Guest
home
beatmaps
rankings
community
store
help
forum
  1. Forums
  2. osu!
  3. Gameplay & Rankings
  4. osu!mania

What do you want in your Top Performance plays?

posted
Total Posts
30

What kind of % do you want on your top performance plays?

99%
26
27.08%
98%
6
6.25%
97%
34
35.42%
96%
12
12.50%
95%
12
12.50%
94%
1
1.04%
93%
2
2.08%
92%
1
1.04%
91%
1
1.04%
Other
1
1.04%
Total votes: 96
Aqo
2,710 posts
Joined April 2010
Topic Starter
Aqo
What kind of plays do you want to see in your top performance plays, if the PP system was changed and you got different PP per score?

Right now most people's top performance plays range between 89% and 95%, so lets say this averages on 92%. Would you prefer if 90% scores gave much less PP and you'd need to 95%+ a song to start getting PP from it? or the other way around? etc

Please vote and write your reasons.

For example right now 600K plays already give a large chunk of PP out of 1M, would you prefer if 600K gave very little to no PP (like 500K does right now) and instead 700K was the lower bar? or other?
go review 20-30 of your own plays and think what kind of plays with what kind of score ranges you want to see in your top performance.
Blocko
osu! Alumni
742 posts
Joined February 2014
Blocko
Gonna vote for 97% here.

97% feels like a solid S to me, unlike 95%s when it's like a low S or a really really really high A (looking at you 95.00% A)
Taking in the fact that 97% plays here are around AAs in SM, top performances should be around that range.

Also, I have a suggestion as to how much pp should give based on score
Score -> pp percentage
500k -> 5%pp
600k -> 15%pp
650k -> 25%pp
700k -> 35%pp
750k -> 50%pp
800k -> 65%pp
850k -> 80%pp
900k -> 88%pp
950k -> 95%pp
975k -> 98%pp
1mil -> 100%pp

Personally, 600k plays shouldn't really give a lot of pp since you can mashpass IN2K6 Lunatic and get a lot of pp from it, provided that you actually combo'd the whole map except for the mashing part. I don't know if the jump between 700k and 850k is quite high, but 750k should be the threshold for about half the pp a map can give.

Don't know how to feel about LN maps though. Those are quite hard to make a good score on.
Last edited by Blocko , edited 2 times in total.
Tristan97
189 posts
Joined November 2013
Tristan97

Blocko wrote:

Don't know how to feel about LN maps though. Those are quite hard to make a good score on.
This should be taken care of/balanced with the new formula, hopefully
Bobbias
1,447 posts
Joined March 2011
Bobbias
Your first post should be reworded. While score an accuracy are closely linked, you should be absolutely clear about which one this would be based off since score is not purely a function of accuracy.

I actually don't think it's necessary to heavily change the pp scaling. To me, passes on harder songs should be given some respect.

Given that PP will be nerfed, I think any PP scaling changes should be left until after the PP calculation nerf, if at all, or we risk skewing things unfairly towards score/accuracy plays.

Consider that 15% of the maximum PP of a relatively difficult song (such as JinJin's Haelequin, which I currently have a 600k pass on) could in theory be worth less than 100% of a much easier song's PP value. As it stands, 15% of Haelequin's max PP works out to 99.15, which is so low it would not even make it on my top scores list (even the second page).

I'm not sure about you guys, but when I pass a song that's actually hard to pass (and Blocko, IN2K6 Lunatic is not an easy pass, even mashing. If it was, I'd have passed it already) I damn well want to be rewarded for finally passing. I'd be pretty pissed if I passed something I've been trying to pass for months only to find out that it's worth less points than some random crap I 99%'d on sight.
Tristan97
189 posts
Joined November 2013
Tristan97

Bobbias wrote:

Consider that 15% of the maximum PP of a relatively difficult song (such as JinJin's Haelequin, which I currently have a 600k pass on) could in theory be worth less than 100% of a much easier song's PP value. As it stands, 15% of Haelequin's max PP works out to 99.15, which is so low it would not even make it on my top scores list (even the second page).

