/corrections/ I will edit this comment multiple times of stuff that i missed. Extra Turn Mechanics (Resurgence) - somewhere in character logic II, maybe even III. - works differently than 100% advance, has a different buff duration interaction and gameplay wise can be hard to minmax.
Yoooo let's go an iceberg I actually know the full depths of lol, been waiting for a video like this! (besides the auto battle source code stuff, I just know some behaviors from community trial and error) As for the RNG crit analysis, I think a general rule of thumb is that 100% Crit Rate always comes first. We have so many Crit DMG and so little Crit Rate buffers atp that even going for 100/140 often results in a balanced 1:2+ Crit ratio. Not only that, but missing even a single Crit for even a multihit move is such a huge damage dropoff that often times it's worth cutting your Crit DMG by as much as, say, 30-40%, to push your Crit Rate up from 90 -> 95% or 95 -> 100%. Heck, the chance of landing 20 Crits in a row with 95% Crit Rate, which can be as little as 1 Acheron Ult or 2 Yunli Ults, is only 35.8%. This is especially important for Seele, who sometimes can't afford to miss ANY Crits, since it means losing a whole extra turn of damage from Resurgence. Biasing towards high Crit Rate also ties a bit into dynamic enemy HP breakpoints, where lowering your Crit DMG but still Critting lets you kill most enemies, while increasing your Crit DMG but missing a few Crits will make you miss the kill. And plus, if your 300k Crits are already killing 300k HP enemies, isn't it better to make that 300k more consistent, instead of trying to raise it to 350k and waste 50k damage on overkill? Chances are, if you need to raise your Crit DMG in order to fish for harder Crits, you should already know that after seeing an enemy left with 1% HP when you land your full combo. At the end of the day though, I won't deny that the objectively best Crit ratio for someone's account will always vary based on what they have available.
its a lot more complicated than that, some characters have more hits which make lower crit rate not as big and has a more reliable mean average. some characters have more even hit splits which helps while other hit splits are small x 7 and one big hit. rng crit analysis then uses random samplings to achieve a fully accurate number, the general rule of thumb is still true, but there isnt a defined number, its just eyeballing.
@ Makes sense, I can see how hit split fudges up cross character calculations. For example, Seele's Skill damage distribution per hit is 20-10-10-60, so missing her last Crit (60% of her damage) is way worse than missing her 2nd or 3rd Crit (only 10%). And short of having 100% Crit Rate, you have no way to control which Crits hit and which ones miss. It's just RNG within RNG lol
It feels like theorycrafting for HSR has fallen into obscurity. In the beginning it was a popular topic because everyone had a limited amount of units and unfinished or improvised teams. As more units released and the team archetypes got fleshed out the theorycrafters split into two camps, one focusing on max score/0 cycle and the other focusing on highly impractical units/teams. The issue is that the first camp can be made obsolete by simply spending money, and the second is pure novelty that would result in a bricked account if an average player tried to replicate it. Very few make content that would help the average player, such as who to use to get full rewards when you dont have all of the newest/meta units. There are a lot of tricks that Ive learned on my own I rarely see anyone use, like which crit based units can be used on a superbreak team without changing their gear. Other than Himeko, Ive had success with Argenti against the Aventurine boss in AS, and Dr Ratio in MOC. Another one is that running s1 Topaz or Himeko with Acheron can actually work in certain situations because their follow ups can apply debuffs and result in faster ults for Acheron, offsetting the damage loss from not having two nihility.
im a bigger fan of weird theorycrafting, making strange teams work through weird synergies. but admittedly, guide theorycrafting or pure meta optimization number crunching is more popular and beneficial. thats why one has significantly less material.
u should have put something on extra turns,, like how seele's ult has to be used before you kill the enemy for ur extra turn to maximize energy gain,, or how boothill, qingque, and blade's extra turn on skill can be used for buff extension. also welt's cut in during enemy turns and rappa's extra turn too
this would also apply to niche synergies like using Acheron's ult in between enemy actions when using Black Swan with her, in order to proc the DMG taken bonus from Black Swan's ultimate
(Surface level) Autobattle reverse engineering is the result when I fine tune builds over time to achieve no hit and fast farming runs, all using the observed auto logic.
You was on my watchlist, now you earned my sub. Great video, for me as genshin tc and new hsr player who understands everything about star rail, but still "floats" and whants to dive deeper... Yeah. Thank you appreciate the work and dedication
imo, at level 1, you generally lead to duo dps imo lol as I type it you mention it but yeah, people like damage, too much supports, even if one big number, might feel bad. Having healer/buff/dps/dps means every other turn see satisfying number, and also if one dps dies or isnt hitting an enemy that its good against, you get some options.
