one button anyway, not including negative edge. Wakeup parry is basically "easier" than a reversal in SF4, because the window is bigger, but there's also a lot of risk to go with it, and it's easily baited.

[Rufus]

That does suck. Well, at least you can reset the "orange" gauge with one of those attacks before a combo to create a baseline you can reproduce later.

Why would they want to make randomness so prevalent in charge partitioning though? Seems rather cruel.

Yeah. I find myself wondering whether up-charge moves will reset the l/r charge partition meter (though I suspect they do). I'm guessing QCF moves don't reset it.

...(Reversals seem tighter, SFIV must've been hella-lenient)...

It feels slow and sloppy to some of us. I remember trying to do some of the trials in vanilla SFIV and it was *difficult* to avoid doing unwanted dragon punches with Dan in one of them.

[onReload]

messing around with frame advance (in cps3emulator.exe), haven't started scripting anything, but here are the beginnings of my findings (inspired by discussion on cmd vs regular throws on #capcom)

Playing with 12 and Hugo, both in idle stance flush against one another:

12 cannot tech Hugo's 360+P or 360+K with his neutral grab at all.

For 12's neutral throw, pressing nothing after the initial input has the throw do 6 hits.
Mashing LP each frame does nothing.
Mashing LP and a different direction each frame (in a 360 motion), reduces his throw to 3 hits.

Interesting note: You can do Hugo's SPD with 4 frames of input (Toward, Down, Back, Up+Punch), but you can't do Makoto's command grab with 3 frames of input. (Toward, Down, Back+Kick). Any idea on the minimum amount of input?

This is all contingent on the idea that I understand frame advance, which, in all possibility, I don't. So if someone would help me out here, this is what I think:

When paused, the emulator allows me to move forward one frame by pressing Enter. Any button I'm holding is considered to be inputted on the frame the Enter key jumps to. For instance, for 12's s.LK, which should be 3 frames of start-up, I'm doing:

[ ] LK+Enter [ ] Enter [ ] Enter [ ] Enter [ ] Enter [ ]
If each set of brackets indicates a frame of animation. So, each frame should be:
[nothing], [frame with input pressed], [1st frame of startup], [2nd frame of startup], [3rd frame of startup], [hit]

...and what I'm seeing is:

[12 is idle], LK+Enter, [nothing has changed], Enter, [12 begins his move by extending his leg], Enter, [12 continues to start the move with the same frame of animation], Enter, [12 continues to start the move with a slightly more extended leg], Enter, [12 extends his leg fully, and Hugo's health bar drops a few pixels indicating the s.LK has connected]

So...simple question here: Am I doing this correctly? If so, I think I could even show this for other newbies with some simple pictures instead of shitty ASCII art.

[CPS2]

I think 12's LK hits on the 5th frame, so startup should be 4 instead of 3. Your startup number is the same as what it says on the SRK wiki, but they have cr.MP as +4 (which is also what I got), so according to their frame data it should link, but I'm pretty sure it doesn't link as it hits on the 5th frame... cr.MP vs a crouching opponent is +5, and I have that connecting as a 1 frame link.

It's a bit confusing though because you can clearly see the health bar go down on the 4th frame, and a red hitbox appears on the 4th frame using Dammit's lua script in FBA-RR 0.0.7.

Basically the way I check frame data is, do any attack and have both characters jump, whoever jumps first has frame advantage, and then count how many frames until the opponent reaches the same jump animation. It can get confusing with 12 and Hugo cos 12 has 1 extra pre-jump frame than normal, and Hugo has 2 extra pre-jump frames than normal, but you can work out frame advantage with them just by being aware of that. Then count your startup frames, but check if your numbers are right by using link combos. It's also difficult with SF3 because there aren't many links... but anyway I'm pretty sure 12's LK hits on the 5th frame, if that helps...

[Maj]

Um, you're kind of wandering into the murky abyss here. The truth is, nobody talks about input lag or even really know how much of it there is. We could measure it on an arcade machine, but there's so many different hardware setups that you never know if it's the buttons or the wiring or the software. We could measure it on an emulator, but you never know if it's the game or the emulation that's lagging.

Personally i think it's better to avoid that stuff and focus on relative timing instead of absolute timing. Scripting pretty much takes care of that for you, because the timing between the two characters in your script is always relative. You don't really care if the script starts on frame 0 or frame 1 or frame 3, because whatever that point may be, you're in complete control after it happens.

