Tari

If you know they're just going to mash 236A, move forward until you're in range for j.236A, and then just hit them in the face with it and get a combo. Or just slowly approach them and get closer until they have to do something else. If she teleports to get away, you can anti-air her or jump up and air-to-air her pretty freely. Hilda isn't scary if all she does is shoot fireballs all day. She's scary and annoying because she has other options to work with, and has a very strong mid-range presence. Aside from her 2A, her close and long range options are not that great. Really, whatever you're doing with Gordeau to get past Hilda 236A spam should work with Yuzu, too.

Remember that if you end a combo with 236C and you still have enough untech, you can cancel it into 214B for oki. Obviously spacing dependent, but it's a good thing to remember. In some cases, you can actually do things like ... 236B > 236A > 236C > 214BD > j.214C for really odd looking combos. Edit: forgot to mention that it's normally best to end strings with 214c for damage and knockdown. Using 236c into 214b is usually not as good, but if the situation allows, can let you get 66c oki, which is nice. There are other situations where 236c is superior, but they tend to not revolve around getting good oki. The 236A > 421C route is arguably more stable (especially useful in netplay) than any of the other routes, but does less damage than the alternative routes. The highest damage confirm is 236B~D > 236A > (delay) 214B~D > etc, but 236B~D > 236B > (delay) 214BD still does more damage than the 421C route. While you can combo that string into 66C, you can't do anything afterwards because you've already spent your two wallbounces/groundslams up with the opening two 236B hits. If people want, I (and/or probably other people in this thread) can try to put together a video of combos that tend to be troublesome for newer players, but I think most of the combos are available in video format in the Yuzu tutorial vids? Not really sure, never watched more than the first one, haha. ========= So I ran into a really strange thing today while playing against our local Carmine player. I jumped and did an aerial C teleport (j.421C) right as he threw a ground pinwheel. We were maybe half a screen apart and he CS'd his pinwheel immediately. The superflash happened and we could clearly see me in the middle of the fade away animation for teleporting, but when the freeze ended, I teleported backwards and to the floor behind me, instead of down to the floor behind him. We tried for a bit to replicate it, but couldn't figure out how it happened.

4c and 2c will both reach in time.

I just want to mention that the 'head-to-head' setup on the stream station was brilliant and should be used at every single tournament ever. I mean, it takes a bunch of space, but it's hella nice. Jona - Gordeau Yraelis - Byakuya

If you're doing a microdash mortal slide, you're pretty much outta luck, though I think it's easier to use the 6+AB dash command for earliest possible mortal slide. Using the 41236 input trick might help you time your input (and help you stop dashing), but that's about all it'll do. If you're just trying to hold a dash and mortal slide out of it, hold your dash with 3 instead of 6 and you'll be able to do 236 when you want to.

Should've said 'rising normals,' sorry. I used the term 'rising overheads,' because that's what the system appears to be designed to punish. It's not a proration on rising moves, it's a damage scaling system that only kicks in when the player being hit happens to be holding down-back when hit by a rising normal. It also affects Eltnum's j.[C], due to the way j.[C] forces her to rise, even when the Eltnum player uses the move as an actual overhead. It's likely a leftover system that was designed before they stripped rising overheads out of the game completely. Proration on a max-range hit sounds like a move-specific property, not a system mechanic.

Go to neutral for 4 frames before doing 236. The input interpreter's buffer is sorta weird, but you can generalize it as working in 4 frame intervals.

Damage and untech proration are almost guaranteed to be unrelated. There's this crazy anti-fuzzy system in the game that nerfs damage dealt from rising overheads normals if the defender is hit while holding down-back (say, while they're in recovery from whiffing a move), but doesn't nerf untech time. The damage nerf from that mechanic is huge. It looks like it immediately applies an extra P2 to the first move of 0.40, and that the proration value scales downwards either for each subsequent hit or over time. Or there's some other crazy non-proration-related explanation for it, but either way, it doesn't affect untech time. Still looking into this, but I can confirm that CS adding hitstun is not an Eltnum only thing. Finally found a combo on Yuzu that works with CS only. On Eltnum's combo, it looks like you get 2 extra frames of untech, which is higher than the previous number I saw. Still hoping to find out if there's a standard for how many frames you get.

The original mook is pretty barebones, honestly. It only gives us the data I listed up above. Seeing as how that video is from one of us, we can definitely explain it, haha. You can always cancel teleport recovery into another stance move, including another teleport. This means that you can exit stance with a d-cancelled teleport, but continue teleporting around as long as you input the next teleport during the previous teleport's recovery frames. The teleports you get from that point on are all slow teleports, though.