I'm not sure about you guys, but when I pass a song that's actually hard to pass (and Blocko, IN2K6 Lunatic is not an easy pass, even mashing. If it was, I'd have passed it already) I damn well want to be rewarded for finally passing. I'd be pretty pissed if I passed something I've been trying to pass for months only to find out that it's worth less points than some random crap I 99%'d on sight.
I agree with this, espcially when a person's top ranks is usually what you want to use to judge a person's skill.

Oh, they have an A on Achromat Black Another? That means more to me than having a 97% on Savior of Song Extra per-say. Okay, so that's about the level they're at. The challenging clears should be rewarded such as in a certain beatmania simulator we all know. However, this leaves the clarification of what constitues as a clear up to discussion and debate. Osu!mania is 'hard clear' in nature, but clears such as sister's noise Lv.42 where Cs can survive through are some something but should not be worth nearly as much as an A or an S on it. There needs to be a certain point where you can go past 'mashing' to 'barely clearing'.


I am probably not qualified to specify that, but I think it shouldn't only be getting the equivalent of the 'mastery' of a SM 'AA'.

Should top pp scores be AAs or should they be clears? Where do you draw the line? That's up to everybody to decide, but let's try not to turn mania into a combo/acc fest or else we'll see the return of an era that was long dominated by autoconverts and that whole messed up system.

Best of luck to everybody to finding terms to agree on to decide how osu!mania will progress in the coming times and big shoutout to Aqo for his dedication to helping with the new SR calculations.

Ago pls
Tidek
1,571 posts
Joined March 2011
Tidek
INSANE accuracy scores on easier maps should be worth a lot more. Its funny as fuck when there is only 1-2pp difference between 990k and 999k score on for example 3,75* map. (as we know there is BIG difference between 990k and 999k)

I mean, ratio between 300g and 300 should be worth A LOT MORE because atm it gives almost nothing. (0,01pp for 30x300g more, are u fking kidding me?)

EDIT:
In my opinion, pp for maps should be more scaled.

For example, we have a 3,75* map. With current system it looks like this: (numbers are not ideal accurate, but they are close to what i mean)

1mln - 175pp
999k - 174,90pp
995k - 174,50pp
990k - 174pp
950k - 164pp
900k - 150pp
800k - 120pp
600k - 80pp

As we can see, 300g is totally ignored in current system, 300g and 300 are the same for pp.

It should be more scaled, something like this: (same map)

1mln - 350pp
999k - 300pp
995k - 200pp
990k - 175pp
950k and under, similar like in above system

Point of this is that its getting more harder to improve our score when its getting more closer to 1mln.
Its A LOT easier to improve a 700k -> 800k score and gets a tons of pp from it than improve 990k -> 999k score and get max 1-2pp with current system.
Last edited by Tidek , edited 1 time in total.
Blocko
osu! Alumni
742 posts
Joined February 2014
Blocko

Bobbias wrote:

Consider that 15% of the maximum PP of a relatively difficult song (such as JinJin's Haelequin, which I currently have a 600k pass on) could in theory be worth less than 100% of a much easier song's PP value. As it stands, 15% of Haelequin's max PP works out to 99.15, which is so low it would not even make it on my top scores list (even the second page).

I'm not sure about you guys, but when I pass a song that's actually hard to pass (and Blocko, IN2K6 Lunatic is not an easy pass, even mashing. If it was, I'd have passed it already) I damn well want to be rewarded for finally passing. I'd be pretty pissed if I passed something I've been trying to pass for months only to find out that it's worth less points than some random crap I 99%'d on sight.
600k seems like a mash pass, but a pass is still a pass no matter how you look at it, and you should be rewarded for it regardless, which brings me to another suggestion:
Obviously, the bigger the star rating, the more pp rewarded, but the pp reward percentages increase as you reach into higher numbers. The percentage boost will stop as you reach higher scores, though. In this way, your plays can be more accurate to indicate where your skill level is around at.
This is actually what we have right now with osu!mania's current pp system, but the differences between 650k and 750k is huge compared to 750k and 850k.

We could also enable DT pp so we can have an overview on how much pp hard maps should give. This is a huge factor as there's not really a lot of hard maps in osu!mania.

