Never thought about it deeply, but very fascinating to see the reasoning behind all those mysterious “impossible” scores on leaderboards like that HDHR SS getting beaten by lower acc play bc of spinners (wild), great videos as always willy!
3:16 Another thing here are people like xKirito, that forced scores like low combo - it can lead to absurd cases. FGSky lost 2 potential game PP records... because plays were worth more PP, but less score to his 4-mod play on Justadice at that time. Yeah...
I played the ouendan games as a kid and I think it's why I react so negatively (not violently but usually an audible wth) to combo breaks. Combo breaks in ouendan could easily halve your score or worse compared to an FC whereas combo breaks in my two current favorite rhythm games barely punish you at all, Muse Dash might knock 5-10k off in a game where 400-500k scores are pretty common on harder charts and the way scoring works in Beat Saber where base score is really granular and dependent on precision and follow-through and the punishment for a combo break is pretty weak (you drop one rank of multiplier) it's pretty likely that a more fluid run with a couple of combo breaks will outscore a more stilted FC.
Score! We all know it, and we all love it. When you click on a note you gain score depending on how accurately you hit. 0 if you miss, 50 and 100 if you time your click incorrectly, and 300 for a perfect hit. Your combo factors into this, increasing the score you get from each subsequent note. Along with mod multipliers, all of it is added to your total for every object. You can’t forget about spinners which can provide bonus score for each extra revolution! So it’s quite simple, if you hit more accurately and get a higher combo you should get a higher score… Oh… okay. The way score is calculated is actually much more complicated. Everything I’ve said so far is still true, albeit a simplification. How much of a simplification? Here is how ScoreV1 works. There are four types of judgements in osu!, the miss, 50, 100, and 300. Right? Wrong. There are actually seven types of judgements including the 100 Katu and 300 Katu, which give no additional score, and the 300 Geki when a perfect combo is completed. What does that mean? We don’t have time. Circles will give the base amount of score. Sliders are broken into base anatomy and whether you hit certain parts determines the judgement given. Heads don’t matter, ticks are worth 10 score, repeats and ends are 30 apiece, and at the end the full slider judgement is given. Spinners will give base score for the judgement as well as two types of bonus points which are not affected by other multipliers, such as mod multiplier or combo multiplier. For every rotation before the spinner is complete you gain 100 bonus points, and after complete 1,100 bonus points. Every single one of these types of scoring objects will raise your combo, and at each point in time of hitting a note. The judgement is multiplied by combo times mod multiplier times difficulty multiplier divided by 25 plus one. What is the difficulty multiplier? To get this the stats of a beatmap, that’s CS, OD, and HP, (AR not included) are added up as well as the return of a piecewise function that’s based on how many hit objects there are divided by the drain time in seconds. And then we multiply that by 5/38ths because it wasn’t arbitrary enough. That! Is how we calculate the score for one single note. Now we add to this to the score total throughout the play, doing the calculation again and again for every single note that is hit. Too complicated? Maybe my simplified explanation was good enough. The current scoring system, officially known as ScoreV1, like all other overly-specific osu! phenomena originates from the Ouendan series. At the core of this is increasing combo leads to exponential gains, because a higher combo play should be in theory a better play. For the most part this system is fine for its original purpose, but a few properties of the formula lead to less than ideal results in reality. There’s a reason why “getting ScoreV1’d” is now a common phrase, directly relating to the tendency for objectively better plays to award less score. You might have already spotted the first glaring issue: When you drop accuracy matters. Because the hit value is increased by your combo, objects later in a beatmap contribute much more score than objects that appear earlier. Two plays with exactly the same accuracy and combo can have different scores. Because one had more accuracy mistakes earlier in the map compared to later on, it gets a higher score. Much to the ire of the playerbase, it is actually very common for slightly low accuracy plays to beat higher accuracy ones on leaderboards. Whether this is due to the aforementioned combo nuance, differences in how fast players spin, or both is highly dependent on the situation. Regardless, this happens so frequently that I am willing to bet every osu! player has encountered a situation where a higher accuracy score doesn’t beat their current record, Or even worse, they accidentally replace their current play with a worse accuracy one. The second big one, that has ravaged many a leaderboard, is with mod multipliers. Because contrary to what many initially think, the multiplication isn’t just done to the hit values, but also for each mod. Famously, DoubleTime alone actually has a lower mod multiplier, 1.12x, than Hidden and HardRock combined although it says they’re equal in-game. This is because 1.06 