[CPS2]

Um, you're kind of wandering into the murky abyss here. The truth is, nobody talks about input lag or even really know how much of it there is. We could measure it on an arcade machine, but there's so many different hardware setups that you never know if it's the buttons or the wiring or the software. We could measure it on an emulator, but you never know if it's the game or the emulation that's lagging.

Personally i think it's better to avoid that stuff and focus on relative timing instead of absolute timing. Scripting pretty much takes care of that for you, because the timing between the two characters in your script is always relative. You don't really care if the script starts on frame 0 or frame 1 or frame 3, because whatever that point may be, you're in complete control after it happens.

This is why I start with frame advantage because measuring startup and looking at animations is relative, until you compare it to hit advantage and what actually links, and then I think you can say for sure what the startup is.

[onReload]

Fuck. I really hope it's not input lag. I've tested the findings a few times, but with savestates, any possibly dropped inputs would stay the same each time I reload, right?

I remember (I think Xenozip maybe?) someone did a video that showed Makoto and Alex both doing their command grabs at the same time, which in theory, require the same input times (both are half circles then one button), but for some reason, Alex's seemed to come out a lot less frequently.

Looks like I'm gonna have to script this stuff, then, and try it at different times during a round. Basically I'm trying to find all the weird throw tech situations, like how I've seen characters tech Hugo's running grab, or Makoto's command grab.

CPS2, I know about the jump thing for frame advantage, and thanks for letting me know about 12 having 1 frame of extra prejump, thought it was only Hugo...poor 12. He's not even big. I'm working with data from the old karathrow.com and now some french site that has a bunch of tables for it. the SRK wiki is pretty messed up from my memory. But am I right in thinking that the frame after pressing enter doesn't count toward startup, as it's the first frame where the game recognizes that I've pushed anything?

[Maj]

This is why I start with frame advantage because measuring startup and looking at animations is relative, until you compare it to hit advantage and what actually links, and then I think you can say for sure what the startup is.

Yeah, that's a good approach to take. Links are always relative, because even if there's input lag, you can press the button earlier (while you still appear to be in recovery) to eliminate that lag from the equation. Then you're left with pure startup vs frame advantage; actual game system limitations.

In fact you won't even notice any of this stuff while you're scripting, because you'll just reduce wait numbers until something doesn't come out. You always find the tightest relative timing between any two inputs. The same is true of frame advance, as long as you're going by pure trial and error.

It only gets confusing when you use frame advance to eyeball timing - which is everyone's first instinct, understandably. But you either have to get used to input delay and take it into account, or avoid that method altogether.

Anyway i wish i could give more specific advice, but i don't really know how throw teching in SF3 works. Being able to tech command throws seems weird to me in general.

[CPS2]

CPS2, I know about the jump thing for frame advantage, and thanks for letting me know about 12 having 1 frame of extra prejump, thought it was only Hugo...poor 12. He's not even big. I'm working with data from the old karathrow.com and now some french site that has a bunch of tables for it. the SRK wiki is pretty messed up from my memory. But am I right in thinking that the frame after pressing enter doesn't count toward startup, as it's the first frame where the game recognizes that I've pushed anything?

I want to say yes, and add that you should also count one extra frame after where you've stopped counting. But where you start and stop counting doesn't matter (it's relative), as long as the number is meaningful.

Also I thought command grabs could tech normal throws, but not the other way around? I haven't tested it though.

[onReload]

Yeaaah I've heard that it's technically the Command Grabber teching the Regular Grabber, but what's the difference, really? I can upload each frame of the whole shebang a bit later. (Have concluded for certain that it is possible, on some damn frame, to have Hugo's neutral throw tech Makoto's Karakusa)

Hopefully the HCB+K not working in three frames isn't just my using a PS2->USB converter, because when I try Keyboard-controlled Hugo's HCB+K grab in three frames, it works without a hitch! Yes, I know, I should be using the same input for both characters, but...well...this was all started up right now outta nowhere, so I'm gonna keep going this way for a second.

[CPS2]

Yeaaah I've heard that it's technically the Command Grabber teching the Regular Grabber, but what's the difference, really? I can upload each frame of the whole shebang a bit later. (Have concluded for certain that it is possible, on some damn frame, to have Hugo's neutral throw tech Makoto's Karakusa)

Hopefully the HCB+K not working in three frames isn't just my using a PS2->USB converter, because when I try Keyboard-controlled Hugo's HCB+K grab in three frames, it works without a hitch! Yes, I know, I should be using the same input for both characters, but...well...this was all started up right now outta nowhere, so I'm gonna keep going this way for a second.