Tristan97 wrote:

Should top pp scores be AAs or should they be clears? Where do you draw the line? That's up to everybody to decide, but let's try not to turn mania into a combo/acc fest or else we'll see the return of an era that was long dominated by autoconverts and that whole messed up system.
Damn, those are thought provoking questions.
I'm all for 97% plays being a good chunk of my top scores, but for harder maps, some cleared plays should be up there as well so people can accurately see what skill level you're on.


Tidek wrote:

INSANE accuracy scores on easier maps should be worth a lot more. Its funny as fuck when there is only 1-2pp difference between 990k and 999k score on for example 3,75* map. (as we know there is BIG difference between 990k and 999k)

I mean, ratio between 300g and 300 should be worth A LOT MORE because atm it gives almost nothing. (0,01pp for 30x300g more, are u fking kidding me?)
That'd be a good idea, but idk if we'd want this game to have more accuracy grinding. For some people, it'd be kinda frustrating to try to get a better acc on a song that you can FC just to squeeze the last pp out of it, wouldn't it?
Tidek
1,571 posts
Joined March 2011
Tidek

Blocko wrote:

That'd be a good idea, but idk if we'd want this game to have more accuracy grinding. For some people, it'd be kinda frustrating to try to get a better acc on a song that you can FC just to squeeze the last pp out of it, wouldn't it?
Not really, when u read my edited post above, u will see that getting 999k on for example 5* map will be rewarded with tons of pp. atm there is no difference between getting 990k and 999k on a 5* map, that is really sad.
Accuracy should be a main factor here on EVERY kind of map.
BilliumMoto
osu!mixer II
190 posts
Joined January 2014
BilliumMoto
Tidek, you're one of the few players who cares so much about accuracy. Instead of trying to vehemently force a change in the game so that more people will be like you, just accept that ego boosts with their shiny SS's by replaying the same map 2000 times is not how most players want to play this game.

Same goes for your hate for overmaps :/
Tidek
1,571 posts
Joined March 2011
Tidek
Are you trolling or what? If not, maybe try to read what I write again, because it seems you dont understand a single thing. Its not about SS, its about score differences and 300g which is totally ignored in current system.
Blocko
osu! Alumni
742 posts
Joined February 2014
Blocko

Tidek wrote:

Not really, when u read my edited post above, u will see that getting 999k on for example 5* map will be rewarded with tons of pp. atm there is no difference between getting 990k and 999k on a 5* map, that is really sad.
Accuracy should be a main factor here on EVERY kind of map.
By doubling the amount of pp reward from 990k to 1million, it's gonna make the game more acc grindy, which is not really what most of us want. It's pretty much catering for accuracy players and not for the ones that actually try to get a good score on it, only to find out that the pp reward is actually a lot less, which is pretty discouraging for the majority of players here.
Honestly, I agree with your suggestion about giving 990k+ scores a boost, but I don't really think this is a good way to change it.
Tidek
1,571 posts
Joined March 2011
Tidek

Blocko wrote:

By doubling the amount of pp reward from 990k to 1million, it's gonna make the game more acc grindy, which is not really what most of us want. It's pretty much catering for accuracy players and not for the ones that actually try to get a good score on it, only to find out that the pp reward is actually a lot less, which is pretty discouraging for the majority of players here.
Honestly, I agree with your suggestion about giving 990k+ scores a boost, but I don't really think this is a good way to change it.
The same discouraging is improving 990k score to 999k on pretty hard map and gets almost nothing from it. (some ppl dont even realize how big difference is between 998k and 999k).
Maybe yeah doubling pp is really too much, but little pp boost for sure is needed in scores closer and closer to 1mln.
Shoegazer
osu!mania Paragon
528 posts
Joined April 2013
Shoegazer
This was a significant problem I had with the pp system, glad that this topic was made. As a score gets better and better, it gets harder and harder to improve the score because of less room for error, it would make absolutely no sense that you already get 85% of the maximum PP with a super low S and 60% of the maximum PP for a super low A. I think a good baseline would be a 97%-98%, that should be about where a person would be getting about 80% of the maximum potential PP from there. That's about 895k?

As for the argument about superb accuracy being reflected... I do think that it should be reflected. Definitely nowhere as much as Tidek's, but it's noticeable enough to differentiate between the the sharpshooters and the normal SSs.