I have no way of confirming if these numbers are true in UNIEL or not, but this is from the UNIB mook (asked someone else to help me read it). I assume that these startup and total numbers don't include the superflash frame, as well, but it could have just been different in UNIB: 236C Startup: 13F Active: 5F Total: 38F j.236C Startup: 18F Active: 5F Landing Recovery: 11F 214C Startup: 14F 1st Hit Active: 3F Landing Recovery: 33F j.214C Startup: 12F 1st Hit Active: 3F Landing Recovery: 33F Infinite Worth (41236D) Startup: 12F 1st Hit Active: 3F Total: 147F 5A Startup: 5F Total: 18F Static Diff: -2F 2A Startup: 6F Total: 20F Static Diff: -3F Since I'm not sure if these are relevant numbers in this game, I'm going to leave them separate from the rest of the frame data.

They just play with fewer players.

j.2[C] into 66C is always going to be easier than j.C into 66C. The latter requires a late enough hit from j.C and an immediate cancel into 66C, whereas the former lets you take your time after you ground the opponent. I like using j.236A > j.6C > etc. because it's high damage and has a good amount of carry distance, but you can't always use that route. It's sometimes necessary or more reliable to use j.236A > j.214AD > 2C or j.236A > j.214B~D > 2C. If you had a good starter (pretty much anything that's not 5A/2A), then you can also use j.236ABA > j.214AD > 2C for slightly more damage when using the j.214AD path. The triple slash combo route has the highest damage out of all the routes when you don't end in double super (double super enders do more damage if you use the j.6C route) and when you don't want to end in super + IW, as the IW will generally whiff. If you just want to spend 200 meter at the end of your combo, using this triple slash route or the j.6C route will deal about the same damage. Without vorpal, the j.6C route does 1 or 2 extra damage if you end in double super when compared to using the triple slash route and ending in IW. Remember you can't use the triple slash route without a good starter, though. As far as making j.C > 66C work, I suggest just trying to catch the opponent lower in your air string, so you have more untech time when you land. You can combo 66C quite high in the air, but it's way harder, and it might not work depending on your combo's proration. When doing the relaunch combo, you can delay the j.B by a decent amount to keep the opponent lower, or just use 2C > 5C, which doesn't launch as much. Pretty much every hit in the relaunch combo can be delayed by a small amount. On the note of relaunch combos, when you have space, using j.2C first does a small amount more damage than using j.B first, but it's hardly important. Yeah, there are quite a few weird exceptions to Yuzu's stances. The one you mentioned first was brought up in this thread some time ago, and no one had a good explanation for it. It would be much more useful if it could be used from a cancel, but alas, it's a neutral exception only. The cancel you get is also a d-paired cancel, so it's slower than a normal one. Still good to know about, though. Tenkai and I were having fun making combos formed around trying to use all her stance exceptions in the same combo: https://www.youtube.com/watch?v=urG3gT5RPSs https://www.youtube.com/watch?v=V7ESf4ybKNM https://www.youtube.com/watch?v=1VfM-TSLjsw And a teleport video: https://www.youtube.com/watch?v=n6YoQ0E3BBE Didn't know about the throw whiff, though, that seems interesting, looking into frames on it. ========= edit: The throw tech is pretty great. I give up on theory crafting this stuff, it just seems like stance specials can carry over some cancel routes from whatever comes before them. You can do weird things like 5A > 236B~[C] > 236A > 421B > Throw. Frames: (point blank) 2A > 214A~D: -8 (point blank) 2A > 214A > Throw Whiff: -15 2A > 236A~D: -7 2A > 236A > Throw Whiff: +3 <-- theoretically useful

Kazuki's combos are pretty much the standards. Off of assault starters, if you can confirm the hit, you can use 5A > 5C > 9D > j.C > 66C > etc. and just do the j.2[C] > 66C combo route from there. I would say it's probably my preferred combo route, since it works well from pretty much any starter. If you're too far to confirm a falling j.C hit into anything other than 66B > slashes, you can use either of the following routes: - 66B > 236A > 214B~D > ender (this will cross up, so your ender should take that into account) - 66B > 236A > 421C > j.214B~D > ender (I generally suggest using this one) For anti-air 5C hits, you generally do one of the following: - (high hit) 5C > j.B > j.6C > j.214B~D > (land) > j.B > j.2C > j.C > 66B > etc. - (mid height) 5C > j.B > j.2C > j.C > 66C > j.236A > j.6C > j.214B~D > 66B > etc. - (CH) 5C > 5A (whiff) > 66B > 236B > 236A > 9D j.2[C] > 66C > j.236A > j.6C > j.214B~D > 66B > etc.

