Summary of cauldrath’s findings/confirmations and stuff I’ve tested:
Dizzytime is how long you have to wait until your stun resets to 0. Dizzytime goes down during anything during the game! Hitstun, blockstun, and knockdown included.
Almost all attacks add dizzytime. The dizzytime just gets accumulated to the current dizzytime you have.
One dizzytime is one simulation frame. However, ST is actually simulated at 75 fps according to sirlin and the drawing is done at 60 fps. Thus, sometimes, the dizzytime counter will go down by more than 1 per frame. The dizzytime immediately starts ticking down as soon as you get it.
Almost all attacks add a random amount of stun between two integers. This stun is simply summed up as you get hit and does not get reduced due to combos. The only way to get rid of stun is for your dizzytime to hit 0 to reset your stun. Stun will not decrease over time like in 3s/sf4.
I haven’t tested, but due to the assembly code snippet cauldrath pointed out, I suspect that it’s easier to get dizzied on faster speeds because the dizzytime probably doesn’t drop down at a faster rate due to the higher speed.
Grabs do not seem to do any dizzy. If you need an excuse to use non-grab throws on a grab char, use it to dizzy.
Chars dizzy at 30 or above. However, characters have some type of modifier that affects how much stun they take per hit. This is either a simple multiplier or there’s a random chance of just subracting/adding 1 to the hit.
Throws do not instantly stun or add to the dizzy timer. Teching does not seem to reduce the stun or dizzy timer.
Dizzytime does not strictly determine how much stun an attack does.
However generally, heavier moves do more dizzy.
Sets of moves with the same dizzy/strength commonly have the same dizzytime. However, you can’t use this when trying to find the optimal stun combos/setups. Moves unintuitively have different stuns that were programmed due to the game designer. For example, n.ryu’s best diagonal jumping stun move is dj.mk.
The stun from a move does not seem to be a uniform distribution. It is harder to get the lower or higher end of the stun a move can do than the middle.
Here are some tests I ran to test how characters receive extra/less stun. As you can see, sim indeed does definitely stun more easily than ryu who stuns more easily than gief.
Note that these tests may be slightly inaccurate since I could only test by repeated trials. I don’t know assembly well enough to figure out the algorithm and numbers from the game.
gief vs sim
nj.8+hp 22-29 100
dj.2+hp 12-18 60
cr.lp 0-7 40
cl.lp 0-7 40
fa.lp 0-7 40
fa.lk 0-7 40
cl.mp 6-13 60 *can be 5?
cl.mk 5-13 60
cl.hp 12-18 80
cr.hk 6-13 130
gief vs ryu
nj.8+hp 22-27 100
dj.2+hp 12-17 60
cr.lp 0-5 40
cl.mp 6-11 60 *can be 5?
cl.hp 11-16 80
gief vs gief
nj.8+hp 21-27 100
dj.2+hp 11-17 60
cr.lp 0-5 40
cl.mp 5-11 60
cl.hp 10-16 80
lariat 9-14 100
ryu vs ryu
*when i got a range of 6 inclusive, i quickly stopped for testing most ranges
moves have the following dizzy properties generally unless otherwise noted:
ground light 0-5 stun 40 time
ground med 5-10 stun 60 time
ground hard 11-15 stun 80 time
dj light 2-7 stun
dj medium/hard 4-9 stun
nj light 2-7 stun 40 time
nj med 6-11 stun 50 time
nj hard 11-16 stun 60 time
move stun time notes
cl.lp 40
cl.mp 60
cl.hp 3-8 40 wierd
cl.hp(2nd part) 11-15 80 2nd hit
cl.lk 40
cl.mk 60
cl.hk 10-14 60+70 2-hit (128)
cl.hk(2nd hit) 2-5
cr.lp 40
cr.mp 60
cr.hp 3-8 40 wierd
cr.lk 40
cr.mk 60
cr.hk 5-10 130
fa.lp 40
fa.mp 60
fa.hp 80
fa.lk 40
fa.mk 60
fa.hk 80
dj.lp 40
dj.mp 40 2nd hit has very little stun
dj.hp 40
dj.lk 40
dj.mk 7-11 50 best dj stun
dj.hk 40
nj.lp 40
nj.mp 50
nj.hp 60
nj.lk 40
nj.mk 50
nj.hk 60
6+hp 2-5 70/70 1st hit
6+hp (2nd) 1-6 70/70 2nd hit
6+mp 2-6 70/70 1st hit
6+mp (2nd) 2-6 70/70 2nd hit
p throw 7-13 100
k throw 8-13 100
srk 10-15 100
tatsu 4-9 90
air tatsu 4-9 90
blue fb 8-13 110
red fb 5-11 100
super 5-10 n/a *a bunch of small hits and this is general result