Been trying to work on a formula that seems fine with several people, this is probably the closest to whatever seems fine:


If score is under 990k
0.95 * e^(score/618750) / e^1.6

If score is over 990k
0.95 * e^(score-10000/618750) / e^1.6 + 0.05 * e^([score-990000]/10000) / e

A secondary function is used for scores above 990k because the exponential graph wouldn't be able to capture the significantly higher difficulty of going from 990K to 1mil. There's probably a better way to do this, but I don't know math at all.
Ciel
170 posts
Joined March 2012
Ciel
I may not be the best person to talk here, given that I believe my highest score on any map ever (including the easiest ones) is somewhere around 970k. (Note that I have 0 SS's in osumania, rare for anyone remotely near my rank).

But the difficulty of obtaining some score for each map is quite different for each (as well as the fact that the upper end of high scores feels largely determined by purely how good you are at accing easy to play maps.) Space time is probably the most obvious example, with a pass usually easily getting somewhere around 700k, having a fairly difficult to climb to 850-870k, and then an easy climb to some other number, more difficult up to 950i-960sh (?) k, and then accuracy fest. This is due to the fact that the majority of this song is easy compared to the difficulty spikes, and even within the difficulty spikes, they come in 2 parts of varying difficulty. (These numbers are probably pretty bs, but I'm just trying to illustrate my point)

Obviously this isn't really feasible to do for a difficulty system though, but it might be better to collect some data on plays in general to find the average curve for score attained.
Tristan97
189 posts
Joined November 2013
Tristan97

Yoshiap wrote:

The difficulty of obtaining some score for each map is quite different for each (as well as the fact that the upper end of high scores feels largely determined by purely how good you are at accing easy to play maps.) Space time is probably the most obvious example, with a pass usually easily getting somewhere around 700k, having a fairly difficult to climb to 850-870k, and then an easy climb to some other number, more difficult up to 950i-960sh (?) k, and then accuracy fest. This is due to the fact that the majority of this song is easy compared to the difficulty spikes, and even within the difficulty spikes, they come in 2 parts of varying difficulty. (These numbers are probably pretty bs, but I'm just trying to illustrate my point)

Obviously this isn't really feasible to do for a difficulty system though, but it might be better to collect some data on plays in general to find the average curve for score attained.
This would require you to detail each map specifically, which isn't a bad idea, but is sadly almost definitely not going to occur, hence an 'all-purpose' general curve is necessary, which is what a number of members in the community is passionately arguing (in the intellectual definition) over. I like what Eze is putting out, but of course, there's still plenty to debate at this time.



I'd like to remind people right now that if the system implements with the ~97% we're polling, only about 20-30 people in the entirety of osu!mania would have Aiae SHD, Imperishable Night 4K Insane or 7K Lunatic as one of their top ranks. For Maniera Masterpiece that number would essentially be in the single digits. Is this a good thing? That's up to our discretion, just wanted to throw that numbers out there.
_underjoy
704 posts
Joined January 2013
_underjoy
On topic, the Top Performance plays should be rewarding mostly players who can S a map. osu! is a game where S is the grade that really makes a difference, look on osu!standard, taiko or ctb, there all players are going for their S'es because A's mean a lot less.

Now of course osu!mania is different, but I'm talking about 95% threshold. The pp for S should be around 70-75% of the max value (scorewise that's a little more than Shoegazer's function but I agree with his). Now the pp for SS'es should be higher than it is, especially in the 990k-999k section as Tidek stated it. Maybe not doubling, but 150% for the 1mln (compared to current 100% for 1mln) seems fair. 999k would be 140%, 998k 125%, 997k 110% and 996k 100%. I gave these numbers without testing but you got the idea. PP for passes is a little too high, when you can get 200-300pp for a B it is kinda meh (more than SSing a 4-4,5* map which isn't accurate at all)

Off topic, the current system isn't as flawed as the star rating system, which needs rework ASAP, because it sucks on so many levels it isn't even funny..
I can find a 4* map that is harder than 6* map, and both maps are not troll maps. This is not normal, provided that star rating should reflect the real difficulty and allow to compare maps. Not to mention the pp for higher * maps scales so quickly it DENIES pp for 4K maps. To get real pp for 4k map, you would have to make a 6,5* map which is IMPOSSIBLE without vibrawalls (where in 7k you can just play a dense, good map). I won't be a 4k fag, but I need to say the pattern recognition (and stamina recognition) should come alive instead of measuring max density as a main factor of song's difficulty.

