noodalls investigates JG (and some GI)

noodalls

noodalls

[10] Knight
Noodalls investigates JG and GI

This was done using a programmable stick of my own creation, combining and Arduino Mega with two MC Cthulhus. Capture was with Hauppauge HD PVR, playback via avidemux on PC.

Fn refers to an input done of frame n. This style will be used throughout. For example, if I say that Natsu does A on F0, this means that on the first frame (frame 0) Taki presses A. Unfortunately, it appears on the screen on F1 (which is different to how Tekken displays command, where I've done most of my work). Again with Natsu's standing A, it is input on F0, it appears on the screen (and she begins to move in her animation) on F1, and it hits (big high mark appears and opponent loses life) on F10. This fits with her known frame data.



Done using Natsu vs Natsu, Natsu's A being the main move used.

First point of interest. Natsu doing A on F0, followed by A on F16 will give nothing, but input A on F0 followed by a second input A on F17 will give another standing A. This hits on F47, i.e. despite the input on F17, it is buffered and used as if it had been input on F37. This would suggest that SC has 20F buffer windows, however that's a discussion for another day. This doesn't change whether the A is blocked or hits, suggesting an absence of hit freeze like SF has.

Blocked standing A punished by a move input on F10 will hit on F39. This would make standing A -8 on block, which agrees with Future Press' frame data. Good.

To guard attack normally, you must be pressing or holding guard the frame before they connect. For example, if you are attacked by Natsu's standing A input on F0 (which will hit on F10), you must input A on F9 to block it. If you input it at the same time as it connects, you will be hit.

If for the same attack (opponent's standing A on F0, hits on F10). You press G on F8 and release on F9, you will get JG. You can get JG for a single frame touch of G all the way down to F3 (pressing G on F3 and releasing on F4 gives you JG). Pressing G on F2 and releasing on F3 will not get you a JG, it will get you hit.

Similarly, holding G for 5 frames will not give you JG. The longest you can hold G to get JG is 4 frames (release on the 5th). However, the release window is the same as for holding it for a single frame. For example, if you press G on F0-3 (where Natsu also presses A on F0) you will get a JG. If you press it 1F before Natsu presses A, and hold it for 4 frames, you will not get a JG.

Interestingly, if you press G on F8 (which would normally give you a JG) you can miss out on the JG by inputting a move on F9 (where Taki's standing A hits F10).

If you JG Natsu's standing A, you will still retaliate at the same speed. e.g. After a JG'd standing A, your retaliatory A will still connect on F39. For the attacker however, inputs are only accepted from F22. The fasted possible standing A following a JG'd standing A will hit on F52, i.e. 5frames slower than it normally would.

EDIT with further testing, it appears that if block stun is long, it will be replaced by the JG block stun, which is much shorter (~20F). I suspect for Taki's standing A the block stun is 19F, so this is not replaced with the JG recovery (would actually be worse off).

People could fairly easily test to see if JG always gives a bonus +5 to guarded moves.

EDIT

Actually, I've gone and tested with A,A B,B K,K as well as AA BB and KK. All of these gives +5 extra frames to the defender. It does not appear to depend on when the JG is input.

EDIT



Cooldown is another important aspect of JG. You cannot simply press G multiple times and expect to get JG each press.
If you press G on F0, the next time you will be able to JG is from F29 (holding it for 4Fs, likely that for a single press the earliest would be F32, though I didn't actually test this). If you try to repeat JG from F28 it will not work. Interestingly, it doesn't matter whether you held or released your initial G, you can release it on F28 and press G again from F29-32 and JG will still occur.


I looked brielfy at GI as well, the parry window is until F2-11.

tldr;
JG can be held for up to 4frames
You must release JG 1-6F before the attack hits.
The earliest you can repeat JG is 29F after the last time you pressed guard.
JG gives an extra 5F of recovery to the attacker. (for the seven moves I tested, including strings and single hits)
Recovery for block stun can be reduced by JG.
GI window is F2-11.


Edited summary to reflect new findings.
 
GJ noodalls

but one thing
JG gives and extra -5F of block disadvantage (for the six moves I tested, including strings and single hits)
Click to expand...
extra -5f, but comparing to what? Because if comparing to normal situation there you blocking the move then it would 't make sense.
It would mean that moves that gives, for example 0 on block, would be still save (-5) using JG on them. Just by watching some vids we can say that is not true.
also
JG does not change the blocking character's recovery.
Click to expand...
It's also not clear to me since we know that JG changes blocking character recovery (half of your post is about this), for me it even replace characters recovery animation from something that depends on the attacking moves to some animation of JG that is exactly the same for all attacks.

Pls, clear this because i know you and I know you check things before posting so maybe I'm misunderstanding something.
 
JG appears to add 5F of JG freeze to the attacker.

Normally Natsu's standing A on block recovers on F37, the opponent on F29. (-8). Natsu's standing A on hits also recovers on F37.
Natsu's standing A JGd recovers on F42, the opponent still on F29 (-13).