Totally expected the relic substat chance conspiracy theories to be on here lol Great vid ^^ you always have some of the most unique content in the community, keep up the good work!
I actually got started on something similar to the build stamina analysis section. originally this was prompted when I saw someone complain about spending 30 self-modeling resin fishing for double-crit SPD boots and not getting anything. I did some napkin math out of boredom to confirm that this was an incredible waste of resources, then got to thinking about relic probability in general, which led to me building a very simple model of the relic system in sheets (with some assumptions to make things easier) while I think there were some interesting conclusions - for one, I now think the new relic release cadence of 3 months is a *very* deliberate choice - I also think that further investigation wouldn't be as useful as approximate rules-of-thumb for the average player trying to clear content. it also left me thinking about other systems and design direction regarding difficulty and powercreep - I'm thinking at some point in the future, maxing traces might become the default for endgame and so this might just become a non-issue.
I think an important point in dynamic HP breakpoints that you didnt bring up is also the point on enemies with Safeguard, and their toughness breakpoints. This would only have been relevant on the Blue dino in Penacony, but all of the Apoc bosses are centered around Safeguard, so this is 100% worth considering as well. Implications of this being being able to use an ult before it gets refilled by the Safeguard break buff, or to use turns/fix team order before the whole team sees the 100% AA, just to name a few.
I would also add "Hits per Action" to "Bonus hits" & "Bounce attack". Although not that important right now, this might become meta once the dev put a lot of focus with this playstyle.
Knowing how Bounce attacks work and how different buffs work like Def shred and Def ignore being the same, how def shreds stacks with itself and thus exponential, how vulnerbability is linear and how certain character stack too much of a stat making another main stat piece viable(example Xueyi and Acheron stack a lot of dmg bonus so Atk orb is always better for Xueyi and sometimes for Acheron ,depending on the team). Like many people still think Aveturine's follow up deals more dmg when he is against more enemies.
It's actually not that unlikely for star rail to be someone's first rpg, this mostly applies to kids, like for example my 13 year old sister who also started playing it
I think my stubbornness on certain MoC patches forced me to be in battle logic 3 lol. I also didn’t realize crit rng had a term. I’ve loosely applied this for certain characters because of the fact they hit multiple times.
No wonder most people struggle against Hoolay He has high HP High Toughness and is so fast that he takes 4 to 6 turns per action he's a nightmare for anyone who doesn't understand speed tuning and his Enchanced states gimmick,I hate how he's so tailor made for FUA
I would say that understanding of Skill Points is a bit more basic and necessary than team synergies, and it's definitely a requirement to clear story during certain checkpoints like Swarm King and Aventurine. You can run a perfect Feixiao Follow-Up team and still lose if you're using your SP on Topaz and Robin more than Feixiao and Aventurine. And that's one of the most SP-forgiving teams right now. (Though at that point, auto-play might be better than manual Lol) Guess I'm somewhere between level 2 and 3 in the iceberg. I understand most in team and battle logic III, but at the same time I can't be bothered with speed tuning and I rarely touch the damage formula.
Rng crit analysis is not really that bad, its just maths about the mean auto battle reverse engineering can be done by anyone paying attention and questioning why the character does this and whether the enemy mechanic is affecting it and you should know the auto battle outcome, without even looking into the code. Sometimes im lazy and i just let auto battle do the rest of the fight if i know it will already do it optimally or close to it
Is team specific counters to specific fights really that low in the iceberg... Maybe its because Im used to turn-based rpgs like SMT but id consider it fairly basic gameplay in comparison to understanding damage formulas (which i dont really calculate as much)
BLADEEE RERUN PLEASEEE I DO ANYTHING FOR BLADIE I MUST GET HIM E6S5 PLEASE IDC WHAT ANYONE SAYS I LOVE HIM AND I WANT HIM SOI'LL PAY FOR HIMMMMMMMM TAKE ME MONEY 4MONTHS OLD PLAYER BTW
15:27 speaking of break, the break value changed since 1.something, it was in multiplier of 30 (30 for basic, 60 for skill etc) and now I see 5,10 What is the change? Is 30 break now 5 break? 10? Help pls
After a bit over 4 months of playing I can confidently say im around a level 1.5 or 2. I did manage to clear all endgame modes by sheer luck but if i were to be thrown into a max level account I would basically only know how to used my two main teams (dr ratio fua and ff break). I honestly don’t bother with high level mechanics like speed tuning/action advance since my relic luck is terrible and I probably won’t get specific breakthrough for a whole team to flow smoothly for another month of constant grinding or so. In the end I feel relatively new still and just watch gameplay videos of people using similar comps but not exactly studying them to see what makes them good. Instead im just like “so this works” or “so these kinds of stats improve dmg”, etc. Thanks for the video, I hope to continue to improve but seeing where I an right now I feel a little sheepish to call myself a “gamer”
on build stamina as well, another factor is how much you can control rng, i remember seeing some things about controlling first relic substat gain level up to 4 (rare for 2, epic for 3, legendary for 4) to control with better odds the other things, very goofy things as well
substat manipulation is a fallacy, it relies on superstition and psychological traps similar often used to sell snake oil. it's not proven correct and the "conclusive evidence" changes every month, theres always an excuse.