It's sad the suggestions are made only after 25 players got over 10kpp, not earlier. I would say that's a little late. Better late than never though :D
akuma123
308 posts
Joined June 2011
akuma123

TheZiemniax wrote:

Off topic, the current system isn't as flawed as the star rating system, which needs rework ASAP, because it sucks on so many levels it isn't even funny..
I can find a 4* map that is harder than 6* map, and both maps are not troll maps. This is not normal, provided that star rating should reflect the real difficulty and allow to compare maps. Not to mention the pp for higher * maps scales so quickly it DENIES pp for 4K maps. To get real pp for 4k map, you would have to make a 6,5* map which is IMPOSSIBLE without vibrawalls (where in 7k you can just play a dense, good map). I won't be a 4k fag, but I need to say the pattern recognition (and stamina recognition) should come alive instead of measuring max density as a main factor of song's difficulty.
THIS
Bobbias
1,447 posts
Joined March 2011
Bobbias

TheZiemniax wrote:

star rating system, which needs rework ASAP, because it sucks on so many levels it isn't even funny..
In case you're not aware, Tom is currently working on a new star rating system for mania, so there's no more need to bitch and complain about what we have until the new one is out and we find all it's flaws.

What about scaling the 'bonus' for going from 990 to 1m (or 995to 1m or whatever) by the difficulty of the map, such that on easy maps there's next to no bonus, but as the difficulty of a map increases, the reward for achieving high accuracy also increases. That would mean players capable of getting impressively high scores on actually hard maps are rewarded but lower skill players are not as encouraged to accuracy race.
Tristan97
189 posts
Joined November 2013
Tristan97

Bobbias wrote:

What about scaling the 'bonus' for going from 990 to 1m (or 995to 1m or whatever) by the difficulty of the map, such that on easy maps there's next to no bonus, but as the difficulty of a map increases, the reward for achieving high accuracy also increases. That would mean players capable of getting impressively high scores on actually hard maps are rewarded but lower skill players are not as encouraged to accuracy race.
This would be great, but you'd have to put it specifically into the SR algorithm, like you'd have to do for a possible slight SR buff for uber high ODs or something. It wouldn't go with the 'one size fits all model', which isn't a bad thing in my opinion, it just matters how/if it would be implemented.
Bobbias
1,447 posts
Joined March 2011
Bobbias
I don't see how you'd need to implement it as part of the SR. My point was that when determining what fraction of maximum PP to give players, you could scale only the scores above some certain threshold based on the SR difficulty.
Tidek
1,571 posts
Joined March 2011
Tidek

Bobbias wrote:

What about scaling the 'bonus' for going from 990 to 1m (or 995to 1m or whatever) by the difficulty of the map, such that on easy maps there's next to no bonus, but as the difficulty of a map increases, the reward for achieving high accuracy also increases. That would mean players capable of getting impressively high scores on actually hard maps are rewarded but lower skill players are not as encouraged to accuracy race.
100% agree.
Jinjin
886 posts
Joined September 2013
Jinjin

Tidek wrote:

Bobbias wrote:

What about scaling the 'bonus' for going from 990 to 1m (or 995to 1m or whatever) by the difficulty of the map, such that on easy maps there's next to no bonus, but as the difficulty of a map increases, the reward for achieving high accuracy also increases. That would mean players capable of getting impressively high scores on actually hard maps are rewarded but lower skill players are not as encouraged to accuracy race.
100% agree.
Yes, the current pp with accuracy scales with a logistic curve (which means that it levels out at the top). I honestly believe that in the top range of the scores, the scaling of pp should not level out. Let's say a map has ~5000 objects. Currently, a score achieved by 4600 MAX's + 400 300's and the score achieved by 2600 MAX's and 2400 300's would not differ significantly in pp, while it should, because people should be rewarded in how accurate they hit. osu! is a rhythm game, and accuracy is a core aspect of rhythm games.
show more
Please sign in to reply.

New reply

0
1 / 30