@Sitri - sucrose C4 mechanics
As an author, It is your responsibility to complete the ticket
- Take advice constructively
- NO disrespectful behavior
- The ticket will be scrapped if: No activity > 1 week or open for > 1 mo.
- When done, type $close
Theory/Finding/Bug: Title of your submission
Evidence: Explanations with calculations and/or Youtube/Imgur proofs
Significance: Conclusion
SitriApr 30, 2021 4:50 AM
Mechanics behind Sucrose C4.Recap: Sucrose C4 creates a 7 hit counter, where after 7 hits your E CD is reduced by 1-7s.
Findings:
Sucrose C4 creates a 7 hit counter when the first normal/charged attack connects an enemy (hitting the air does not create, nor further the counter). However, this counter starts whether your E is on cooldown or not. If you do 5 normal attacks, then E and put your E on cooldown, you only need to do 3 more to hit the counter and get your cooldown reduced.
Switching characters does not delete this counter either, and it exists for a very long time, tested up to 30s, so guesstimate is 30s plus. Death also does not delete the counter.
The cooldown reduction hard caps at each charge's cooldown. If you get a higher reduction than the current E charge CD, it will reset, and not carry over to potentially 2nd charge's cooldown.
Example: E is on 3s CD, you get a 5s CD reduction, you instantly gain E charge and timer start at 15s for the remaining charge.
Sucrose C4 CD reduction probability distribution:
https://docs.google.com/spreadsheets/d/1jMODclkPguPHpTsLmGafbvQKIcXk1ddc9NWYPjhqqsc/edit?usp=sharinghttps://cdn.discordapp.com/attachments/837549621936652309/837957383111049216/unknown.png
Although the distribution does not fit nicely into the Gaussian curve, mean ~ 4s suggests there is no overly skewed cooldown reduction to either lower or higher bound, at large N.
Evidence:
Switching character does not reset Sucrose counter:
https://www.youtube.com/watch?v=Q32IuH-Eu9g
Counter starts whether your E is on CD or not:
https://youtu.be/Q2zyolyA8r0
Counter exists for 30s+:
https://youtu.be/qTboBW3A7QwSignificance:
This very slightly makes on field enabler Sucrose comps better, as swapping on Sucrose is very lenient + better understanding of Sucrose C4 RNG.(edited)
Switching character does not reset Sucrose counter
Sitri
Counter starts whether E is on CD or not
Sitri
Sucrose counter exists for 30s plus
@INVEST₼CLOSEYOURFUCKINGTICKETS
probability distribution of 1-7s reduction I will work out this weekend
INVEST₼CLOSEYOURFUCKINGTICKETSApr 30, 2021 4:51 AM
noice
@gimmeabreak we got some behavior findings
gimmeabreakApr 30, 2021 4:57 AM
nice! what happens if you go up to 7 normals when both stacks are available? and then use E, does the next auto reduce the cd or do you have to redo all 7 again?
SitriApr 30, 2021 5:01 AM
your next auto reduce the CD
as long as your counter hasnt reset cause you go over 7
you should be fine
gimmeabreakApr 30, 2021 5:07 AM
ah ok, so if you do over 7 then you start over regardless if you got any benefit out of it? ok that makes sense, let me go fix up my code thanks!!
SitriApr 30, 2021 5:12 AM
yeah, its like you are trying to reduce 0s cooldown
mol | anemo fanApr 30, 2021 5:48 AM
https://streamable.com/anmrud i was testing sac frags to see if E proced even without damage, but the C4 did. figured you all would be interested to know
also please let me know if there's any ID / name info in there, trying to crop that stuff out(edited)
SitriApr 30, 2021 6:29 AM
So it procs on hit
not on damage
nice
thanks
mol | anemo fanApr 30, 2021 6:31 AM
yeah i thought that was weird. i think i tested on some flowers when i was walking around and it didn't work, which i also found weird
so i think it has to hit an enemy but it doesn't have to do damage (edited)
oooh... can baron bunny / gouba be targeted?
