Delta Red Top Secret Briefing: Cammy Frame Data and Hitbox

Classic Fighting Games Ultra Street Fighter IV
1

Hi fellow Cammy players, I just downloaded the leaked SSF4 hit box video and I have analysed the Cammy move sets frame by frame and here are my findings:

Note: I don’t know if the leaked hit box video is the final build or not so that’s the Caveat… The video was recorded @ 60 fps and I counted 1 frame at a time (I am so disappointed that Cammy’s Cannon Strikes were recorded with a Camcorder @ only 30 fps, so I couldn’t do any accurate analysis on it…) Sigh…

(s) startup frames
(a) active frames
® recovery frames

(h) frame advantage on hit
(b) frame advantage on block

Green – with the positive frame advantage, it is possible to LINK into another move with equal or less startup frames for a combo
Red – with the negative frame advantage, Cammy is vulnerable to 3 frame attacks such as Ryu’s SRK, cr.lp (3 frame startup) etc…

Normals

standing far:

:lp:
LP: 3(s), 3(a), 7® +4(h) +1(b)
:mp:
MP: 7(s), 4(a), 14® -1(h) -4(b)
:hp:
HP: 7(s), 2(a), 15® -1(h) -3(b)

:lk:
LK: 4(s), 3(a), 6® +5(h) +2(b)
:mk:
MK: 7(s), 2(a), 13® +2(h) -1(b)
:hk:
HK: 8(s), 2(a), 22® +2(h) -6(b)

standing close:

:lp:
LP: 3(s), 2(a), 8® +4(h) +1(b)
:mp:
MP: 4(s), 4(a), 15® +6(h) +2(b)
:hp:
HP: 4(s), 4(a), 15® +3(h) -1(b)

:lk:
LK: 4(s), 2(a), 8® +3(h) +1(b)
:mk:
MK: 4(s), 3(a), 17® +4(h) -6(b)
:hk:
HK: 9(s), 5(a), 24® -7(h) -11(b)

crouched:

:d::lp:
LP: 3(s), 2(a), 8® +4(h) +1(b)
:d::mp:
MP: 5(s), 4(a), 10® +3(h) 0(b)
:d::hp:
HP: 6(s), 4(a), 12® +7(h) +2(b)

:d::lk:
LK: 3(s), 3(a), 9® +2(h) -1(b)
:d::mk:
MK: 6(s), 4(a), 12® +1(h) -2(b)
:d::hk:
HK: 7(s), 2(a), 24® kd(h) -8(b)

Neutral Jump:

:u::lp:
LP: 4(s), 4(a)
:u::mp:
MP: 5(s), 4(a)
:u::hp:
HP: 5(s), 3(a)

:u::lk:
LK: 5(s), 5(a)
:u::mk:
MK: 5(s), 7(a)
:u::hk:
HK: 5(s), 3(a)

Note: All jumps attacks, during the startup frames, Cammy’s legs are invulnerable to projectiles
Note: All jumps have 4 pre-jump frames before airborne and during pre-jump frames, you are invulnerable to grabs.

Jump towards/away:

:uf::lp:
LP: 5(s), 5(a)
:uf::mp:
MP: 5(s), 3(a)
:uf::hp:
HP: 6(s), 4(a)

:uf::lk:
LK: 4(s), 7(a)
:uf::mk:
MK: 5(s), 4(a)
:uf::hk:
HK: 6(s), 5(a)

Note: All jumps attacks, during the startup frames, Cammy’s legs are invulnerable to projectiles
Note: All jumps have 4 pre-jump frames before airborne and during pre-jump frames, you are invulnerable to grabs.

Specials

Spiral Arrow:

:qcf::lk:
LK: 7(s), 19(a), 19® -17(b) Distance Number(1.350, 2.550) don’t know what these numbers mean yet…
:qcf::mk:
MK: 7(s), 19(a), 19® -17(b) Distance Number(1.750, 2.950)
:qcf::hk:
HK: 7(s), 19(a), 19® -10(b) Distance Number(2.150, 3.350)

:qcf::2k:
EX: 7(s), 19(a), 19® -10(b) Distance Number(2.150, 3.350)

For EX:
1st startup frame can be attacked, grabbed or hit by projectiles.
From 2nd startup frame to the 25th frame, Cammy is completely invulnerable to projectiles but can be grabbed and attacked.
From frame 26 (end of her startup+active frames) until the end of the move, she’s can be attacked, grabbed or be hit by projectiles.

