3rd Strike Frame Data


my understanding of it has always been the exact opposite. ie. that you can cancel dash “recovery” with anything (jump, block, attack, etc) except another dash. i based this off the fact that makoto has to wait a noticeable amount of time between two consecutive dashes and esn’s chart shows that she has the longest “recovery” frames, so i put two and two together and ended up with this theory. could be wrong but this is how i’ve always interpreted whatever dash “recovery” is.


@ESN @telesniper

After parrying ibukis toward mk, why can yun punish it with lp lk mp TC only when she couch guards (has joystick in down back). if she stand guards, the TC will not punish her toward mk on parry.

I am aware you can select the recovery state of some special moves but this move is a normal move.

If you guys don’t know why this occurs that’s ok, just curious. Thanks


Yuuki- Very interesting…
Maybe the game treats towards forward more like a jump-in attack? I seem to recall that you can choose your recovery state on your airborne attacks.

Semi related can ESN, tele, etc also look into ibuki twd’s rh frame data. It is listed as -4 on block but testing in OE it seems to be completely safe as ibuki can block both mak’s sa1 and ken’s sa3.

Edit. Further testing on towards RH reversal gigas can grab it (holding up after executing move)
Yangs sa2 hits only if Ibuki tries to crouch block it similar to Yuuki’s example.

My guess is that when executing an “airborne” attack you have one less landing and/or recovery frame if you chose to stand block instead of cr block.


question…is there a way to determine hit advantage of special moves? i cant seem to find it on ESN’s site :frowning:


you can find that information at game restaurant.
to find the frame data for a character, select the character picture on the left table. once youve gone to the character page, click 一覧 on the far right of the header table.

under H&G硬直 you’ll see the advantage for moves that are guarded (G) moves that hit the opponent standing (立) and moves that hit the opponent crouching (屈)
the D in that information stands for “down” as in the move knockdown the opponent.

hope you find this helpful. @djdjw

let me know if you have any further questions on how to read the frame data or any part of the website.


tyvm sir :slight_smile: very helpful


demon 0 frame!



This was mainly a response to Hax, but this portion fits better here…

I assume you’re referring to the red damage on Chun’s life? Damage actually occurs one frame before your reel animation, meaning you need to, in effect, be blocking/parrying at least 2 frames before the move connects visually (5/6 realtime with the reported input lag). Honestly always wondered if that actually has something to do with the varied frame data reported, some being early by a frame and such… maybe they counted until the red? Or factored in how many frames you’d have to block it?

@ESN How did you measure start up again? Before the red? When the hit color showed? Would this mean that potentially start up could be accurately measured as 1 frame before the red, as on that next/current frame, it’s a “done deal”? Or another frame earlier?


The startup definition i used is:
the number of frames from when the game registered an attack until the frame when opponent is considered as being hit, without including that last frame. On screen, it will corresponds to “red bar - 1”, and “sparks/char hit - 2”.

Full example of how i processed for Gouki cl.LP, first number is current internal frame number of emulator, second is P1 attack variable, third is P2 attack variable (which also show if he is being hit):

  • 79471 : 0 / 0 --> button pressed, startup = 0
  • 79472 : 4 / 0 --> P1 attack registered, startup = 1
  • 79473 : 4 / 0 --> startup = 2
  • 79474 : 4 / 0 --> startup = 3
  • 79475 : 4 / 1 --> P2 registered as hit, red bar, hitstun = 1
  • 79476 : 4 / 1 --> sparks/char bit, hitstun = 2
  • 79477 : 4 / 1 --> hitstun = 3
  • 79478 : 4 / 1 --> hitstun = 4
  • 79479 : 4 / 1 --> hitstun = 5
  • 79480 : 4 / 1 --> hitstun = 6
  • 79481 : 4 / 1 --> hitstun = 7
  • 79482 : 4 / 1 --> hitstun = 8
  • 79483 : 4 / 1 --> hitstun = 9
  • 79484 : 4 / 1 --> hitstun = 10
  • 79485 : 4 / 1 --> hitstun = 11
  • 79486 : 4 / 1 --> hitstun = 12
  • 79487 : 4 / 1 --> hitstun = 13
  • 79488 : 4 / 1 --> hitstun = 14
  • 79489 : 0 / 1 --> hit advantage = 1
  • 79490 : 0 / 1 --> hit advantage = 2
  • 79491 : 0 / 1 --> hit advantage = 3, P2 can do something (jump for example) so it’ll execute on last frame
  • 79492 : 0 / 0 --> P2 starts to jump

So we got:

  • startup = 3
  • hit advantage = 3

On a side note, iirc that’s not the same on OE. For this same example, sparks would appear only at 79477 while “char hit” would still happen at 79476.
Will check this again tonight if i can.


