forum

Performance Points feedback and suggestions (Standard)

posted
Total Posts
2,750
show more
DroidBass
Ehm... you're right, there was no reasson to mention about DT topic... but I guess we can continue about the powerful nomod FC's that are oftenly seen in low pp's players. well I've seen cases of players at 1600 global pp doing a 98 pp rank at rungran with around 400's combo and some 2000-2400 pp friends of mine that has rungran aswell with 89-92% FC accuracy. Rungran is a commonly pp'ed map at 1600-2600 pp players. So yes, I have proofs that players of a wide range can do this map :P


You have these 3 cases of friends of mine that when they were 1900-2100 pp they made dmc rungran FC with A and one that at 1679 had almost done FC on it. So I would not underrate the power of low pp players that with retries can rank a 550's combo on an extra map that is built mainly on long streams. Long streams map combo are practically shorter in time lenght than a map built of 550 combos out of jumps. The shorter a map is, the more tryhardable is, that is why I know some 2000 pp players can even FC dmc rungran if they tryhard it enough.

A similar case happens with 07th Expansion - rog-unlimitation in which some players at 2800-3300 pp can do aswell a 88-91% FC on there gaining an almost 200 pp rank.

SPOILER
Player profiles that pp'ed rog unlimitation
92.34% A FC [3364 pp player]
90.61% A FC [3165 pp player]

Both of these maps are OD7 wtih high stars from streams with spacing, and both are commonly done by unexpectively low pp players.

So there is another map in which low acc FC is common and grants more than decent pp. Well yes I can't do this second one to be honest, but I see this commonly on people's best performance. Seems that a large group of people have no real problems about doing long streams wtih that spacing with CS3, I'm unsure on what to think about the pp efficiency of that map ... it's nothing easy but it's too commonly seen even on people at below 3000 pp, because the 2 friends of mine did rog unlimitation when they were a bit smaller than 3000 pp.


So ... if its true that 99% acc high OD is the most common way to have best perforamnces, it's commonly seen that some people don't need accuracy at all to make their ranks and if you expect that skill is lineally relationated to accuracy you're wrong. Likely we see DT or nomod maps that we see too often on people's best performance with 99% acc there are a group of low acc FC maps that we can see sometimes aswell and that MANY people have complaints againist these maps aswell, mostly talking about dmc rungran in which some people hates the much pp it gives to themselves.

But yes, this kind of player is not rare, but we have to admit that DT high accuracy players are most common for a group of reassons that some people on this topic have already talked about this.
Endaris
But low-acc FCs give a decent award just like you stated with two examples.
Are they underrewarded? I think not.
OD has an exponential scaling but a scaling that just ENDS at some point. So the difference in accuracy has to be huge to make up a huge difference in combo which seems fair for me.
DroidBass
I would say that they are not underrated on some nomod maps, mostly acesible at OD7 maps with high spaced streams, but they are rarely seen on other kind of maps.

The thing that I do not like is that accuracy is completely independant of map's level. This makes some weird exceptions doing unexpected high pp, more than maps with 1 more whole star at SS. Don Omar ft. Lucenzo - Danza Kuduro [Experto] is a map that with DT has OD9.67 and 4.25 stars ... a very strange case of a map with too low stars but such OD. FOLiACETATE - Heterochromia Iridis [Another] is an difficult another map of 223 bpm streams, anoying reversals and theorical star level of 5.25 stars but OD7.