Cannon Spike:

:dp::lk:
LK: 5(s), 21(a), 32® -30(b)
startup frame 1~4 Cammy is completely invulnerable to everything.
Only Cammy’s legs can be attacked @ 5th frame (last startup or first active frame) and from the 6th frame until the end of the move she can be grabbed and attacked.

:dp::mk:
MK: 5(s), 26(a), 30® -30(b)
startup frame 1~5 Cammy is completely invulnerable to everything.
she can be attacked and grabbed starting from 6th frame of the startup frames until the end of the move.

:dp::hk:
HK: 5(s), 31(a), 29® -30(b)
Same properties as MK.

:dp::2k:
EX: 5(s), 26(a), 30® -30(b)
startup frame 1~6 Cammy is completely invulnerable to everything.
she can be attacked and grabbed from the 7th frame of the startup frames until the end the move.

Cannon Strike:

:uf::qcb::lk:
LK: 12(s), 11(a)
:uf::qcb::mk:
MK: 12(s), 11(a)
:uf::qcb::hk:
HK: 12(s), 11(a)

:uf: or :u: or :ub: :qcb::2k:
EX: 12(s), 16(a)**

** looks like EX C.strike is active until ground… so number of active frames is variable depending on starting height.

Tiger Knee Cannon Strike:

:qcb::uf::k:

(pj) pre-jump frames

LK, MK, HK: 4(pj), 12(s), 2(a), 5® +11(h) +8(b)

{TK Cannon strike information is from Nerfherder’s analysis}

Quick Spin Knuckle:

:hcb::lp:
LP: 35(s), 11(a), 9® +4(b)
:hcb::mp:
MP: 33(s), 11(a), 10® +3(b)
:hcb::hp:
HP: 36(s), 11(a), 10® +3(b)

For all 3 non-ex version:

frame 1 of her startup frame is completely projectile invulnerable from head to toe but vulnerable to grabs.

From frame 2 until frame 25 of her startup frames she’s vulnerable to grabs, during the same period her legs are completely projectile invulnerable but can be low attacked and her head can be hit by projectiles and high attacks during the same period.
Her mid-torso is completely invulnerable to all attacks during the same period.

From frame 26 until the end of the move, she’s can be attacked, grabbed and hit by projectiles.

:hcb::2p:
EX: 33(s), 11(a), 14® +3(b)

From frame 1 of her startup until frame 25 of her startup frames, she’s invulnerable to everything except against grabs. From frame 26 until the end of the move, she’s can be attacked, grabbed and hit by projectiles.

Hooligan Slide:

(pj) Hooligan Pre-Jump Frames

:hcf::uf::lp:
LP: 2(pj), 63(s), 9(a), 12® -1(b) Distance Number(2.101, 3.001)
:hcf::uf::mp:
MP: 3(pj), 55(s), 9(a), 13® -1(b) Distance Number(3.262, 4.162)
:hcf::uf::hp:
HP: ???
:hcf::uf::2p:
EX: ???

Hooligan pre-jump frames are grab invulnerable.

Problems with her HP and EX versions are due to the fact that when the slide happened they hit the test dummy, thus making it impossible to tell how long the active and recovery frames were… Wished the tester would do hooligan slide with test dummy right next to them, probably not a Cammy user…

Grabs

:lp:+:lk:
ground grab: 3(s), 2(a), 20® Distance Number (0.130, 0.798)
Ouch… 20 frames (almost 1/2 second) of recovery for a missed grab…

:u: or :ub: or :uf: :lp:+:lk:
air grab: 3(s), 2(a) Distance Number (unknown)

Hooligan grab:

