Please post here next time: Ibuki Q&A Thread: Ask simple questions here!
“light move SJC Ultra 2” is almost always for show, and rarely if ever useful in a real match.
But to answer your :qcf::uf::qcf: (which I will refer to as M1) vs :qcf::qcf::uf: (which I will refer to as M2) question, first you need to understand how SJC U2 works. The full notation of a SJC U2, would be something like
st.MK xx super jump xx U2
That is, the st.MK is cancelled into a super jump, and that super jump is then cancelled into U2. The timing window to cancel st.MK into a super jump is pretty easy, because you can buffer the super jump (similar timing to how one could do st.MK and buffer Tsumuji).
However, the window to cancel the super jump into U2 is a small 3-4 frame window. This is because what you’re actually cancelling is the prejump frames of the super jump, which is about 4f. Once Ibuki is off the ground, you cannot cancel into U2. You also cannot buffer the U2 or do it too early, otherwise you won’t be in the pre superjump frames yet, and instead you’ll get something like st.MK xx EX dp.
Now onto the differences between the two methods of SJC U2:
M1 where you do the :uf: first, has the advantage of executing the super jump earlier. If you’re using a normal like cr.HP which has a lot of hitstop frames to use, then this may be a null advantage. But with a normal with minimal hitstop frames, such as cr.LK, you might notice it can be difficult to even do something like cr.LK xx super jump without being quite fast on the stick.
M2 where you do the :uf: last, has the advantage in timing the 3-4f pre superjump window. Since when using this method, you should be aiming to actually start your superjump around the time you perform the final :qcf::uf:. In doing so, you reduce the guess work, and the U2 input becomes a 3-4f link immediately following your :uf: input. In practice however, the super jump is still buffered a little bit earlier, but the guess work in timing the U2 input is still reduced.
So basically, M1 you super jump cancel the st.MK on the first frame of cancellability, and take a bit of a guess as to when to hit the ultra buttons. You might have trouble with this method if you don’t have a very strong sense of timing.
M2 you super jump cancel st.MK somewhere between the middle to last frame of cancellability, which reduces the guess as to when to hit the ultra buttons. You might have trouble with this method if you are too slow to perform the super jump cancel.
Finally, the brute force and simplest approach to figuring out which of the two SJC U2 methods is better for you is to just try them both and see which one you have better success or consistency with. I would actually recommend this before trying to dissect your execution, as usually your hands will automagically know which one is most comfortable to use.
tl,dr answer to question 1 :qcf::uf::qcf: is probably much easier for you when doing something like cr.LP xx SJC U2, unless you are incredibly fast (Sako-like) on the stick.
Assuming you have read the above, we can now move onto your second question:
cl.st.HK xx super jump xx command dash
I think I should first give a special mention to normals with special properties regarding SJC U2:
Any crouching normal xx SJC U2, as they have one special property that needs to be taken into account. The special property is that using the down input for the crouching normal cannot count for the super jump input as well.
e.g. :d::mp::qcf::uf: does not get you cr.MP xx super jump.
In order to get that super jump cancel, you either have to go to neutral and then perform M1, or just perform M2. Or you can perform M1 with an extra :uf: at the end (i.e. :qcf::uf::qcf::uf:), but this kind of defeats the purpose of M1 and still ends up being M2 anyways.
Second property, normals that are jump cancellable as well as super jump cancellable (e.g. cr.HP, cl.st.HK, and TC10 enders). Note that Ibuki can also do a regular jump cancel, and then cancel that jump into a special or ultra, similar to her super jump.
e.g. cl.st.HK xx jump xx U2
(fun fact: Ibuki can cancel her jump into a FA, but not cancel her super jump into FA. This is opposite to CViper)
In my research, I came to the conclusion that the jump cancel window is about 2f compared to SJC’s window of 3-4f. Another fact is that at the beginning of the cancel window (possibly only the first cancellable frame), jump cancelling takes priority over super jump cancelling.
An easy way to see the cancel timing difference between regular and super jumps, is to do cl.st.HK xx super jump. Too early and you’ll get a regular jump instead.
Combining the the two special properties I mentioned, and we can see that the window to cancel a jump must be smaller than that of a super jump. Consider cr.HP for example. It is a crouching normal, so as long as you do something like
you will always get a regular jump cancel, no matter what. But if you do
:d::hp: :n: :qcf::uf:
you have a much higher chance of executing a super jump cancel, since going neutral allows you to perform a super jump cancel, and it also adds a delay for you so that you don’t cancel the cr.HP too early and get a regular jump instead.
Now onto the test, try doing
and then try doing
:d::hp: :n: :qcf::uf::lk:
Both are technically correct inputs for cr.HP xx jump/superjump xx LK cd, but depending on your jump, the LK cd will or will not be easy to time correctly.
tl,dr answer to question 2: make sure you’re doing
cl.st.HK xx super jump xx command dash
cl.st.HK xx regular jump xx command dash
PhD research: Ibuki Combos and Glitches