As a by-product of testing teleports, I've found out that in-stance teleports on the ground go invuln on frame 4, which is actually the frame before any wind effects appear. I'm unsure if this includes throw invuln, though, and don't plan to test it. This would imply that all my numbers for teleport invuln are slow by 1 frame, so I've adjusted the frames accordingly. If anyone happens to know of any supers that are 1+4, or any 'blockstrings' that have exactly 15 frame gaps, please let me know so I can test this more concretely.

Against Byakuya, I just want to point out that his web traps actually drain your GRD away if you break them or block them. It's a fairly interesting mechanic. On block, they take half a GRD block. On destruction, they take a full GRD block, and then on hit, they take 1.5 GRD (not sure if this is true in combo). I checked this earlier today, but I was actually told that they took none on block and only half if destroyed by our local Byakuya player, so I'm not sure if I'm missing something. Byakuya does gain some grd if his webs are blocked or hit his opponent(0.5 grd), and the webs do more chip damage if shielded by their respective shield types (ie: aerial shield vs air web, grounded shield vs ground web).

The 66C > j.236A > j.6C > j.214B~D > 66B ender path is universal. If it's dropping, it's probably a timing issue. I can record the combo on every character if it'd help as reference, but I'm not sure how to describe it. I usually just delay the j.236A and not the j.6C. Also not sure why Wald is falling too fast for you in the other combo. I don't use it, but it seems to work fine in training mode. Try delaying your 4B, perhaps? You can teleport away at the end of any blockstring if all you want is space, or you can end with 5B > 5A (whiff) or 2B, both of which are +/-0 on block, and then just back away normally or start over.

The move list isn't exactly well labeled (the moves that are labeled are mostly wrong, it seems) or sorted, and Yuzu is just... not there. edit: yuzu is up!

Haven't had too much chance to play this matchup a lot, but the few times I did play it, it felt like Yuzu just has an overwhelming advantage in most situations. Chaos' pet can be destroyed with any special move, or any B or C normal, which is incredibly handy. He can't summon the pet very safely after it's destroyed in this matchup, since Yuzu can literally just slash at him and destroy the pet on appearance by happenstance. The fact that he doesn't have a meterless reversal (does he even have a metered one?) really doesn't help him. That said, Chaos has decent corner pressure and oki, and his hitbox can be annoying for combos.

? I was told that debug reported Eltnum's jab as 3f, when it's 5f. If that's not the case, then I don't know what's going on with the numbers in debug mode.

Tried it with Eltnum 236B(3) > 236C > 5B > 5[C], with a chainshift between 5B and 5C. It didn't seem to work there, but I don't know if that gap is too long or if I just suck at timing chain shifts. Human error is too influential in my test. :P Might be a special property on those specials, then. I'll edit that. I am considering looking into it more, since 236B(1 hit) > EX Gunshot > 236B(2 hits) has increased untech time when the initial 236B is a CH, but 236B(1) > EX Gunshot > 236A has no extra untech on CH starter. Might be reduced SMP penalty, as well, not really sure. Could also be because Orie 623A is multi-hit. I would appreciate it if other people could chime in on their ideas behind counterhits and/or correct anything I wrote, because the system isn't very well explained.

Untech time scaling? Because I don't get any extra untech on hits done after CS between having 0/3/6 GRD. If there's extra scaling, I guess it applies to any CS? edit: Yuzu 2C > 5C without CS has 35 frames of untech, whereas 2C > CS > 5C has 36 frames of untech with any amount of GRD. Maybe you get +1 frame? lol. Doing a wallbounce combo into 5A with or without CS, though, has the same amount of untech time.

Iirc, debug mode counts frames from 0, so you have to add 2 to debug mode numbers. This issue showed up when people posted Eltnum frame data from debug mode and it was 2 frames off from the hand-counted numbers as well. This is, of course, assuming the debug mode is reliable.

I generally try to push CD as Yuzu starts sheathing her sword. That's the point in time where you can cancel. If you're any earlier or later it won't work. On an unrelated note, I am still somewhat baffled by Carmine and Waldstein corner pressure.

If you get a really high hit on 5C and you confirm into j.B, you can use this route: 5C > j.B > j.6C > j.214B > (land) j.B > j.2C > j.C > 66B > etc. If you get a low anti-air 5C hit, it's best to go straight into the relaunch loop with j.B > j.2C > etc. and then combo into the 66C route.

I meant that the 4C > 236ABA > 9D > j.2[C] part was more difficult to perform because of the delay timing after 4C, but I forgot we were talking about punishes and not random hits, so it actually shouldn't really matter. In any case, 4C > 236ABA > etc. does 5061 damage in vorpal, so it is more optimal. Is there anything else we can add? :P If you have 3+ or 6+ grd, it might be worthwhile to do ...66B > 236ABA > CS > 66B > 236ABA > IW as the ender instead. Haven't tested it, though. (edit: tested, does less damage)