So, the defending character's recovery has not changed. The attacking character has incurred an additional 5F of recovery (due to the move freezing immediately post being JGd for 5F).

I tested on 6 different moves and found this was consistent. Can you give examples of where the difference is not 5F? Not because I doubt you, but because it would add more information to the puzzle.
 
I agree with CheeseOftheDay. Hayate from the future press guide also said to us that it's better to just guard a move which has a long recovery In my opinion Just Guard act like focus attack in Street fighter 4 so i don't understand this universal -5. If it's true just guard has no interest (risk/reward). Perhaps it is the reason why it cost no meter
 
If I set Astaroth to do 22_88B (-10 on block) then I set second action to "All Guard"

if I Just Guard 22_88B I can hit him with Patroklos 66B (i21) which is more than -15.

Astaroth 6K -14, Just Guard, hit him with Pat 66B (i21) which is more than -19.

I am hoping this is not an error caused by Training Mode functions failing to guard properly.

I can't say how it works. I would like to say that Just Guard negates blockstun while letting you take advantage of "whiffed recovery frames" but that's just fumbling in the dark. Both moves are ridiculously slow on whiff...
 
Thanks for the example. I think my initial description is both wrong and right, and not the whole story.

I looked at the Astaroth 6K example. Normally Astaroth recovers on F60 after 6K, after being JGd he recovers on F65.
However, normally the opponent recovers on F47. With JG, they recover on F36.

If we look at the Natsu standing A example, normally the opponent spends 19F recovering. The JG in the Asta example is 20F. Perhaps the JG recovery is either the normal recovery or JG recovery, whichever is shorter.


I would also suspect that there are only so many block animations, if we could list them it might be possible to work out exactly how much JG helps each move by class.
 
I have an HDPVR too if you need help noodals ^^ but i don't think i can reproduce just guard situations
 
Did a look at some of the different moves of Astaroth. The format is
(a/b/c)
which is
(initiation/recovery on guard/recovery on JG)


A (20/42/39)
B (21/43/40)
K (13/35/33)

66B (16/42/35)


6K (17/47/36)
66K (18/46/37)
6B (20/46/39)
66B (16/42/35)

4A (19/48/38)
4B (16/40/36)

4A+B (21/46/40)
3K(18/42/37)

So, at worst you are 7F better off (Asta standing K reduces defender's recovery by 2F, and the attacker does spend an extra 5F recovering.)

At best you are 16F better off (Astaroth 6K adds an extra 5F of recovery to Astaroth, plus the opponent recovers 11F faster.)

One additional note, which would be hard to get rid of using the current testing method, JG has no pushback, whereas normal guard has a variable amount. Given that I used standing A as a testing method, these values could be out be a frame or so (as when done from far, standing A appears to impact on i21, not i20.) However, that too is an important advantage of just guard, it leaves you right in their face where your attacks should connect at the earliest initiation frames.
 
c_nul, I've checked the training mode, and it does appear to be very accurate. If you wanted to check all the moves for a given chracter, you need to record at 60fps, (I beleive the m2ts format is the best way to achieve this). Set CPU to JG then AA and count how long after the move input is completed until the AA punishment hits.

Then repeat this process using guard followed by AA. The difference between these two numbers is the JG improvement.

I may well do this for Asta in total soon.
 
I think Daish said that Just Guard would give you variable frame advantage based on the move just guarded?

Are these results supporting what he once said, or is it something else entirely?
 
It seems like. With some moves (AAs and BBs) you can only do sneaky things, like slipping in Guard Impacts, auto-GIs, things of that nature. Maybe duck or quickstep punishing, I haven't tested.

With other moves, punishment potential is greatly increased.

Essentially, there are two versions of block data, the regular (Asta 6K -14) and the Just Guard version (Asta JG 6K -30).
 
AA_BB_KK of natsu all gives the same block stun animation and it can be true for most AA, BB's in the game. Just like in all 3d FG there are dozens of block animations and there is no sens to find a formula for what advantage JG will give to all moves because because it will depends on type of block animation and recovery of the certain move.

In the future the only case will be what advantage you will get from JG'ing these 'pain in the ass' attacks/strings that normally would be hard to deal.

For me I've tested it with alphaPata 2A and 9B with are really safe on block (-4, -5), don't give small block stun animation and after JG they can be punished with 13, 14 frames moves or something like that.
 
Can you check if it's possible to make JG safe by holding G at the end of the JG window? (Hold G 6 frames after release)
 
ShenYu said:
Can you check if it's possible to make JG safe by holding G at the end of the JG window? (Hold G 6 frames after release)
Click to expand...

I don't think you can do this. When I tried it would just guard normally and ruin an otherwise valid JG.
 
Schneider-X said:
I think Daish said that Just Guard would give you variable frame advantage based on the move just guarded?

Are these results supporting what he once said, or is it something else entirely?
Click to expand...

Just Guard has a set recovery animation every time. The move you're just guarding, obviously, does not. Therefore just guarding a move with slower recovery gives you more advantage. That's what Daishii was talking about.
 
You must log in or register to reply here.
Back