(pj) pre-jump frames

:hcf::uf::lp: near opponent :lp:+:lk:
LP: 2(pj), 32(s), 1(a), 25® Distance Number(0.030, 1.030)
:hcf::uf::mp: near opponent :lp:+:lk:
MP: 2(pj), 37(s), 1(a), 12® Distance Number(0.060, 1.060)
:hcf::uf::hp: near opponent :lp:+:lk:
HP: 1(pj), 32(s), 1(a), 12® Distance Number(0.090, 1.090)

:hcf::uf::2p: near opponent :lp:+:lk:
EX: 1(pj), 32(s), 1(a), 16® Distance Number(0.144, 1.144)

Hooligan pre-jump frames are invulnerable to grabs.

Ultra II

:qcb::qcb::3p:
Cammy CQC: 1(s), 35(a), 61®

Holy cow!!! 61 frames of recovery that’s 1 full second… Yikes… 1 word, don’t miss with Ultra II…

Dashes

:l::l:
Back Dash: 6(s), 16® – Total 22 frames

:r::r:
Forward Dash: 18 frames Total

All 6 back dash’s startup frames are completely invulnerable to everything.
Forward dashes can be attacked, grabbed, hit by projectiles.

Focus Attacks:

:mp:+:mk:

Level 1: 21(s), 2(a), 35® – Super Armor starts from frame 1 of startup frame and lasts for 10 frames total
Level 2: 50(s), 2(a), 35® – Super Armor starts from frame 1 of startup frame and lasts for 40 frames total **
Level 3: 65(s), 2(a), 36® – Super Armor starts from frame 1 of startup frame and lasts the entire startup frames + the active frames, in this case 67 frames total **

** Level 2 and Level 3 are difficult to track because both have a minimum charge time and a maximum charge time, so depending on the tester, the startup frames (and the associated super armor frames) can vary a bit. These values are directly from the videos.

Another note on focus attacks: For level 1 and level 2, there’s a brief window of vulnerability (when super armor protection frames runs out while you are immobile but before your active frames come out, for level 1, this window is from the 11th frame to the 20th frame of the startup frames for a total of 10 frames of vulnerability) where your opponent can hit you for free without any protection from super armor.

2

WHAT DO VARIOUS COLORS REPRESENT:

Green Outline Boxes: - Vulnerable Boxes (she is vulnerable to any solid red attack box)
Blue Outline Boxes: - Grabbable Boxes (she can be grabbed whenever the blue outline box is there)
Solid Yellow Boxes: - Collision Boxes (This is mainly used by the developers for detecting object collisions)
Solid Red Boxes: - Hit boxes (Whenever this Solid red box crosses into any green outline boxes, it registers as a hit to the game engine)

Cammy Cannon Strikes:
starting from left - LK, MK, HK and EX
Cannon Strikes

Cammy Specials:
starting from top left - Spiral Arrow Startup, Spiral Arrow In-Progress, Quick Spinning Knuckle Last Active Frame (AKA Spinning back fist)
starting from bottom left - Cannon Spike LK, MK, HK, EX (the extra full invincible frame)
Cammy Specials

Jump Towards/Away Attacks:
starting from top left: LP, MP, HP
starting from bottom left: LK, MK, HK
Cammy Jump Towards/Away Attacks

Neutral Jump Attacks:
starting from top left: LP, MP, HP
starting from bottom left: LK, MK, HK
Cammy Neutral Jump Attacks

Crouch Kicks:
Starting from left: LK, MK, HK
Cammy Crouched Kicks

Crouch Punches:
starting from left: LP, MP, HP
Cammy Crouched Punches

Close Standing Kicks:
starting from left: LK, MK, HK
Cammy Standing Close Kicks

Close Standing Punches:
starting from left: LP, MP First Active, MP Final Active Frame, HP First Active, HP Final Active Frame
Cammy Standing Close Punches

Far Standing Kicks:
starting from left: LK, MK, HK
Cammy Standing Far Kicks

Far Standing Punches:
starting from left: LP, MK, HP
Cammy Standing Far Punches

3

reserved 2 (just in case)

4

This is awesome stuff, Faux! Thanks for collecting this data.