Awesome, thanks for clarifying that again! That’s interesting about the difference on OE too!

Speaking of… I wonder about certain OE chains, maybe it’s just a “game speed” issue, but for instance I can Jab Jab Short with Mak so much more consistently on FBA, GGPO, Shmup and Groovy than I can on OE. On Emu I can just double tap and be pretty sure it will combo, but on OE it feels like the pushback of Jab either happens too fast, or the chain window is just smaller, or perhaps happens 1 frame too late (making it a 50% “chance” to not combo) for that particular move… Or it’s an issue with self cancelling moves on OE? Haven’t any issues with Jab - Short, just the Jab Jab.

Probably just me messing the timing up, but I was curious if the self cancel window is the same on r608 as it is on 512. Thanks for that answer again.


My bad, i was a little wrong about the OE differences.
After checking with 120fps videos, i got (tested with Chun far HK):

  • no differences found between chars animations (sprites change at the exact same frame)
  • sparks/char hit happen at the same frame on both version
  • OE red bar happen 1 frame later than fba-rr, same frame as sparks
  • P1 super bar is filled 1 frames later on OE, same frame as red bar

That alone makes confirm harder on OE for those watching for super bar or health. (not my case)


Needing help figuring out how towards+jab combos into st mp with dudley. I tried reading the frame data from ESN’s site but i still dont quite understand if anyone can help please check it out for me. frame data pertaining to the towards jab being only +1 on hit and st mp having the 4 frame startup how does it combo? i was just in training mode then i noticed that with the correct timing the st mp would combo. I was just messing around with towards jab since its never used but it has super cancellable properties and other boxes colored in the frame data list. came up with this combo and was just curious thanks in advance :slight_smile: http://www.youtube.com/watch?v=Lm4sX0ZN9NY&feature=youtu.be


It’s a target combo.



0_0 totally missed that…sorry guys :frowning: thanks iso


dekuscrub everybody! :stuck_out_tongue:


ok so besides ->+jab , mp target combo, do any other of dudleys target combos allow the frames of st mp to cancel/link into ducking super? i have tried but no success :frowning:


According to the frame data, Oro’s close standing jab is +5 on hit, and his close standing forward is 4f startup. That should be a 1f link, right? I’ve never gotten that link to work. Is this an error in the frame data, or am I doing/reading it wrong?

I put the training dummy on “jump” and “parrying” in 3SO, and I noticed that if I mash jabs to get the dummy at the right height, I can time a cs.MP and it they just get hit by it, sort of like a Darkstalkers unblock. Are there unparry frames on landing, or is this just a training mode glitch?


Oro’s close forward might have two key frames for hitboxes, one up real close and the second one out more towards his knee which makes that link impossible. I don’t think his close jab has the same 5+ property so that’s probably out of the question too.


since you’re waiting the full time till the jab is done youll be as far away as the pushback sends you i imagine (for that move).
so isnt it possible his mk just doesn’t reach?

i mean think about it. that would be the kind of confirm everyone who plays oro would use all the time. have like a bajillion frames to see the jab hit and then react and if it doesn’t hit, you’re still at advantage anyway.

thats like some sf4 shit. punch/kick repeatedly until something hits then just keep pressing buttons for more damage.


I’m not sure I understand. Do you mean that it’s just hitting a frame later or something?

Close jab pushes him out far enough out that he can’t get another cs.jab without walking back in, but cs.MK activates really far away for a “close” move. It’s definitely in range, it just won’t link for some reason. If Dander is telling me that I’m just getting pushed out far enough that the MK is hitting a frame or two later, that makes sense, but it seems kind of weird.

Yeah that’s why I want to do it!

Oh well. Any idea about the parry thing?