forum

Performance Points feedback and suggestions (Standard)

posted
Total Posts
2,750
show more
dennischan
I agree that we should get pp for simply passing hard songs.
It could work like this:
pp gain from passing freedom dive+senketsu no chikai+ DT warheit
You can just calculate the difficulty to pass those songs:

The formula should be something like this: beatmap difficulty*HP drain*log(longest combo length)
and then we get a fixed amount of pp for each song.
For example, we get 50pp passing freedom dive, 45 for senketsu, and 80 for DT warheit
Then total gain of pp: 80*100%+50*95%+45*90%
then weight them all and then get the pp tweak



Also I think long songs are undervalued a lot.
It takes same effort to hold 200 combo at a long song than an short song at same difficulty, but you gain more pp by playing the short one

200/220 combo at anime opening
200/2000 combo at crazy hard renard song

more pp for anime opening... That just dosen't make sense.
I think we should get more pp from long songs, and it should work exponentially

something like: base^(length of song in minutes)*original pp

and of course FL bonus should scale with the length of the song...
and it should work exponentially also
nooblet

dennischan wrote:

I agree that we should get pp for simply passing hard songs.
It could work like this:
pp gain from passing freedom dive+senketsu no chikai+ DT warheit
You can just calculate the difficulty to pass those songs:

The formula should be something like this: beatmap difficulty*HP drain*log(longest combo length)
and then we get a fixed amount of pp for each song.
For example, we get 50pp passing freedom dive, 45 for senketsu, and 80 for DT warheit
Then total gain of pp: 80*100%+50*95%+45*90%
then weight them all and then get the pp tweak
Getting PP for passing songs is probably not a good idea. Someone who can mash their fingers fast can /probably/ pass chipscape or mad machine or any other crazy stream map (NF easily negates HP, since that's only 0.9*PP) with half-decent accuracy, but that doesn't mean they can play the game.

I agree with the long song / FL stuff, but it'd also overrate easy marathons like this. Dunno, should probably be scaled.

It's hard to determine, though, because a lot of songs are hard not because of the overall difficulty of the map, but rather a single part. For example, Snow Goose would be a piss-easy map if not for the deathstream / short streams near the end. Especially since that stuff is all at the end, getting a combo for the beginning of the map wouldn't be too hard, and thus the map becomes free PP.
DT-sama

dennischan wrote:

I think we should get more pp from long songs, and it should work exponentially

something like: base^(length of song in minutes)*original pp
That would make marathons overrated, and length from not-so-relevant would become the only relevant factor for pp. Some marathons are so easy you can hold multiple 1000x combos with no effort at all.

I agree that length is underrated, but I think the length bonus should just decay slower.
jesse1412
Long streams are so overrated it hurts me inside. I spent like 40 mins trying to plot graphs for how the pp bonuses for longer maps could work and 4d comes along every time with it's bullshit and ends up with an insane bonus. The system is meant to measure difficulty to fc the maps and 4d is insanely overrated in that department. I'm either going to assume that the current system takes into account the length but not that well when calculating stars which is making it all go funky OR freedom dive is overrated when it comes to difficulty to FC.
dennischan
the mighty jesus has spoken
on topic: why dont nerf long streams
example: CFDgimme/chase the light /4 dimensions whatever
and maybe boost banned forever...
haha5957
I just saw Tom's reply saying that combo is factored as performance*(combo^0.8/maxcombo^0.8.)

However I personally think there should be a reasonable amount of pp given just by passing the map itself.

For example - your pp reward= performance*0.35+peformance*0.65*(combo^0.8/combo^0.8)
In this case I counted the "pass" reward 35%, so obviously if you pass a map with 180/1800 combo it would yield you nearly no pp with current system, however this should reward you a bit if the map's difficulty is super high.

maybe this number can be differed according to the HP Drain(and this should be reasonable solution to award high drains slightly more).



think of airman or FD. You barely passed it nomod and gets rewarded no pp.. I find this slightly unreasonable.
tfg50
The problem with just passing is that this system doesn't punish playing with NF, it gives the same PP as playing without it (iirc). Also, I know people that can pass insanely hard stuff when comparing what they can pass to easy maps that they can't even fc (https://osu.ppy.sh/s/120002 + dt vs https://osu.ppy.sh/s/55945 nomod)
nooblet

tfg50 wrote:

The problem with just passing is that this system doesn't punish playing with NF, it gives the same PP as playing without it (iirc). Also, I know people that can pass insanely hard stuff when comparing what they can pass to easy maps that they can't even fc (https://osu.ppy.sh/s/120002 + dt vs https://osu.ppy.sh/s/55945 nomod)


No Fail reduces it by 10%.