Well the case is this:

Case a) Don Omar ft. Lucenzo - Danza Kuduro [Experto] +DT
has 4.25 stars, 195 bpm, AR9.67, OD9.67 & CS4, contains 223 circles & 175 sliders

Case b) FOLiACETATE - Heterochromia Iridis [Another]
has 5.25 stars, 223 bpm, AR9, OD7 & CS4, contains 361 circles & 213 sliders

maps pp ranges by tillerino (because too much lag I can't upload images this time)

For case a)
95% = 82 pp
98% = 117 pp
99% = 143 pp
100% = 185 pp


For case b)
95% = 143 pp
98% = 156 pp
99% = 165 pp
100% = 178 pp


Observations:
1.- Meanwhile a) value is "trash" from FC, it's massive OD scaling ends beating b) which is a much harder map overall.
2.- map a) gives as much pp at 99% acc as map b) but with 95% acc.
3.- Elements distribution (again)
a) contains 223 circles & 175 sliders
b) contains 361 circles & 213 sliders

Toughts
Well, map b) is very bitch to FC because anoying reversals, unconfortable positions at some circles and it is even difficult to FC for players over 4500 pp. map a) is just hard to accurate, any player that has enough aim to FC an Hard +DT can do map a) with FC and with retries can gradually rise his accuracy on map a). Result is that a much more wide range of players (even ME! at 4436 pp) can gain pp from this 4.2 stars map. Meanwhile players of even higher rank than mine have real difficulties to FC map b) for a worthless amount of pp.

That's how OD is OVERRATED (more than just "accuracy is op"). I have nothing more to say.
Endaris
The actual mapdesign and complexity is much more important.
I also thought we weren't talking about DT?

Anyway, enjoy SSing this for massive 286pp:
https://osu.ppy.sh/s/290040
E m i

ReynBolt wrote:

Well, map b) is very bitch to FC because anoying reversals, unconfortable positions at some circles and it is even difficult to FC for players over 4500 pp
right, things that pp doesn't take into account. if anything, it makes the map b underrated, not the map a overrated.
DroidBass

Endaris wrote:

The actual mapdesign and complexity is much more important.
I also thought we weren't talking about DT?

Anyway, enjoy SSing this for massive 286pp:
https://osu.ppy.sh/s/290040
argggg 138 bpm streams is too hard!!! ... seriusly the more far to 185-225 bpm the harder is it to me!!! (regardless if it's slower or faster than the given range)
silmarilen

ReynBolt wrote:

The thing that I do not like is that accuracy is completely independant of map's level.

silmarilen wrote:

acc should be a function of map complexity, not just of map length and od. too bad there dont seem to be any ways to do that.
Endaris

ReynBolt wrote:

Endaris wrote:

The actual mapdesign and complexity is much more important.
I also thought we weren't talking about DT?

Anyway, enjoy SSing this for massive 286pp:
https://osu.ppy.sh/s/290040
argggg 138 bpm streams is too hard!!! ... seriusly the more far to 185-225 bpm the harder is it to me!!! (regardless if it's slower or faster than the given range)
Please don't act like the bpm is your problem. It's the technical complexity and high OD that keeps you from getting anything close to good acc on it.
Reyvateil

Endaris wrote:

The actual mapdesign and complexity is much more important.
I also thought we weren't talking about DT?

Anyway, enjoy SSing this for massive 286pp:
https://osu.ppy.sh/s/290040
> thelewa did 99%

I pass.
Yauxo
Short: Can the PP algorhythm spot Sliders which are hit too early (100) but still give full points (300) and take that in consideration when PP'ing?



I was wondering about a few things today and wanted to ask following;

- Is the PP calculated with the replay sent to the server or "just the score"?
- In case of replay, does the replay save the Sliderhits in a value which is usable for the PP algorhythm (ex. Slider gave 300, but the initial hit was in the 100 area (xy ms off the actual timing)?

If yes, I was wondering if it'd be able to take "300 Sliders with a hit on 100/50 window" in consideration and adjust the PP with that.