Unfortunately, however, I don’t think watching for Cammy to return to idle stance is gonna be a good indicator on recovery. I do believe that there are such things as “transition” frames that go from the last recovery frame to idle stance to smooth out animation that actually count as idle animation, which means she recovers a few frames before the idle stance actually begins. Almost all of the numbers for recovery given in the posts above are longer than the recovery frames listed in the original SFIV Wiki page, and I doubt they upped the recovery of ALL her moves. So there’s definitely something off with the recovery frames, but they are different in length, it seems, for every move. I have no idea how we would properly calculate it without doing experiments on our own or using a programmable controller.

  • James
5

Yeah… From the videos, it is impossible to tell when she actually recovered from her move, so neutral stance hit box was the only thing I can consistently count/rely on. I wish the tester would have back dashed after every move (he did on some of them, but not all), then it would have been extremely easy to tell where last recovery frame was.

I found quite a bit of new info by watching these videos such as the gap in focus attacks, jump attacks with her legs having projectile immunity during startup frames etc…

I tried to write down all the findings in this post to share with everyone here. The ultimate goal is to get all the correct frame data here and understand the game engine completely so we can predictably reproduce in game scenarios/combos using this information.

6

I just watched the vid, and I can safely say that it should be used as intended. For hitbox data only.

All of the startup and active frames are exactly the same as in PRIMA’s SSFIV frame data guide.

Analyzing recovery from that vid is impossible because of transition frames James mentioned.

Faux123, you did a mistake with n .j.Lp startup (4 frames), n j. Lk startup (5 frames), Mp SBF (I know it looks like it has 34 frame startup, but there’s one frame where Cammy doesn’t move, it shoudn’t be counted because it is a video encoding problem) and EX SBF startup (33 frames). CStrike test is inaccurate becasue it runs at 30 fps. Everything else (I’m talking startup and active frames) is correct.

One disappointing thing, the person who made this never did her airthrow, but judging by the fact that all of the data from PRIMA’s guide is correct, I think it is faster. On the other hand, backdash seems to have same 22 frames and not 20.

7

Yeah, there’s a lot of inaccurate stuff posted.

http://shoryuken.com/wiki/index.php/Cammy_(SSFIV)
Gilley took the data from the japanese guide and I’ve verified all of it, fixed the bits that were wrong (cr.LP +2 on hit? lol no), and added some stuff that I researched myself.

Also, I made a bizarre discovery regarding Cammy’s super. It hits on 6f, but is sometimes regarded as hitting on 5f even though it can be visibly seen to hit on 6f. Scroll to the bottom of the wiki page for a few extra details.

8

http://www.megaupload.com/?d=6WDU8RR4

It was brought to my attention that the recordings in the hitbox torrent for the cannon strike were only 30fps. I decided to record them myself at 60fps for all the cammy players.

This was done with a programmable stick, so hopefully the move is done with the same timing each time. The inputs should be uf on the first frame, and kick on the 22nd frame (with the qcb motion starting on the 20th frame.)

Input display is on, CH is off. I used her forward throw to set up the same distance for each video. For the first series (videos 1-4), I did throw then one jump back, so the first strike would whiff and the second hit. For the second series (5-8) I just did forward throw then the command.

9

Updated 1st and 2nd post.

First post updated with latest frame information from various sources.

Second post updated with pictures of the active frames for most of Cammy’s attacks.

Hope these will be helpful for some… Carpark you may use my thread to do your hitbox/frame analysis. I can leave the 3rd post for you.

10

Thanks.:tup:
I’ll try to fit the whole thing into one post, when I finally finish the whole thing.

It takes forever to compare hitboxes for various moves.

11

I love Cammy’s close RH. I know it’s supposed to be an anti-air, but this is what happens when you fuck up a command with a kick:
HK: 9(s), 5(a), 24® -7(h) -11(b)

12

yeah I hate when I do this when i mess up a TKCS.

13

Quote: {TK Cannon strike information is from Nerfherder’s analysis}

Thanks Faux for giving me credit. I couldn’t have figured this out without Carpark though. He deservers some of the credit too.