Basically one of them is going to hit first, and if the normal throw hits first there's a window afterwards where the tech can be inputted, and during that window if the input is a command throw (the active part) then the tech should work. I think the murky area would be if a command throw and normal throw hit on exactly the same frame (they're going to have different startup numbers so you'd want to check sequential frame timing for each type of throw to be sure, and you can't use link combos to varify the startup so you pretty much have to trial and error it). I'm not sure what happens and haven't tested any of that stuff as far as I can remember. To show that a normal throw can tech a command throw you have to show that the command throw is hitting first, or they're both hitting on the same frame.

I think all this stuff would be better scripted than with frame advance, because it's all input based and not visual at all. If it's faster to use frame advance as a one off thing then do that, but probably keep some notes... I'd advise using FBA-RR and recording a movie file, because then at least you have a number on each frame and can use that for reference.

[onReload]

Yeah you're right about that last part. the way i've been concluding what can tech what is by calculating how to have their first active frames be simultaneous, and i keep shifting it around to see if it would make a difference, but if only shift one character's throw attempt around, instead of trying them both, well, then...argh. yeah it's gonna be easier to script, but my body is trying to run away from that idea. I've got time before the pizza arrives, anyway...

[Buttermaker]

You can test an emulator's input delay in the game's service mode (input test). CPS-3 has zero delay in MAME and FBA.

[onReload]

Never thought about doing that...but what I'm more concerned about is dropped inputs, things like that, that you can't really test for as easily.

...but has anyone ever heard of another game (with mashable throws) that allows the offensive character to reduce hits all by him/herself?

[Q80Warlock]

Here is an interesting info on SF3's juggle system, you can actually know if it's possible to followup with more hits depending the the combo hits message. If the combo message appears immediately no more juggle is allowed if the combo message appears after a slight pause it mean you still have juggle points.

[onReload]

yup, i love that little engine quirk. i've been surprised by it a few times in 538 videos

[CPS2]

Unfortunately it doesn't work with dizzy combos, and they have completely different rules for what juggles and what doesn't.

[onReload]

the rules aren't completely different, are they? i thought it just became that no move will reset the opponent

[his1nightmare]

To clarify it for real... :
There isn't simply one but a crapload of rules for the possibilities of juggling opponents. I will try to list the momentarily needed ones in a logical order.
- The "2-frame-rule"; every attack which has 2 frames startup or less (on the needed hitbox) will combo if cancelled from a connected normal, without exception; this is the main reason why you can do things like s.hp xx SA2 with Ryu but not s.hp>SA1.
- Rather an exception than an extension to the first rule following applies: Certain attacks in the game leave the opponent in a juggle-able state for longer than 2 frames. An example is the first hit of Remy's c.s.hk, or s.hk of Alex.
-> Both of what's written above has pretty much absolutely nothing to do with juggle points, as they apply even when 0 points are left, but it is optically visible by what Q80Warlock wrote.

Mathematically you can have even less than 0 juggle-points in this game,... an example of what happens in this case would be the following:
If you launch Chun-Li (or any other character with a very huge vertical hitbox while falling down) with Yang and then perform s.lk xx SA2 as late as possible, it will combo, but because s.lk has absolutely no juggle-points, the SA will not combo into the follow-up-hits. So everything ends in a 2-hit combo and a whiffing super though the first hit connected without any spottable problems.

- If an attack leaves the opponent airborn with juggle points left, this attack can combo into certain supers with more than 2 frames startup; this is similar to the second "rule" I wrote, but is obviously a different mechanic and covering another aspect of gameplay. This is why Ryu can combo c.s.lp xx SA1 on airborn opponents, not just reset them and then perform the super.
- Everything written and also any other rules does not simply apply to attacks, but to hits, this is why I wrote i.e. "first hit of Remy's c.s.hk", some attacks (basically) don't have juggle points at all on the first hit of a 2-hits normal, but on the second (or the opposite), so the second connects only because of what I summed up above. This leads to attacks which leave the opponent with more/less juggle points depending on how many hits of the respective normal/special/super connected. The most common example is Ryu's f.mp, another one would be Remy's EX Cold Blue Kick. Both times you can juggle longer if the first hit doesn't connect.