times 1.06 equals 1.1236, which is rounded for the display. It can be quite annoying to see a DoubleTime SS which is beaten by a lower accuracy Hidden HardRock score. Nowadays on maps you can submit a score for each different mod combination, but before an update in 2016 only one play was saved no matter what mods you played with: the one with the highest score. So let’s say you’re Cookiezi on April 15, 2013 setting the first ever DoubleTime FC on Raijin no Hidariude… well that’s a shame because you already have a Hidden HardRock FC. Due to the extra multiplier that one’s worth more score, and your new FC (that no one could match for the next three years) doesn’t submit! Finally, as a cherry on top, it is practically impossible to calculate the score of a play without taking the exact inputs from a replay. These issues are nothing new, and several proposals for new scoring systems have popped up over the years. To make tournament matches more fair, ScoreV2 was created in 2015, and has been the standard for competitive play ever since. All plays are Standardized to be a maximum of 1,000,000 points before mod multipliers and spinner bonuses are included, with 70% of your score coming from combo and the other 30% from accuracy. Lazer, the next version of osu!, uses a sort of "ScoreV2.1" which shares the same base concepts with room for changes. Currently there is debate over the percentages and how to balance the mod multipliers, but it will continue to inprove as Lazer continues development. Hey, and if the 1,000,000 point max isn't your cup of tea, Lazer also includes a setting which can convert the Standardized scores into very close estimations of V1 values, but without the aforementioned drawbacks associated with the old system. Thinking a little outside the box, the purpose of a scoring system is to reward better plays with more points, so many imagine a future where performance points can replace score entirely. Even though pp famously has its own glaring problems, it may be worth considering down the line. Maybe there's another, even better system which we haven't even dreamt of yet. Until then, we'll have to live in a ScoreV1 world, and deal with the downsides that come with it, as we have for 15 years already. Score has always been a mess, but at least we’re working towards a future that is much less messy. Who knows where scoring will go next.
What I still don't understand is how is it possible that mod multipliers don't work with low object counts? Like the leaderboard on the Ultra Begginer diff on MIIRO vs. Ai no Scenario is riddled with NFEZHT scores tied with HDHRDTFL. Also, if CS, OD and HP gets accounted for the hitobject values, why do the first few notes add their base value to the score? No matter what, the first few notes will always add 300 (or 100, 50 or 0) to your score. It's weird x3
The note density and difficulty stats on Ultra Beginner are so low that the difficulty multiplier is only 0.45, which rounds down to 0, and that's the part of the formula that mods affect. Combo scaling also has a -2 as part of the formula, so the first 2 notes of a combo are only worth base value and you start getting combo + mod multipliers from the third note.
Can someone please direct me where I could find out more about what was said at 1:03 ( where he mensioned that we dont have the time to explain that) I would like to learn about it throgh a video since reading about it doesnt help me as much. I tried to find videos but none of them actually explain it. Any help would be appriciated!
Like how in catch an A or S can be better than an SS because of score gotten from spinner. Mania's scoring is sooo good. It makes the game fun. I don't always worry about misses, I just try to get a good score that I feel good about. The 7k mania pp record LITERALLY has 30 misses. :(
I used to really enjoy playing mania for this exact reason. But I do have to say that actually getting an fc for a good score feels so much nicer on std. For me atleast, mania FCs feel really redundant and not important or special at all. Also imo even if its more of a grind it makes the pp records seem more important and special because it is what you need to get a good score, whereas with mania you couldve missed like 10 times and still gotten a really good score.
@@xd3athclawx554 Hm, I guess it really depends on the score for me tbh. It sucks that in standard you HAVE to fc to get ANYTHING. Catch requires fcs but not as much it feels like.
@@Imaproshaman7 Yeah I get that, not getting anything from a play because it wasn't an fc is kinda annoying but I usually don't mind because I don't usually go for pp unless I know I can fc the map in the first place, and then if I 1 miss I'm annoyed because I 1 missed not because I didn't get any pp.
what?¿?¡'¿??¿ i have peen playing for 10 years and i didnt know that u could have several scores with different mod combinations. 7 years without knowing
Idc that it emulates the original game stepmania was a ddr emulator but it didn't force itself to copy its frame based timing window wanna know why? because its shit so is the original osus scoring system so idk why any sane individual would look at that complex mess that favors sloppier play without misses and think they should copy it
Because osu! is a simulator first, original game later. So peppy developing the original score system from the osu tatakae ouendan make sense. And it's not like Peppy don't want to make it, he does, but people don't like changes. See ScoreV2 and lazer.
