Fix him constantly facing and attacking the wrong way when the opponent is slightly off axis.
Fix his moves going through people that are standing still.
No that's how it works. There's 2 frames at the end of the window to input a and get the JF, 1 frame slower and you get RLC a again. 1 frame faster and you get RLC aa non-JF.
your 66b~rlc a:a dmg numbers are high, i think you had counter on. should be rlc k (53) rlc b (55)
nevermind he just has a just frame version of rlc a:a i didnt even know existed
rlc aa, 66 rlc b (58)
ch b6, 66~9bb tech catches all (including no tech) but back (2nd hit wiffs)
It seems pretty reliable to me, but I don't know for sure. All 3 seem to win clean in i19 situations as well.
It's possible there are multiple active frames and the moves are hitting on different frames in different situations.
Can you share some examples of cases where 3b doesn't hit?
*edit*...
Interesting factoid about 6aa that doesn't fit nicely in frame data charts:
If the first attack hits the second attack is uninterruptible. If the first attack is blocked the second attack clashes with i20.
https://8wayrun.com/threads/mitsurugi-general-discussion-trips-tricks.20010/page-2#post-771112
2a for cervy, mitsu, sophie certainly appears to be i12 in my tests.