- Expanding the very first, "2-frames-rule" again, actually it should be called "8-frames-rule". Why? Because it's not the case that the follow-up of an attack has to hit the opponent within 2-frames after hit-freeze, but precisely within 2+6(the amount of frames the hit-freeze lasts) frames from the moment onward the opponent was actually hit. This plays a role when you i.e. combo from projectiles shot a while ago, so you have 8 frames of time from the moment your projectile hit the opponent to extend the combo, which makes it possible to juggle opponents after they were hit by projectiles which leave them with 0 juggle-points left.

- Pretty much everyone knows, but I will simply state it too: All attacks in this game "add" either 0, 1 or 6 juggle-points upon launch/first juggle or subtract any number from 1 to 5, but always leaving at least 1 if 2 or more were present. I'm not 100%ly aware of how this part of the mechanics works, but certainly some attacks take away more than 1 juggle-point when juggling someone filled with 3+ points and also some attacks seemingly take away more points than usually when used multiple times (in a row). I guess no one knows exact values and calculations because there is simply no need to discover something this bizarre.

[onReload]

Anyone know why you can't do Crossup Air Tatsu, SA1 with Akuma?

[his1nightmare]

The 180° spin backwards has no juggle points.

[onReload]

Hmmm. Not sure why they would do that, Ken's EX does...but it costs meter I guess...and Akuma would get the whole super that way. It's just weird

[his1nightmare]

During pretty much all of my match sets, I seemingly pulled off, or saw being pulled off, rather hard karas without the intention of doing them. I.e. f.hp kara Power Bomb with Alex.
I tried to get to the bottom of this, and found an answer: In several occasions this game forbids you to perform anything but a normal (no special or super; didn't test for jumps, taunts, throws, etc.).
For fun, I repeatedly performed c.s.lp>SA2 with Alex (1-frame link). Using pianos, Alex always performed a normal at the last frame of the lp's recovery, causing the super to receive a kara (mostly f.mp).
In the end I put the things stated above in a video, showing the described mechanics with EX Flash Chop and SA2: http://www.mediafire.com/download/pqa1z ... sfiii3.avi

If anyone already knew, is there something to be added to this, boosting my understanding a bit further?


Topic #2:
Akuma can occasionally combo c.light xx Hadouken. And by occasionally I mean randomly, no c.light receives any frame advantage from crouching opponents, and I am not talking of stunned ones (usually doesn't work either).
The Hadouken comes out exactly 1 frame too late to combo, yet it works every now and then. Saw it happening on crouching Chun-Li and standing Gill (at least from what I can remember right now).
Does anyone know why?

[Maj]

Pretty funny reading some of the old mystery posts in this thread. I'm pretty sure "Progressive Hit Frame System" or whatever slick name they were trying to put on the back of the box just meant "more hitboxes that update every frame."

Anyway two random questions that i never got around to looking into:

1) NG/2I/3S: Why does interrupting Ryu's HCF+K stepkick make the opponent fly toward him?

2) NG: Is there a trick to doing those ShinSho loops or is it just dummy-specific? I tried to do that against Alex once and couldn't get more than three j.HP reps.

[Maj]

Before!? Is that the same with Chuns as well?

Yes, which makes the whole thing a lot more impressive.

that's a common misconception
there is a two frame window after the super freeze that you can parry
it's just easier to parry before it

One more question. If you look at Evo Moment #37 clips, you can see Daigo stutter-stepping forward once he gets down to zero life and has no other option except to parry Chun's SA2.

We know that parry input is active for something like 10f if you tap F/D, and it gets shortened to like 5f if you hold those directions. I guess it's a penalty for not being precise enough, or maybe they didn't want to give every walk/crouch movement the full parry window. Anyway after those 10f, there's a pretty long cooldown where you can't attempt another parry, maybe like 15-20f. I don't remember the exact numbers but they're roughly in this ballpark.

When Justin activates SA2, you can see Ken is right at the beginning of a tap F attempt. Daigo looked to be doing a bunch of them in anticipation, so he almost definitely got lucky with the timing.

My question is, if he hadn't got lucky with the anticipatory attempts and Chun happened to activate super during one of the cooldown windows, would Ken still have a chance to go for that post-superfreeze 2f attempt?