lexis

It can indeed and does so very often, the thing above was just an attempt at truly stupid damage from a very simple combo/hitconfirm.

So, I've done some research into what causes 2-hit Chokkogatta Dandy (air super) to be untechable, and threw together this (rather messed up) gif. Alpha is really messed up in it, but it should still serve for illustration. Gif is here! Note that the hitboxes are approximate and only for illustration, although they still resemble the real thing (as seen through Pasky's hitbox viewer). Basically, the tail must hit on its own first. This then causes the sprite to change, which also changes the hitboxes. If this change makes the dive portion hit them then congratulations, 2 hit Chokkogatta Dandy (air super) for Massive Damage! Notes: Chokkogatta Dandy (air super) moves you forward, so if you hit Chokkogatta Dandy (air super) when you're around the middle of the enemy their sprite will change direction, and fly that way too. Outside of the corner this ends up being almost impossible to combo from. Characters with thin top arc air sprites and huge vertical air-hit sprites are ideal for 2-hit untechable Chokkogatta Dandy (air super), examples include Order-Sol and Anji. Characters with vertically large top arc air sprites are harder to work with, since if both parts of the super hit at the same time you get the techable version instead. This could be old news for everyone involved, but the (bad) animation might help illustrate it. Finally, here's a setup for your Fuujin stuffing needs: http://www.youtube.com/watch?v=1SM-TKIF-9w&feature=youtu.be Edited: Mixed up the names for the supers ._.

Quoted from a reply to an old PM asking about this issue, received after posting to this thread. According to which it has been posted before, although I'm not sure exactly where or I'd quote that instead ^^. But yeah, knowing that is rather important for doing analysis, and although a bit confusing during combos it's still very possible.

I used this tool for some analysis in the past, and I got the feeling that it somehow rendered the hitbox data two frames faster than the actual game was moving at, but I wasn't able to confirm it. Your screenshots might show the same behavior (if so, maybe 3f vulnerable at the end of the dash, then two frames of his standing hitbox before he returns to standing?). The sprites update three frames after a collisions happens, instead of on the next frame, and so on. Has anyone else experienced this, or is it an isolated problem? Or am I simply mistaken? Example sprites rendering slower than hitboxes at collisions: http://imgur.com/a/jsVEP#0 You can clearly see where the collision happens in the hitboxes, but the sprites don't update to match until two frames later.

I decided that I really wanted to know what would break the 2147B loop, so I did some research in the only game I could get 60 fps footage from, BB:CS1. Now, I really don't know exactly how Hazama's 5C, 2C och 2147B have changed since then except for the normals getting lower hitboxes, but the basic theory behind this should still be valid. Turns out that there's four ways to drop it in CS1, namely: 1. Doing 2C too early or too late. There's 3 frames in total that will work, one more and the 5C can't be canceled anymore, one less and they're too low. I've got pictures showing exact heights for these three frames + if you do it one frame too early. 2. Jump canceling 2C too late. Again, there's 3 frames here you can JC at and still keep the combo going, the first possible frame, and the two after it. 3. 214B not coming out on the first frame possible in the air. 4. 5C not hitting on the first possible frame. Nr 4 means that something must have been changed since then unless I've misunderstood something about advanced input for step type characters. In CS1 it appears impossible to do a dashing 5C, since you cannot afford even one frame of dashing animation come out. Please feel free to inform me of exactly what has changed, or of how I've misunderstood advanced input dashing normals for Hazama. Nevertheless, this most likely highlights the difficuly level of the loop: one 3f real link, must JC early and do an early TK, and one 5f advanced input link at the end. Not that hard if you know what actually matters in it. I had no idea that the JC had to be immediate, and hopefully someone else than me learned something from this too ^^ Pictures may be editable to a format that is actually viewable if people ask for it PS, For those saying that a 2147B input will give you a superjump, what are you guys doing? You won't since YOU'RE ALREADY HOLDING 2 FROM THE 2C. Mind you, a forward jump is still better since you get one frame of forward/backward motion in the air before the TK comes out, but accidental superjumps are not the real reason to use 963214B or 2149B