From time to time I see post about Sliders being to easy as theyre a guaranteed 300. ppy wont change this as it's just too big of a deal, it'd change too much in all the (leaderboard-)scores we have now. Can the PP algorhythm spot these faulty 100 Sliders?
Because I think that that'd kinda solve the problem to some extend, as people dont really care about their actual scores, but rather about their PP. That's just my view though.
GhostFrog
Replay data isn't used, no. Just combo, accuracy, and 300s/100s/50s/misses. Even if pp could be calculated based off of replay data for new scores (and that would significantly increase the time required for calculation, which is already pretty long because of the ridiculous number of scores in existence), replay data isn't saved for current scores that aren't top x, so it couldn't be applied retroactively.
Yauxo
I see, that kills the idea then. Thanks for the answer
jesse1412
I think that was the longest time that this thread hasn't been posted in since it was made.
Yauxo

jesus1412 wrote:

I think that was the longest time that this thread hasn't been posted in since it was made.
I broke it, rip wr
Vuelo Eluko
dead game
I Give Up
I wish pp scoring was acc based like mania or at least acc based pp and with some combo bonus pp. never gonna happen :{
jesse1412

KukiMonster wrote:

I wish pp scoring was acc based like mania or at least acc based pp and with some combo bonus pp. never gonna happen :{
That would be quite retarded, combo > all.
E m i
miss count plz
Reyvateil

jesus1412 wrote:

That would be quite retarded, combo > all.
specially for you :')
-Makishima S-
Combo -> Acc

When acc/speed pp split?
DT-sama

KukiMonster wrote:

I wish pp scoring was acc based like mania or at least acc based pp and with some combo bonus pp. never gonna happen :{
Mania doesn't have an aim component like osu does so combo is irrelevant there.
TheLeviathan
Hello guys,so i might get somehing wrong but... I just finished a map with full combo, map that was weighted 111 pp and in my top ranks list i see that map on my #1 place, and it's weighted 100% (111 pp). But i didn't got 111 pp to my score, i got something like 30 or maybe little bit mor. So what i got wrong about PP system?
jesse1412

TheLevitian wrote:

Hello guys,so i might get somehing wrong but... I just finished a map with full combo, map that was weighted 111 pp and in my top ranks list i see that map on my #1 place, and it's weighted 100% (111 pp). But i didn't got 111 pp to my score, i got something like 30 or maybe little bit mor. So what i got wrong about PP system?
Weightage system

For transparency on how your final pp is calculated, pp gained from the beatmap is given along with weight (% of eligible pp to be gained for final pp). Only your highest pp score's beatmap gives you full pp. Other scores' pp values will be multiplied with decreasing percentage. The percentage is always rounded to the next whole number for displaying, but for calculations it is not. For example a score can be weighted by 0.48% despite the display saying "0%". If n is the amount of scores giving more pp than a given score, then the score's weight is 0.95^n .
It follows, that your total pp is computed as follows. Let PP contain each score's pp value. PP[i] denotes the ith score's pp value, sorted decreasingly, where i goes from 1 to n, and n is the amount of scores you have.
Total pp = PP[1] * 0.95^0 + PP[2] * 0.95^1 + PP[3] * 0.95^2 + ... + PP[n] * 0.95^(n-1)
Click for an example of the formula usage (scroll down to see the example).
There is also bonus pp based on the number of ranked maps you have a score on. The bonus is:
416.6667 (1- 0.9994^Number_of_scores).
laeamminlakana
I haven't seen anyone talk about patterns and "reading" pp so I'm gonna make a post real quick about them.
Please remember that I have very little information on the topic of pattern's pp values as I only read like 6 of the last pages and have vague memories of reading the first 30 or so, and they said that pattern recognition is a thing of the distant future...