Your list is very thorough! Nice work. I decided to compare all available lists of frame data from regular and Super SF4 and I found out that they all have mistakes in them. Or at least combinations of hit/block stun and frames that don’t add up. This then makes it pretty much impossible to figure out if the list I now have is accurate. Your list however appears to be the one with the least amount of flaws.

One example: in the wiki frame data far HK has one frame les recovery. The total amount of frames hasn’t changed. That can’t be right. Does that mean the old data was wrong or the new? I don’t know. I’m assuming the new data in this list is correct.

Also: I think the odd numbers with the spiral arrow represent the amount of squares in the training stage that the SA travels. I remember multiple occasions where Seth Killian mentioned Capcom using these squares as a reference.

The main reason why I wanted to compile my own list is to see which moves have changed in Super. I have found nothing new on the subject. Nothing that leads to new combos or links anyway.

14

Moral of the story: Get punished for using CQC at the wrong time. SMDH…

15

Checked the recovery for all normal moves, it matches what is in the SSF4 mook. Standing HK is 38frames, this was the case in SF4 as well from my testing, so I think we can assume the previous data was wrong.

Standing	Crouching	Close

LP 13 13 13
MP 25 19 23
HP 28 22 23
LK 13 15 14
MK 22 22 24
HK 32 33 38

16

FA
First number refers to earliest frame where FA can be done, 2nd number to latest. L1 refers to the fact that only a Level 1 FA is possible if it is to combo (this actually applies to all her FA stuff.)
Where FAs on only certain frames will combo that is annotated.

Standing		Crouching		Close							
LP	3	14	3	13	3	13						
MP	3	21	3	19	3	16	Close 3-16 combos		Hold F3-27 combos (rel28)			L1
HP	0	0	0	0	3	18	Close 3-17 combos		Hold F3-26 combos (rel27)			L1
LK	3	15	3	14	3	14						
MK	0	0	3	20	3	17						
HK	0	0	0	0	0	0						
												
Cannon Spike												
LK	3	20										
MK	3	20										
HK	3	20										
EX	3	20										
												
Spiral Arrow												
LK	0	0										
MK	0	0										
HK	3	28	3-25 combos	Hold 3-34 combos (rel 35)			L1					
EX	3	28	3-25 combos	Hold 3-34 combos (rel 35)			L1					
												
TurnKnuckle												
LP	27	50	Combos 27-50		Hold F27-60 combos (rel61)			L1				
MP	26	48	Combos 26-48		Hold F26-58 combos (rel59)			L1				
HP	29	51	Combos 29-51		Hold F29-61 combos (rel62)			L1				
EX	26	48	Combos 26-48		Hold F26-58 combos (rel59)			L1
17

this is… confusing to say the least.

i get that your talking about using focus attack in combos, but the presentation is not clear at all. also you don’t really give any reasons for why anyone would want to use FA in combos.

edit: holy necrobump batman… i thought that this was just posted… oh well, only half a year late.

18

It just seemed like a reasonable topic to cover, to see which moves are focus cancellable, and whether doing so would lead into any useful followups. It was always unlikely to be match-applicable, but might hold value for combo videos. However, in finishing there didn’t seem to be much generated, and given that it’s been 6 months with no reply, that’s probably the case.

Given which, I reread it now, and the formatting makes sense to me still.

19

While this thread is bumped, I was wondering if anyone knows what the frame situation is after a blocked FADC cannon spike? The move is not like a typical SRK because it has a different animation and recovery if it is blocked vs if it whiffs, so I don’t think it’s possible to just calculate the frame advantage from the normal data.

20

-5

Trying a new method of testing frame data. If you load up the M2TS files in avidemux, advance until you get to the frame post FADC where P2 Cammy just ducks prior to jumping. Allign this as frame 0. The see how many frames until P1 Cammy ducks just prior to jumping. For all variations, this is 5 frames later, hence -5 on block.

PStick was used for the CS FADC portion. Hauppauge HD PVR 60fps for the recording.