Also not sure if jesus is serious when he says 4D is overrated ...
Vuelo Eluko

nooblet wrote:

Also not sure if jesus is serious when he says 4D is overrated ...
i always thought 4d was one of those ridiculously epic songs and anyone who fcs it should feel quite rewarded for it.. the map is basically osu! lore...

i honestly think the overrating problem is more prevalent on shorter maps like chase the light
Shimatora

- By zxmuffinxz


- By myself.

I was considering FLing this map to squeeze some more PP out of it but come on. Seriously? SS no-mod is currently considered to be worth more than FL+HD, albeit an S and 1.49% lower accuracy.

~Insert FL-is-underrated-still rant here.~
Vuelo Eluko
if the map is easy to FL [I.E low aim rating, notes arent too far apart] then FL doesnt give it much PP

presumably hidden acts on aim difficulty too so
Shimatora
Considering I put a good 100 tries into FL FCing the map earlier, and considering it took me under 20 to SS the map, I'm just finding it hard to believe that a FL+HD with a couple of misses is worth less than a full combo that had none, considering the blind jumps. I'm not expecting it to be worth 200pp or something stupid, I'm just expecting PP to actually reflect the difficulty of a map correctly. Hence my bringing it up. I feel that the PP awarded does no match the skill required to pull off.
silmarilen
the map is od8 and has virtually no aim rating, it's only natural when you consider the way pp works that this would give less pp.
comparing it with tp, you lose 16 acc and gain 15 aim.
Shimatora

silmarilen wrote:

the map is od8 and has virtually no aim rating, it's only natural when you consider the way pp works that this would give less pp.
comparing it with tp, you lose 16 acc and gain 15 aim.
It's just dumb how FL seems to be completely obsolete now. The only difficulty mod that you shouldn't bother using if you want to gain PP because, as my friend so eloquently put, "memo isn't a real skill."
silmarilen
the thing with FL is that the only difference with nomod is that you need to memorize where everything is, it doesnt require you to actually be faster/more accurate, and only kind of needs better aiming because you have to blind aim. FL gives a 50% boost to aim for this, and im not sure how you would otherwise increase the bonus of FL without breaking it.
Shimatora
Considering how small the area becomes at 200 combo, I'd say it requires quite a bit more "skill" than HD does. You don't even see where the circles pop up, let alone the approach circles if you add HD. It shouldn't be on par with DT of course, it's no where near the skill required to perform well with DT. But, as the current system has it, FL might as well just be removed from PP calculation altogether, it makes that little difference unless you SS. Basically, we're pretty much forced to either go all the way or fall short of the PP we might have had for HD or no-mod SS/higher accuracy.

EDIT: Not saying the balancing of bonuses is easy of course, just putting in food for thought.

DOUBLE EDIT: If adjusting it even slightly messed things up, the only alternative would be to have completely separate PP rankings for each mod, or something of the sort. In before I get flamed for this suggestion.
jasian
Just bear in mind that FL focuses on muscle memory and memorization alone. You have to complete the same jumps and streams in the same time as nomod so applying difficulty bonuses similar to DT/HR is out of the question. The only thing FL is comparable to is HD and the bonuses of HD have already been reduced. Yes, if you play Airman with FL then you should get a HUGE bonus because it proves that your muscle memory must be insane; but at the same time you should not get a huge pp bonus if the song is very easy to SS nomod.
Vuelo Eluko
today silmarilen agreed with me
im all fuzzy
Shimatora
I guess the main misconception I have is more tries to FC != harder. Paradoxical statement, but that's what my issue boils down to. The harder the map, the more PP. The harder to FC or SS, the more PP. If someone puts the effort in to learn a map then (of course, in my own opinion) they should be rewarded for it, not put off even attempting it (AKA just no-mod SS over FL S).

Effort != PP. Gotcha.
Vuelo Eluko
the amount of pp from a play, as i understand it, is based upon how quickly the map requires you to move your cursor, how hard it is to land on the notes, and how precisely the notes are tapped. I don't think such hard to quantify metric such as 'memorizing' a map is taken into consideration because it does not directly make the map technically harder in terms of how you have to move your fingers.