Substat manipulation is fake due to the simple fact that if it were real, there would be zero people complaining about shitty relics or daily stamina limit. (I'm a day one player, so both of those things aren't really an issue for me lol)
/corrections/
I will edit this comment multiple times of stuff that i missed.
Extra Turn Mechanics (Resurgence) - somewhere in character logic II, maybe even III.
- works differently than 100% advance, has a different buff duration interaction and gameplay wise can be hard to minmax.
I didn't understand what the Seele mains were talking about all the way back in 1.0, but I understand now as an E2 Firefly haver.
@@lucadivine3862wait, what were they saying?
@@MaybeNooonseele best waifu
Relic RNG "Manipulation/Prediction"
you should add relic substat manipulation--the spreadsheets are insane, at least to my untrained eye. it's unknowable domain worthy
I like when casuals play the way I played pokemon as a kid all damage dealers
HSR official discord is full of those, at least they posted a lot in the game's first quarter. 😂
Seele + Serval + Himeko + Natasha
Mr Yang, he synthesized
Today I learned my account is so bricked I accidentally became a theorycrafter. Thanks, Waveman.
A pioneer of our community 🫡 o7
Convergent evolution lmfao
Yoooo let's go an iceberg I actually know the full depths of lol, been waiting for a video like this! (besides the auto battle source code stuff, I just know some behaviors from community trial and error)
As for the RNG crit analysis, I think a general rule of thumb is that 100% Crit Rate always comes first. We have so many Crit DMG and so little Crit Rate buffers atp that even going for 100/140 often results in a balanced 1:2+ Crit ratio. Not only that, but missing even a single Crit for even a multihit move is such a huge damage dropoff that often times it's worth cutting your Crit DMG by as much as, say, 30-40%, to push your Crit Rate up from 90 -> 95% or 95 -> 100%. Heck, the chance of landing 20 Crits in a row with 95% Crit Rate, which can be as little as 1 Acheron Ult or 2 Yunli Ults, is only 35.8%. This is especially important for Seele, who sometimes can't afford to miss ANY Crits, since it means losing a whole extra turn of damage from Resurgence.
Biasing towards high Crit Rate also ties a bit into dynamic enemy HP breakpoints, where lowering your Crit DMG but still Critting lets you kill most enemies, while increasing your Crit DMG but missing a few Crits will make you miss the kill. And plus, if your 300k Crits are already killing 300k HP enemies, isn't it better to make that 300k more consistent, instead of trying to raise it to 350k and waste 50k damage on overkill? Chances are, if you need to raise your Crit DMG in order to fish for harder Crits, you should already know that after seeing an enemy left with 1% HP when you land your full combo. At the end of the day though, I won't deny that the objectively best Crit ratio for someone's account will always vary based on what they have available.
its a lot more complicated than that, some characters have more hits which make lower crit rate not as big and has a more reliable mean average. some characters have more even hit splits which helps while other hit splits are small x 7 and one big hit.
rng crit analysis then uses random samplings to achieve a fully accurate number, the general rule of thumb is still true, but there isnt a defined number, its just eyeballing.
