Server Icon
KQM | Genshin Impact
itto-buffering-mechanics
64 messages
Avatar
TiBotThu, June 23, 2022 at 03:13 AM
@wlukaz - As an author, it is your responsibility to complete the ticket

Guidelines
- You can rename your ticket with /rename <ticket name> or with the button below
- If this ticket was created by accident or it can be deleted, you can use /delete within the first 5 minutes, otherwise ask a Scholar.
- When you are ready to submit the ticket, compile everything into one message following the format below and pin it. Then type /close or click the button below; the ticket will automatically be moved to be reviewed.
- To add contributors to your ticket you can use /contributor add <user>.
- The ticket will be scrapped if: no activity >1 week or open for >1 month.
Write-up Format
Theory/Finding/Bug: Title of your submission

Evidence: Explanations with calculations and/or YouTube/Imgur proofs

Significance: Conclusion
Avatar
steno119Thu, June 23, 2022 at 01:20 PM
:Hmmge:
Avatar
beeMon, June 27, 2022 at 09:30 AM
just to be sure, is it a normal normal attack or a burst normal attack?
Avatar
wlukazMon, June 27, 2022 at 12:49 PM
it shouldn’t matter but in his ult
Avatar
TiBotMon, July 4, 2022 at 01:00 PM
:dinkdonk: This channel hasn't been active in the past week!
Deadge2
Avatar
wlukazSat, July 9, 2022 at 01:54 PM
someone will do this eventually
right
:artesians:
Avatar
aftermathrarSat, July 9, 2022 at 04:00 PM
just gonna log my findings since i'm recording stuff (frames counted in avidemux so no frame counter overlay on attached vids)

holding LMB from idle to go straight into a CA should give us the minimum buffer time. it takes ~14 frames (0.233s) from startup until the "stack consumed" flash appears
when using n2+, the normal cannot be canceled into a CA until after the frame the attack would be activated

here are two videos, one holding LMB as early as possible in the swing, around .5s before the CA flash, and one delaying until near the .233s buffer window. you'll see that they still end up going into the CA on the same frame of the swing animation
in this clip, i hold LMB after the n2 would have hit and it also has a buffer time of 14 frames, starting the CA during the n2 recovery frames
same idea for n3. the early clip shows holding LMB for over .5s before going into the CA
second clip holds for .316s and both go into the CA on the same frame
checked in burst and the buffer timing from idle -> CA was still the 14 frames, so i think we can say this holds true in itto's burst as well
and for n4 we know that you can basically start holding LMB whenever and it'll buffer the CA asap
tl;dr:
CA input takes 14 frames to activate
other than N1, itto cannot start a CA before the hitbox activates on the normal
if you start buffering at least 14 frames before hitlag ends, itto will go straight into a CA afterward
@wlukaz that's basically all i can think of for now

i will want to double check whether hitlag counts toward the buffer input frames, but it seems like it should
Avatar
wlukazSat, July 9, 2022 at 06:10 PM
:WICKED:
Avatar
aftermathrarSat, July 9, 2022 at 08:57 PM
ok, some interesting results when looking at buffering inputs on hitlag. main takeaway is that the hitlag frames count toward the CA input, so it's pretty lenient

although n1 acts a little differently than n2/3

for n2/3, if you start buffering during the hitlag frames, it'll start the CA teleport ~14 frames from that point, meaning you get the teleport during the recovery frames of the normal you did

if you buffer CA out of n2 perfectly, the earliest it will start is 4 frames after hitlag ends. looking at itto's sword rotation, frame 1 being when it starts moving again

for n3, the earliest teleport starts 6 frames after hitlag ends. i judged hitlag mainly from itto's kick rotation. if you count frame 1 as when he starts rotating again, he disappears on frame 5 and the stack flash is on frame 6

n1 if you start holding lmb during the hitlag, it'll buffer the n2 attack and start the teleport on-hit of the n2. however, this teleport is delayed by the hitlag frames, but is ~4 frames faster than buffering the ca out of n2 normally (since it skips that 4 frame delay)
somewhat related, the earliest an input for the next normal is accepted is during hitlag, so a reliable way to buffer an n3 -> CA is just to double click and hold when you see the n2 connect. it's pretty easy to build the muscle memory to space out the inputs, since itto's hitlag is consistent compared to most characters and you have around half a second of leeway
@wlukaz not sure if all that made sense, lemme know if you want clarification
here's the odd behavior of buffering during N1 hitlag

the normal CA teleport is 24 frames, if you buffer like this it's 32 frames since the hitlag is 8 frames
Avatar
wlukazSun, July 10, 2022 at 01:45 AM
that all seemed to make sense yeah
so basically it works normally, but his n1 is weird bc if you buffer the n2 with a click during the animation, and then start holding during the hitlag (is it just the hitlag? or can it be slightly earlier/later?) you can save 4 frames of the ca coming out?
or like
you buffer the n2 click during the n1 animation, then the ca during n1 hitlag? or is it during the n2 hitlag
Avatar
aftermathrarSun, July 10, 2022 at 06:53 AM
it's just holding during the n1 hitlag. if you watch the keyboard overlay in that vid, you'll see that i only make two inputs. i click once to start the n1 and then hold LMB during n1 hitlag which both queues up the n2 and buffers the CA for it.