clamp is technically a mathematical function by its nature, but is more or less referred to as something in programming. Its more complex math than I have studied (it looks like it anyways, and I haven't ever seen it before) but what I found was this, clamp(x) = max(a,min(x,b)) ∈ [a,b] it looks similar to the thumbnails clamp where it has clamp(8 objects/time , 0,16)) so it could be that. I think that part is just a way to simplify a part of the math on calculating the speed or bpm of the map through the amount of circles (maps using 6ths and such) because in the editor you can go from 1/1, all the way up to 1/16 or 16ths, and in the clamp it goes 0,16, stating the range of the function, ∈ [a,b], ∈ [0,16] or a part of the set 0 to 16. I also found someone say "Looks like clamp(x) is a projection onto an interval" Which would fit in with the interval ∈ [0,16]. Hope this is what you were looking for :)
scoer : good
more scoer : better
less scoer : :(
"if you dont understand it, make it simple"
Never thought about it deeply, but very fascinating to see the reasoning behind all those mysterious “impossible” scores on leaderboards like that HDHR SS getting beaten by lower acc play bc of spinners (wild), great videos as always willy!
you could get a limit of like 600 at the time
Score.
scoer
scroe
screo
we all love it and we all know it
You have a lot of it.
I love SCOER!!!!
I agree
I also agree
I love respektive
I love scfreeeoeoerrr🎉🎉🎉🎉🎉🎉🎉🎉🎉🎉
\o/
3:16 Another thing here are people like xKirito, that forced scores like low combo - it can lead to absurd cases.
FGSky lost 2 potential game PP records... because plays were worth more PP, but less score to his 4-mod play on Justadice at that time.
Yeah...
I played the ouendan games as a kid and I think it's why I react so negatively (not violently but usually an audible wth) to combo breaks. Combo breaks in ouendan could easily halve your score or worse compared to an FC whereas combo breaks in my two current favorite rhythm games barely punish you at all, Muse Dash might knock 5-10k off in a game where 400-500k scores are pretty common on harder charts and the way scoring works in Beat Saber where base score is really granular and dependent on precision and follow-through and the punishment for a combo break is pretty weak (you drop one rank of multiplier) it's pretty likely that a more fluid run with a couple of combo breaks will outscore a more stilted FC.
Score! We all know it, and we all love it.
When you click on a note you gain score depending on how accurately you hit.
0 if you miss, 50 and 100 if you time your click incorrectly, and 300 for a perfect hit.
Your combo factors into this, increasing the score you get from each subsequent note.
Along with mod multipliers, all of it is added to your total for every object.
You can’t forget about spinners which can provide bonus score for each extra revolution!
So it’s quite simple, if you hit more accurately and get a higher combo you
should get a higher score… Oh… okay.
The way score is calculated is actually much more complicated.
Everything I’ve said so far is still true, albeit a simplification.
How much of a simplification? Here is how ScoreV1 works.
There are four types of judgements in osu!, the miss, 50, 100, and 300. Right? Wrong.
There are actually seven types of judgements including the 100 Katu and 300 Katu,
which give no additional score, and the 300 Geki when a perfect combo is completed.
What does that mean? We don’t have time. Circles will give the base amount of score.
Sliders are broken into base anatomy and whether you hit
certain parts determines the judgement given. Heads don’t matter, ticks are worth 10 score,
repeats and ends are 30 apiece, and at the end the full slider judgement is given.
Spinners will give base score for the judgement as well as two types of bonus points which are
not affected by other multipliers, such as mod multiplier or combo multiplier.
For every rotation before the spinner is complete you gain 100 bonus points,
and after complete 1,100 bonus points. Every single one of these types of scoring
objects will raise your combo, and at each point in time of hitting a note.
The judgement is multiplied by combo times mod multiplier times difficulty
multiplier divided by 25 plus one. What is the difficulty multiplier?
To get this the stats of a beatmap, that’s CS, OD, and HP, (AR not included) are added up as
well as the return of a piecewise function that’s based on how many hit objects there
are divided by the drain time in seconds. And then we multiply that by 5/38ths because
it wasn’t arbitrary enough. That! Is how we calculate
the score for one single note. Now we add to this to the score
total throughout the play, doing the calculation again and again for every single note that is hit.
Too complicated? Maybe my simplified explanation was good enough.
The current scoring system, officially known as ScoreV1, like all other overly-specific
osu! phenomena originates from the Ouendan series. At the core of this is increasing combo leads to
exponential gains, because a higher combo play should be in theory a better play.