@ Makes sense, I can see how hit split fudges up cross character calculations. For example, Seele's Skill damage distribution per hit is 20-10-10-60, so missing her last Crit (60% of her damage) is way worse than missing her 2nd or 3rd Crit (only 10%). And short of having 100% Crit Rate, you have no way to control which Crits hit and which ones miss. It's just RNG within RNG lol
It feels like theorycrafting for HSR has fallen into obscurity. In the beginning it was a popular topic because everyone had a limited amount of units and unfinished or improvised teams. As more units released and the team archetypes got fleshed out the theorycrafters split into two camps, one focusing on max score/0 cycle and the other focusing on highly impractical units/teams. The issue is that the first camp can be made obsolete by simply spending money, and the second is pure novelty that would result in a bricked account if an average player tried to replicate it. Very few make content that would help the average player, such as who to use to get full rewards when you dont have all of the newest/meta units. There are a lot of tricks that Ive learned on my own I rarely see anyone use, like which crit based units can be used on a superbreak team without changing their gear. Other than Himeko, Ive had success with Argenti against the Aventurine boss in AS, and Dr Ratio in MOC. Another one is that running s1 Topaz or Himeko with Acheron can actually work in certain situations because their follow ups can apply debuffs and result in faster ults for Acheron, offsetting the damage loss from not having two nihility.
im a bigger fan of weird theorycrafting, making strange teams work through weird synergies. but admittedly, guide theorycrafting or pure meta optimization number crunching is more popular and beneficial. thats why one has significantly less material.
Using HuoHuo against Kafka gets you a bunch if free action forward
THIS IS INOVATIVE!
Every new player can watch this video and learn about the game mechanic in a fun way
u should have put something on extra turns,, like how seele's ult has to be used before you kill the enemy for ur extra turn to maximize energy gain,, or how boothill, qingque, and blade's extra turn on skill can be used for buff extension. also welt's cut in during enemy turns and rappa's extra turn too
True
Extra turn also won't consume buff
this would be somewhere in level 2, i knew i forgot some more seele stuff 💀
this would also apply to niche synergies like using Acheron's ult in between enemy actions when using Black Swan with her, in order to proc the DMG taken bonus from Black Swan's ultimate
(Surface level) Autobattle reverse engineering is the result when I fine tune builds over time to achieve no hit and fast farming runs, all using the observed auto logic.
You was on my watchlist, now you earned my sub. Great video, for me as genshin tc and new hsr player who understands everything about star rail, but still "floats" and whants to dive deeper... Yeah. Thank you appreciate the work and dedication
Build stamina analysis was the thing I never knew I needed.
imo, at level 1, you generally lead to duo dps imo
lol as I type it you mention it
but yeah, people like damage, too much supports, even if one big number, might feel bad. Having healer/buff/dps/dps means every other turn see satisfying number, and also if one dps dies or isnt hitting an enemy that its good against, you get some options.
hypercarry often wins because people often times dont know basic math and addition, one big number > two slightly smaller numbers
Totally expected the relic substat chance conspiracy theories to be on here lol
Great vid ^^ you always have some of the most unique content in the community, keep up the good work!
I actually got started on something similar to the build stamina analysis section. originally this was prompted when I saw someone complain about spending 30 self-modeling resin fishing for double-crit SPD boots and not getting anything. I did some napkin math out of boredom to confirm that this was an incredible waste of resources, then got to thinking about relic probability in general, which led to me building a very simple model of the relic system in sheets (with some assumptions to make things easier)
while I think there were some interesting conclusions - for one, I now think the new relic release cadence of 3 months is a *very* deliberate choice - I also think that further investigation wouldn't be as useful as approximate rules-of-thumb for the average player trying to clear content. it also left me thinking about other systems and design direction regarding difficulty and powercreep - I'm thinking at some point in the future, maxing traces might become the default for endgame and so this might just become a non-issue.
I think an important point in dynamic HP breakpoints that you didnt bring up is also the point on enemies with Safeguard, and their toughness breakpoints. This would only have been relevant on the Blue dino in Penacony, but all of the Apoc bosses are centered around Safeguard, so this is 100% worth considering as well.
Implications of this being being able to use an ult before it gets refilled by the Safeguard break buff, or to use turns/fix team order before the whole team sees the 100% AA, just to name a few.
Great vid, was wating for something like that for a long time, and you did it quite well. my appreciaton for hard work sir
Cooked once again, Waveman! Great video!
I would also add "Hits per Action" to "Bonus hits" & "Bounce attack". Although not that important right now, this might become meta once the dev put a lot of focus with this playstyle.
Oh. I, uh, thought this was gonna be a very *different* kind of Theorycrafting...lemme come back this with some snacks.
Knowing how Bounce attacks work and how different buffs work like Def shred and Def ignore being the same, how def shreds stacks with itself and thus exponential, how vulnerbability is linear and how certain character stack too much of a stat making another main stat piece viable(example Xueyi and Acheron stack a lot of dmg bonus so Atk orb is always better for Xueyi and sometimes for Acheron ,depending on the team).