for the other normals, the single hold input during hitlag will just start your CA during the recovery frames of the normal instead of queuing up the next normal + buffering
if you click/hold before the hitlag connects, itto won't start the next normal in the chain

for the n1, if you hold for n2 after the hitlag ends, it'll do the CA with the expected timing, which is a few frames after the n2 hitlag ends

not really sure what's up with the "CA at start of hitlag" behavior, but it shouldn't matter much because there's not really a point where you start a CA out of N2 anyway
i think the takeaway for the normal buffering is something like

during a normal's startup swing, before the hit connects:
- click inputs are too early to start the next normal in the chain
- hold inputs will start the CA out of the current normal after 14 frames or after hitlag, whichever is later

during a normal's hitlag:
- click inputs will start the next normal in the chain
- hold inputs will start a CA during the recovery frames of the normal

after a normal's hitlag:
- click inputs will start the next normal in the chain
- hold inputs will buffer a CA after the next normal's hitlag ends

from this, imo the optimal way to buffer itto's combos is to click once during each swing's hitlag and then start holding right after hitlag ends. this will queue up the next normal as early as possible and give you a very wide window to start holding LMB in order to buffer the CA
@wlukaz ping here since i forgot to reply to a message
Avatar
wlukazSun, July 10, 2022 at 11:33 AM
:WICKED:
nice to know
can i use that for the finding/significance of this ticket
or do you just want to type it up @aftermathrar
Avatar
aftermathrarSun, July 10, 2022 at 11:39 AM
i think you'll be more succinct than i was, so if you want to copy or reword this message and maybe the tl;dr one, i think it has the info for both findings/significance
i wrote a lot to say "hold left click to charge attack" @wlukaz
Avatar
wlukazSun, July 10, 2022 at 02:11 PM
lol
idk i’m on mobile so i don’t wanna type a big thing rn
w/e i’ll do it
oh i need those videos uploaded to yt or imgur so i can use them as evidence
Avatar
wlukazSun, July 10, 2022 at 02:13 PM
specifically this one and
Avatar
wlukazSun, July 10, 2022 at 02:14 PM
just a few of these videos that show 14f, maybe smush them together into one video
if you could just upload those to yt i’d really appreciate it
Avatar
aftermathrarSun, July 10, 2022 at 02:19 PM
sure
Avatar
wlukazSun, July 10, 2022 at 02:21 PM
that’s probably good
Avatar
aftermathrarSun, July 10, 2022 at 02:26 PM
n1 buffering: https://youtu.be/MXn0lpZBvH0
itto n1 buffer 3x
Recording showing that itto's CA starts ~14 frames after holding LMB
n2/n3 https://youtu.be/PpVaIyjd3Ak
itto n2/n3 buffer comparison
The start of the CA will not interrupt n2, n3, or n4. Instead, it will only activate after the hitlag has ended.
This video shows holding LMB at different timings before the hitbox is active and the CAs starting on the same frame.
n1 exception (comment with description) https://youtu.be/wdi78tpEnDI
itto n1 buffer exception
Itto cannot start a CA during the recovery frames of his n1. Instead, it is delayed until the n2 connects, at which point he will start the CA teleport. However, this teleport is extended by the number of hitlag frames.

It does end up being ~4 frames faster than a normal CA out of n2, but there are no combos which would realistically utilize th...
Avatar
aftermathrarSun, July 10, 2022 at 02:42 PM
i'd add this in significance since that's what people will mainly look at in the tcl

"the most consistent way to buffer itto's combos is to click once during each swing's hitlag and then start holding lmb as soon as hitlag ends. this will queue up the next normal as early as possible and give you a very wide window to buffer the CA"
Avatar
wlukazSun, July 10, 2022 at 04:00 PM
ok
Finding:
Itto’s CA has a uniform buffering time - it takes 14 frames from holding LMB to the stack flash, including hitlag frames.
Itto’s N2 allows a CA to come out a minimum of 4 frames after hitlag ends. N3 has a minimum of 6 frames after hitlag.