Shimatora wrote:

Precisely why I said Flashlight is obsolete. Useless outside of score. The new system is excluding FL as something that actually makes the game harder and you should be rewarded for using. Basically, we both agree. I understand WHY it happens, I don't need explaining, just offering feedback.
Should FL really, REALLY be encouraged though?
Shimatora
Precisely why I said Flashlight is obsolete. Useless outside of score. The new system is excluding FL as something that actually makes the game harder and you should be rewarded for using. Basically, we both agree. I understand WHY it happens, I don't need explaining, just offering feedback.

EDIT: Bassist Vinyl, you've literally summed up ppv2 in a nutshell. That is my quote of the fucking year right there. FL is something that shouldn't be encouraged, but frowned upon.
haha5957
Is There a clear thread, or maybe some kind of wiki, consisting of the factor what pp accounts in? I mean, I can guess that we take combo, accuracy, difficulty, etc.. but you know, the details. like, losing 10% pp with nofail, losing 5% pp with spunout, combo penalties you by combo^0.8/maxcombo^0.8...

would be much easier to feedback if we had a thread or wiki that consists of information that tom94 has currently given out..


Also I strongly disagree about FD4D being overrated(actually I think it is underrated.) but I agree that medium high bpm(150~180) long streams are bit overrated ). like, say, thousand 150bpm stream would be honestly not that difficult but even 32 streams of 222bpm is ridiculously hard. example would be Mendes, with it having low enough bpm to long stream with ease, it seems to have high difficulty and it gives me the most pp, while it was one of the most easiest map to 99%/FC it.


On the other hand, so called "0108 style maps" are very underrated. maybe high SV is bit underrated. dance number and scarlet rose are super hard but its difficulty is calculated very low, maybe because you don't actually have to "press the keyboard" that much to play those maps due to tons sliders. Assuming you are taking slider correctly into the difficulty, it seems like high SV is bit underrated on current calculation. I see jumps are correctly applied(according to the airman's difficulty) but maybe generally high SV, or map that consists of lots of small jumps are bit unreasonably underrated.

https://osu.ppy.sh/s/25271
take a look at "bloody rose" difficulty. It is not that hard, but when you play it I'm pretty sure this is hard enough to be an insane difficulty but calculated as "hard" which is ridiculous, because regular 2.49 star diffs are notably easier than this one. Compare it to https://osu.ppy.sh/b/156346&m=0 hard diff which has 2.46 diff but it is easy enough to DT it with ease. The point is, the thing that makes "bloody rose" diff hard is the high SV (which seems underrated right now)

https://osu.ppy.sh/b/27639&m=0
Intense diff has 1.91 star rating? you see nearly no DT on the ranking, which indirectly tells you how hard this map is. characteristics? high SV.

https://osu.ppy.sh/b/27737&m=0
yeah.
Ziggo

haha5957 wrote:

Is There a clear thread, or maybe some kind of wiki, consisting of the factor what pp accounts in? I mean, I can guess that we take combo, accuracy, difficulty, etc.. but you know, the details. like, losing 10% pp with nofail, losing 5% pp with spunout, combo penalties you by combo^0.8/maxcombo^0.8...

would be much easier to feedback if we had a thread or wiki that consists of information that tom94 has currently given out..
https://osu.ppy.sh/wiki/Performance_Points
dennischan
I still maintain that long songs should be boosted.
Maybe linearly, such that easy insane tvsizes are pp farms.
With current system, TV size maps are tons of pp, while really hard maps aren't worth anything...
FL should also be boosted a bit more at long songs (3mins or more) since it is really hard to memorize everything
Cifica


discuss
Full Tablet

Cifica wrote:



discuss
There are 2 different maps with the same song and difficulty name, but one is harder. Are you sure you are comparing the same map?