Like many people still think Aveturine's follow up deals more dmg when he is against more enemies.
He didn't cook. He ASCENDED and CREATED.
It's actually not that unlikely for star rail to be someone's first rpg, this mostly applies to kids, like for example my 13 year old sister who also started playing it
ilysm
Love,
A Gentlethem Fan ♥
I think my stubbornness on certain MoC patches forced me to be in battle logic 3 lol. I also didn’t realize crit rng had a term. I’ve loosely applied this for certain characters because of the fact they hit multiple times.
No wonder most people struggle against Hoolay He has high HP High Toughness and is so fast that he takes 4 to 6 turns per action he's a nightmare for anyone who doesn't understand speed tuning and his Enchanced states gimmick,I hate how he's so tailor made for FUA
I would say that understanding of Skill Points is a bit more basic and necessary than team synergies, and it's definitely a requirement to clear story during certain checkpoints like Swarm King and Aventurine.
You can run a perfect Feixiao Follow-Up team and still lose if you're using your SP on Topaz and Robin more than Feixiao and Aventurine. And that's one of the most SP-forgiving teams right now. (Though at that point, auto-play might be better than manual Lol)
Guess I'm somewhere between level 2 and 3 in the iceberg. I understand most in team and battle logic III, but at the same time I can't be bothered with speed tuning and I rarely touch the damage formula.
I’m surprised that Motion Value isn’t mentioned
Rng crit analysis is not really that bad, its just maths about the mean
auto battle reverse engineering can be done by anyone paying attention and questioning why the character does this and whether the enemy mechanic is affecting it and you should know the auto battle outcome, without even looking into the code. Sometimes im lazy and i just let auto battle do the rest of the fight if i know it will already do it optimally or close to it
Is team specific counters to specific fights really that low in the iceberg... Maybe its because Im used to turn-based rpgs like SMT but id consider it fairly basic gameplay in comparison to understanding damage formulas (which i dont really calculate as much)
BLADEEE RERUN PLEASEEE I DO ANYTHING FOR BLADIE I MUST GET HIM E6S5 PLEASE IDC WHAT ANYONE SAYS I LOVE HIM AND I WANT HIM SOI'LL PAY FOR HIMMMMMMMM TAKE ME MONEY 4MONTHS OLD PLAYER BTW
15:27 speaking of break, the break value changed since 1.something, it was in multiplier of 30 (30 for basic, 60 for skill etc) and now I see 5,10
What is the change? Is 30 break now 5 break? 10? Help pls
multiply by 3. 30 toughness back then is now 10 on game descriptions, 60 is 20. the formula is still the same, using the original 60/30 values,
@ ok thanks
Can't wait for 3.X to add to these🤣🤣🤣🤣
Idk man I think you overestimated casuals. Most of them have no idea what’s going on at all and just see pretty character yayy
Please help i cant beat memory of chaos 6 my team is kafka and dang heng IL and Argenti and bailu
After a bit over 4 months of playing I can confidently say im around a level 1.5 or 2. I did manage to clear all endgame modes by sheer luck but if i were to be thrown into a max level account I would basically only know how to used my two main teams (dr ratio fua and ff break). I honestly don’t bother with high level mechanics like speed tuning/action advance since my relic luck is terrible and I probably won’t get specific breakthrough for a whole team to flow smoothly for another month of constant grinding or so. In the end I feel relatively new still and just watch gameplay videos of people using similar comps but not exactly studying them to see what makes them good. Instead im just like “so this works” or “so these kinds of stats improve dmg”, etc. Thanks for the video, I hope to continue to improve but seeing where I an right now I feel a little sheepish to call myself a “gamer”
on build stamina as well, another factor is how much you can control rng, i remember seeing some things about controlling first relic substat gain level up to 4 (rare for 2, epic for 3, legendary for 4) to control with better odds the other things, very goofy things as well
substat manipulation is a fallacy, it relies on superstition and psychological traps similar often used to sell snake oil. it's not proven correct and the "conclusive evidence" changes every month, theres always an excuse.
Substat manipulation is fake due to the simple fact that if it were real, there would be zero people complaining about shitty relics or daily stamina limit.
(I'm a day one player, so both of those things aren't really an issue for me lol)
interesting ok 👍 never looked into it that much just came across it while browsing, the more you know!
You forgot Relic RNG Manip/Prediction, it's interesting enough imo
relic rng manipulation is not real and is founded by sketchy data
Ö
WaveMan77
shoutout to my gentlethems fr
fr ong