The exception to this is Itto’s N1, which allows you to hold LMB during its hitlag and will then start an N2 and immediately buffer a CA, which skips the minimum 4 frame delay after hitlag, making this the most efficient way to do a CA out of N2.

Evidence:
Standing CA: https://youtu.be/MXn0lpZBvH0
CA out of N2/N3: https://youtu.be/PpVaIyjd3Ak
N1 exception demonstration: https://youtu.be/wdi78tpEnDI

Significance:
This gives us a greater and more quantitative understanding of Itto’s CA buffers, as well as potentially opening up an easier way to do certain combos with the N1 exception.

The most consistent way to buffer Itto’s combos is to click once during or a bit after hitlag for each NA, then hold click as soon as the final NA animation begins. As long as you start holding click at least 14f before the earliest possible CA (which is 4f/6f after hitlag for n2/3) you will get a frame perfect CA out of it.
itto n1 buffer 3x
Recording showing that itto's CA starts ~14 frames after holding LMB
itto n2/n3 buffer comparison
The start of the CA will not interrupt n2, n3, or n4. Instead, it will only activate after the hitlag has ended.
This video shows holding LMB at different timings before the hitbox is active and the CAs starting on the same frame.
itto n1 buffer exception
Itto cannot start a CA during the recovery frames of his n1. Instead, it is delayed until the n2 connects, at which point he will start the CA teleport. However, this teleport is extended by the number of hitlag frames.

It does end up being ~4 frames faster than a normal CA out of n2, but there are no combos which would realistically utilize th...
does that look good and a correct analysis of what you discovered @aftermathrar
Avatar
aftermathrarSun, July 10, 2022 at 04:23 PM
mostly, i might be reading this part wrong:
"As long as you start holding click at least 14f before the end of the NA (including hitlag of the hit) you will get a frame perfect CA out of it."

something like this makes more sense to me "As long as you start holding click at least 14f before the earliest possible CA cancel (including hitlag of the hit) you will get a frame perfect CA out of it."

edit: also if you click before the hitlag of the normal, you won't continue the combo, so i'd amend the start of the last paragraph to say only "during or just after hitlag"
but "earliest possible CA cancel" is kinda clunky and might not be clear
also thanks for typing it up lol
Avatar
wlukazMon, July 11, 2022 at 03:02 PM
ok
Finding:
Itto’s CA has a uniform buffering time - it takes 14 frames from holding LMB to the stack flash, including hitlag frames.
Itto’s N2 allows a CA to come out a minimum of 4 frames after hitlag ends. N3 has a minimum of 6 frames after hitlag.

The exception to this is Itto’s N1, which allows you to hold LMB during its hitlag and will then start an N2 and immediately buffer a CA, which skips the minimum 4 frame delay after hitlag, making this the most efficient way to do a CA out of N2.

Evidence:
Standing CA: https://youtu.be/MXn0lpZBvH0
CA out of N2/N3: https://youtu.be/PpVaIyjd3Ak
N1 exception demonstration: https://youtu.be/wdi78tpEnDI

Significance:
This gives us a greater and more quantitative understanding of Itto’s CA buffers, as well as potentially opening up an easier way to do certain combos with the N1 exception.

The most consistent way to buffer Itto’s combos is to click once during or a bit after hitlag for each NA, then hold click as soon as the final NA animation begins. As long as you start holding click at least 14f before the earliest possible CA (which is 4f/6f after hitlag for n2/3) you will get a frame perfect CA out of it.
itto n1 buffer 3x
Recording showing that itto's CA starts ~14 frames after holding LMB
itto n2/n3 buffer comparison
The start of the CA will not interrupt n2, n3, or n4. Instead, it will only activate after the hitlag has ended.
This video shows holding LMB at different timings before the hitbox is active and the CAs starting on the same frame.
itto n1 buffer exception
Itto cannot start a CA during the recovery frames of his n1. Instead, it is delayed until the n2 connects, at which point he will start the CA teleport. However, this teleport is extended by the number of hitlag frames.

It does end up being ~4 frames faster than a normal CA out of n2, but there are no combos which would realistically utilize th...
Avatar
TiBotMon, July 11, 2022 at 03:02 PM
Ticket closed by @wlukaz. If there are any issues with it - it can be reopened by the owner or staff by using the buttons below or /open.
Avatar
itzsomebody | chrisTue, July 12, 2022 at 12:45 AM
Note if this makes it to PR: Make sure to credit Aftermath#7658 entirely
Avatar
TiBotSun, July 31, 2022 at 04:41 PM
@Tibot added @aftermathrar as contributor
Ticket verified by @BowTae
Ticket verified by @eriri🌸
Creating transcript...