Edit: they aren't the same map.
Illkryn


I think this should be attended to, bugged hard diff
Aloha

Illkryn wrote:



I think this should be attended to, bugged hard diff

Topic Starter
Tom94
All beatmaps are being re-calculated to 1.5x stars, so things like that are bound to happen. As soon as everything's done pp will go back to normal, don't worry.
GoldenWolf

Cifica wrote:



discuss
You have nofail on the SS
Rewben2
dennischan
about when will the pp recalculation finish?
I still see maps with wrong star difficulty...
Will graveyarded and pending maps have a recalculation of pp?
Akari-

haha5957 wrote:

example would be Mendes, with it having low enough bpm to long stream with ease, it seems to have high difficulty and it gives me the most pp, while it was one of the most easiest map to 99%/FC it.
have you considered the fact that you might be an exception and that you happen to be really good at high accuracy on low-medium bpm long streams?

there are only 37 people who have a higher acc than you on mendes, only 11 used a mod and it's a frequently played map

most people find it difficult to break 93%
haha5957

Akari- wrote:

haha5957 wrote:

example would be Mendes, with it having low enough bpm to long stream with ease, it seems to have high difficulty and it gives me the most pp, while it was one of the most easiest map to 99%/FC it.
have you considered the fact that you might be an exception and that you happen to be really good at high accuracy on low-medium bpm long streams?

there are only 37 people who have a higher acc than you on mendes, only 11 used a mod and it's a frequently played map

most people find it difficult to break 93%
i understand that my strong point might be medium bpm stream and weakpoint might be fast SV.

I still think High SV maps are underrated. Popular Emerald sword and Strangeprogram, Dance number, scarlet rose, even myself's map those have 2.88SV feels way too hard compared to maps that have same difficulty. This is why I am "discussing" this here. Is it just that I suck at fast SV and maybe little better on medium bpm streams? Like i said, I see nearly no doubletime on emerald sword Intense difficulty which barely has 3 star difficulty. Dance number, scarlet rose is famous for its insanity but it doesnt even have extra difficulty. They have fast SV in common.
Akari-

haha5957 wrote:

i understand that my strong point might be medium bpm stream and weakpoint might be fast SV.

I still think High SV maps are underrated. Popular Emerald sword and Strangeprogram, Dance number, scarlet rose, even myself's map those have 2.88SV feels way too hard compared to maps that have same difficulty. This is why I am "discussing" this here. Is it just that I suck at fast SV and maybe little better on medium bpm streams? Like i said, I see nearly no doubletime on emerald sword Intense difficulty which barely has 3 star difficulty. Dance number, scarlet rose is famous for its insanity but it doesnt even have extra difficulty. They have fast SV in common.
yeah high SV maps are underrated because sliders are almost a nonfactor in difficulty calculation (calculation assumes that movement to and from a slider is 100% optimal iirc) and it doesn't help that high bpm singles are underrated unless they have big spacing like some of WWW's scores

this is a difficult issue to fix though, hard sliders are the exception and not the norm, and it would be tricky to calculate the difference between spaced streams, tolerable bpm singles, and high bpm singles
haha5957
It would be much easier to help out if we could know how actually slider difficulty is calculated because I really think high SV maps (or map with complex slider rhythm) are way too underrated. I generally agree with the star rating difficulty but high SV centered maps are significantly underrated
GhostFrog
This might help you. I think the current iteration of the algorithm is the same up to overall scaling, but with exact values used where the code uses approximate values. I haven't actually looked through it, so no promises.

Also, it's very much a known issue that sliders are underrated. The difficulty lies in trying to fix that without breaking repeat sliders and sliderstreams.
Oskur
Although since the topic of single taps being undervalued to some extent came up, how much should we expect pp to change when/if a solution/change is applied to the algorithm? (relative value, i.e. "holy crap some people with have giant rank shifts" or "meh it'll affect some people but relative rankings should stay about same")
Topic Starter
Tom94

Oskur wrote:

Although since the topic of single taps being undervalued to some extent came up, how much should we expect pp to change when/if a solution/change is applied to the algorithm? (relative value, i.e. "holy crap some people with have giant rank shifts" or "meh it'll affect some people but relative rankings should stay about same")
Relative rankings should stay about the same since most people who are good got a lot of other good scores aside from these singletapping ones. Some people will profit, though.
Keep in mind, that this is one of the lowest priority aspects at the moment, since there are way more important things to do at a larger score and in the other modes.
Kert
I still suggest nerfing of DT (high speed or whatever) / increasing the profit from playing smaller circles, because this is just too silly



I also got absolutely nothing (it's not even in my top performance) for HR FC this thing which is very difficult https://osu.ppy.sh/s/8442
show more
Please sign in to reply.

New reply