For the most part this system is fine for its original purpose,
but a few properties of the formula lead to less than ideal results in reality.
There’s a reason why “getting ScoreV1’d” is now a common phrase, directly relating to the tendency
for objectively better plays to award less score. You might have already spotted the first
glaring issue: When you drop accuracy matters. Because the hit value is increased by your combo,
objects later in a beatmap contribute much more score than objects that appear earlier.
Two plays with exactly the same accuracy and combo can have different scores.
Because one had more accuracy mistakes earlier in the map
compared to later on, it gets a higher score. Much to the ire of the playerbase, it is actually
very common for slightly low accuracy plays to beat higher accuracy ones on leaderboards.
Whether this is due to the aforementioned combo nuance, differences in how fast players spin,
or both is highly dependent on the situation. Regardless, this happens so frequently that I
am willing to bet every osu! player has encountered a situation where a higher
accuracy score doesn’t beat their current record, Or even worse, they accidentally replace their
current play with a worse accuracy one. The second big one, that has ravaged
many a leaderboard, is with mod multipliers. Because contrary to what many initially think,
the multiplication isn’t just done to the hit values, but also for each mod.
Famously, DoubleTime alone actually has a lower mod multiplier, 1.12x, than Hidden and HardRock
combined although it says they’re equal in-game. This is because 1.06 times 1.06 equals 1.1236,
which is rounded for the display. It can be quite annoying to see
a DoubleTime SS which is beaten by a lower accuracy Hidden HardRock score.
Nowadays on maps you can submit a score for each different mod combination, but before an update in
2016 only one play was saved no matter what mods you played with: the one with the highest score.
So let’s say you’re Cookiezi on April 15, 2013 setting the first ever DoubleTime FC
on Raijin no Hidariude… well that’s a shame because you already have a Hidden HardRock FC.
Due to the extra multiplier that one’s worth more score, and your new FC (that no one could
match for the next three years) doesn’t submit! Finally, as a cherry on top, it is practically
impossible to calculate the score of a play without taking the exact inputs from a replay.
These issues are nothing new, and several proposals for new scoring
systems have popped up over the years. To make tournament matches more fair,
ScoreV2 was created in 2015, and has been the standard for competitive play ever since.
All plays are Standardized to be a maximum of 1,000,000 points before mod multipliers
and spinner bonuses are included, with 70% of your score coming from
combo and the other 30% from accuracy. Lazer, the next version of osu!,
uses a sort of "ScoreV2.1" which shares the same base concepts with room for changes.
Currently there is debate over the percentages and how to balance the mod multipliers,
but it will continue to inprove as Lazer continues development.
Hey, and if the 1,000,000 point max isn't your cup of tea, Lazer also includes a setting
which can convert the Standardized scores into very close estimations of V1 values,
but without the aforementioned drawbacks associated with the old system.
Thinking a little outside the box, the purpose of a scoring system is to reward better plays
with more points, so many imagine a future where performance points can replace score entirely.
Even though pp famously has its own glaring problems,
it may be worth considering down the line. Maybe there's another, even better system
which we haven't even dreamt of yet. Until then, we'll have to live in a
ScoreV1 world, and deal with the downsides that come with it, as we have for 15 years already.
Score has always been a mess, but at least we’re working towards
a future that is much less messy. Who knows where scoring will go next.
so true 😘
oi u spoilin the vid ;-;
We need to stop doing this💀
Nice transcript
I ain't readin allat
I did my final AP Calculus project on osu!'s scoring system. It's way more in depth than I thought it was lol.
Man I tend to not play the same map after I passed it for the first time but I still managed to find myself getting scorev1'd lmao
0:14 how many tries did this take to get this
0:07
And they recently made a pr on lazer to have scorev2 instead of the thing they had
i once forgot to turn off nofail then beat my pp record and it was like 3k points above previous score
What I still don't understand is how is it possible that mod multipliers don't work with low object counts? Like the leaderboard on the Ultra Begginer diff on MIIRO vs. Ai no Scenario is riddled with NFEZHT scores tied with HDHRDTFL. Also, if CS, OD and HP gets accounted for the hitobject values, why do the first few notes add their base value to the score? No matter what, the first few notes will always add 300 (or 100, 50 or 0) to your score. It's weird x3
The note density and difficulty stats on Ultra Beginner are so low that the difficulty multiplier is only 0.45, which rounds down to 0, and that's the part of the formula that mods affect.