might be worth testing
i'll check
Ayzel | netlimited blade worksApr 30, 2021 6:36 AM
are you planning on putting the probability distribution on this ticket or making a separate one(edited)
SitriApr 30, 2021 6:37 AM
i will throw it here
no need to make another ticket
Ayzel | netlimited blade worksApr 30, 2021 6:38 AM
ok sounds good
honestly tho
i think it kinda deserves its own separate one
you can dump the info here and we can just publish it as two tho
SitriApr 30, 2021 6:38 AM
thats fine yeah
mol | anemo fanApr 30, 2021 7:02 AM
just confirming that i checked for some even weirder targets (targetable: pyro flower; not targetable: campfires, baron bunny/gouba, electro crystals) and nothing else worked. so it is very specifically on enemy hit, including abyss mage shields
SitriApr 30, 2021 7:08 AM
@INVEST₼CLOSEYOURFUCKINGTICKETS new finding when im testinfg distribution, the cooldown reduction does not carry over to the 2nd charge
if you have 0 charge, and like 3s CD left before filling 1st charge, even if you get a 5s CD reduction, it will only insta fill that charge, instead of carrying over
INVEST₼CLOSEYOURFUCKINGTICKETSApr 30, 2021 7:09 AM
limits the potential
SitriApr 30, 2021 7:11 AM
yeah
make counting distribution more annoying too
since now I have to select samples
cant use the ones that are <7s before CD that insta reset CD
hmm
but does that skew the data?
it shouldnt no?
ah screw it, i'll push it to n = 200
time to auto 1600 times on sucrose
mol | anemo fanApr 30, 2021 7:20 AM
does the AA count carry over across charges? I haven't noticed in gameplay, but i have noticed that it always bottoms out at the first charge and starts the next at 15 if you hit a cooldown proc. i think the attack count carries over to charge 2, so it might be "ideal" to wait if you are at 6 and 1s left(edited)
SitriApr 30, 2021 7:20 AM
yes, AA count carry over
1
its just reduction hard cap at 0s CD of each charge(edited)
mol | anemo fanApr 30, 2021 7:30 AM
what distribution are you testing? the amount of cooldown it deducts? might be easiest to just test on one stack when it is near max cooldown remaining so you get the full value deducted(edited)
then you can use those values to indicate how likely you are to be wasting cooldown if you happen to hit 7 attacks without much cooldown left(edited)
hypothetical: "with 3s left of cooldown 25% of C4 procs will be partially wasted"
gl testing it, that is a tough one!
now that i'm thinking about it the fastest way is probably to: use 2 Es, then record AAing as quick as possible. just remove every CD reduction that hits 0s, because any one that doesn't is evidence for the overall distribution of 1-7s cooldown reduced.
SitriApr 30, 2021 7:37 AM
yeah
im doing that rn
but inital counting and this doesnt look nice at all
mol | anemo fanApr 30, 2021 7:37 AM
you have the patience of a saint
SitriApr 30, 2021 7:37 AM
values seem random af
its not in intervals of 1s
mol | anemo fanApr 30, 2021 7:39 AM
yeah when it says between 1-7 seconds i never noticed a 1s interval in casual play
however given those 11 values the distribution looks very normal around 3.5ish
at a glance
once you get to 25 check if the average and median time are fairly close together, imo
the data doesnt fit that well in the normal distribution though
INVEST₼CLOSEYOURFUCKINGTICKETSMay 1, 2021 7:10 PM
Legwork like this
Is what i miss about being just a regular tcer
Faranight | too deep in swirlMay 1, 2021 7:11 PM
I wouldn't expect it to fit into that normal distribution
INVEST₼CLOSEYOURFUCKINGTICKETSMay 1, 2021 7:11 PM
Its a pita in the moment but i think youll be proud of it later
gimmeabreakMay 1, 2021 11:56 PM
what kind of messed up distribution is that..
my god mhy...
oh but the mean is close enough to 4.. so does that mean it's good enough to call it uniformly distributed?
SitriMay 2, 2021 12:18 AM
best is it we can dig through the files and find the value, but unless we go to N = 1000 i doubt the data would look better, but even then its not surefire
gimmeabreakMay 2, 2021 12:20 AM
ya for sure thanks for the hard work! at least we know the mean is in the middle! not something stupid like.. mean of 2
SitriMay 2, 2021 12:57 AM
@INVEST₼CLOSEYOURFUCKINGTICKETS i think this ticket is done right?
I dont recall anything else interesting I need to note down
INVEST₼CLOSEYOURFUCKINGTICKETSMay 2, 2021 12:58 AM
yeh
just condense the graph and stuff
into the conclusion of the original post
and its done and dusted
SitriMay 2, 2021 1:04 AM
@INVEST₼CLOSEYOURFUCKINGTICKETS is this good?
INVEST₼CLOSEYOURFUCKINGTICKETSMay 2, 2021 1:05 AM
ye
thats about right
Faranight | too deep in swirlMay 2, 2021 4:51 AM
It's just uniform. I really would not expect Mihoyo to make it anything but uniform or slight alterations, they probably want to keep the RNG in their mechanics fairly simplistic. Good to have data on it though.
thanks!!
So it procs on hit
1
3
thanks for the hard work! at least we know the mean is in the middle! not something stupid like.. mean of 2
Save transcript
Reopen Ticket
Delete Ticket