Combo scaling also has a -2 as part of the formula, so the first 2 notes of a combo are only worth base value and you start getting combo + mod multipliers from the third note.
mania players looking at score V2: Look what they need to do to achieve a fraction of out power
i'm sorry for beating Aricin's DT SS on Shinsekai with HDHR 99.78% FC :C
Can someone please direct me where I could find out more about what was said at 1:03 ( where he mensioned that we dont have the time to explain that) I would like to learn about it throgh a video since reading about it doesnt help me as much. I tried to find videos but none of them actually explain it. Any help would be appriciated!
thank u willy
scorev1 troll
2:41 lol that surprised me
Spinners my beloved
I love the fact that we’ve had an objectively better scoring system for the past 8 years and we just… won’t switch to it
Its not fun
Combo game
@@MrTerrorM cause getting scorev1'd is fun yup mmmhmmm
Just don't break combo. Its a combo oriented game.
Not getting acc drop on slider ends is nice tho
I love score! And osu! And cookiezi shigetora chocomint!
I hope someone noticed but good job on the sojourn collab fc
Like how in catch an A or S can be better than an SS because of score gotten from spinner. Mania's scoring is sooo good. It makes the game fun. I don't always worry about misses, I just try to get a good score that I feel good about. The 7k mania pp record LITERALLY has 30 misses. :(
Some #1s on catch convert lbs are D scores
@@geiboi77 Yeah, exactly.
I used to really enjoy playing mania for this exact reason. But I do have to say that actually getting an fc for a good score feels so much nicer on std. For me atleast, mania FCs feel really redundant and not important or special at all. Also imo even if its more of a grind it makes the pp records seem more important and special because it is what you need to get a good score, whereas with mania you couldve missed like 10 times and still gotten a really good score.
@@xd3athclawx554 Hm, I guess it really depends on the score for me tbh. It sucks that in standard you HAVE to fc to get ANYTHING. Catch requires fcs but not as much it feels like.
@@Imaproshaman7 Yeah I get that, not getting anything from a play because it wasn't an fc is kinda annoying but I usually don't mind because I don't usually go for pp unless I know I can fc the map in the first place, and then if I 1 miss I'm annoyed because I 1 missed not because I didn't get any pp.
Math is all
what?¿?¡'¿??¿ i have peen playing for 10 years and i didnt know that u could have several scores with different mod combinations. 7 years without knowing
Idc that it emulates the original game stepmania was a ddr emulator but it didn't force itself to copy its frame based timing window wanna know why? because its shit so is the original osus scoring system so idk why any sane individual would look at that complex mess that favors sloppier play without misses and think they should copy it
Because osu! is a simulator first, original game later. So peppy developing the original score system from the osu tatakae ouendan make sense.
And it's not like Peppy don't want to make it, he does, but people don't like changes. See ScoreV2 and lazer.
I Love You Willy.
i have been playing score v2 for about a year now, best decision ever made
SHOUTOUTS:
arlos
large fridge
very small giant
this is why i play beat saber
yes
Funny how osu! is such a simple game but so complex at the same time
what is "clamp" in that caculation?
clamp is technically a mathematical function by its nature, but is more or less referred to as something in programming.
Its more complex math than I have studied (it looks like it anyways, and I haven't ever seen it before) but what I found was this, clamp(x) = max(a,min(x,b)) ∈ [a,b]
it looks similar to the thumbnails clamp where it has clamp(8 objects/time , 0,16)) so it could be that.
I think that part is just a way to simplify a part of the math on calculating the speed or bpm of the map through the amount of circles (maps using 6ths and such) because in the editor you can go from 1/1, all the way up to 1/16 or 16ths, and in the clamp it goes 0,16, stating the range of the function, ∈ [a,b], ∈ [0,16] or a part of the set 0 to 16.
I also found someone say "Looks like clamp(x) is a projection onto an interval" Which would fit in with the interval ∈ [0,16].
Hope this is what you were looking for :)
#SCOER
The mania scoirng system is cool, its basically: your acc = your score-ish
Now the Mania pp syatem is even better: your acc/100% acc
meanwhile, in osu standard pp scales almost linear with combo
miss in the middle of the map to halve the pp rating
This is more math then i've ever done in my life
Eyy a new willy video!
scoer
ok
Did you really just skip over the one thing people are obviously wondering about by saying "we don't have time"
cool i guess?
omg you sound like the voice from the osu tutorial
and this video could be from an osu turtorial
This video is an osu tutorial
uwwu
ok