So after thinking a lot about patterns that are more difficult than others I've come to some conclusions and I'll post them here.
First, overlaps are increasingly hard to read/play as AR and CS get lower
Second, if fast (160+ 1/4 bpm) snapping between 70° and 180° hasn't been given any proper value yet, it deserves some.
Third, jumps Are hardest at 180°, 90° and 0° there should be a small increase in value for each of these, linearly falling to the default value where the halfway points(i.e. 45° and 135°). I guess the value increase should also not be too large, like at max 1,1 times higher aim value for jumps like these
Fourth, Jump sets where combo changes deserve some extra value, as I personally, and most likely a lot of other people too, rely on followpoints when reading jumps. (Granted, maps where combo changes aren't logical shouldn't make it to ranking, thus making this nigh irrelevant on the topic of pp)

Just typing out some of my thoughts on map difficulty values, I'm sure I missed a ton of points I've thought of during my personal performance evaluation system planning.
Kert

Brimroth wrote:

...
t/331848
dung eater

Brimroth wrote:

...
p/3553419 there's been a lot of talk, but nothing done about it

if there was an easy way (easy to find, stuffed to your face) to get all the circle/slider positions and timings for several maps and just write an algorhitm to calculate stuff i think people would experiment with pattern and other difficulty calculation stuff more
death_bestow
I really hate how I can beat my score and get a lower PP value than the lower score. How and why is this possible, and can it be changed to take your highest pp attempt rather than your highest score?
Granger
It can and theres work being done for this; in the future you'll have multiple "ranked" plays for each mod but only your highest PP one contributes to your PP total. (if i remember correctly)
xJFx
my only complain is that I've been correcting some songs that I had in C because I couldnt play them properly at that moment, but it has happened that I get a better rank and accuracy in the song and my global accuracy drops...I dont get it...
Endaris
The accuracy on your profile is weighted by the accuracy of your top plays.
The worse a play is the less influence it has.
Therefore a C without any combo and trash acc has no influence on your weighted accuracy but a B with mediocre acc and high combo does have influence and will therefore lower your weighted accuracy.
julchiar
double time weighting


I believe the reason it gives more pp than it should (difficulty-wise) is because it gives a bonus for higher OD despite not increasing OD at all.

Higher OD decreases the time window in which you need to hit to score a 300. It therefor increases the time window between those 300 hit windows which is where the difficulty/accuracy challenge comes from.
This is not the case for double time. Double time merely decreases the time window required for a 300 but evenly decreases the time window between hits (instead of increasing it).

Double time makes you need to play faster, not more accurate than no-mod at all.
E m i
Is that also why stream maps are easier to acc than singletap maps?
Is it the reason why the fastest players in the game don't play 175-200bpm HR and 210-240bpm DT maps anymore?
dung eater
pp bonus for more time interval between hits (lower bpm, pauses) would be nice, because it's tuff
DT-sama

julchiar wrote:

double time weighting


I believe the reason it gives more pp than it should (difficulty-wise) is because it gives a bonus for higher OD despite not increasing OD at all.

Higher OD decreases the time window in which you need to hit to score a 300. It therefor increases the time window between those 300 hit windows which is where the difficulty/accuracy challenge comes from.
This is not the case for double time. Double time merely decreases the time window required for a 300 but evenly decreases the time window between hits (instead of increasing it).

Double time makes you need to play faster, not more accurate than no-mod at all.
To keep the same relative accuracy with DT, you need to have a higher absolute accuracy.
While it's true that relative accuracy is underratednot rated at all (140bpm OD10 is way harder to acc than 180bpm OD10 but they count the same), it's not like DT accuracy is a piece of cake, the absolute increase in accuracy is still there.

The relative vs absolute problem comes up in reading too, with density. DT keeps the relative reading difficulty (density) the same, but the absolute reading diffficulty (same density, higher AR) is higher.
Senko-san
More of a question than feedback, but regardless.

Does Hidden mod by itself affect the PP gained from the map or does it only increase score? Asking Tillerino while playing with both nomod and hidden seem to result in it guessing the same PP
Gigo
Of course it affects pp, even if it's by a small amount.
Reyvateil
Hidden gives a fixed amount to the aim value and accuracy bonus of a score. If I'm not mistaken it's an 18% boost to aim and a 2% boost to accuracy, so it's best used in jump/small circle maps with high OD.
WeabooTrash
I feel alive
Noodle Knocker
The lack of PP you get for doing a HD play. I think a couple years back HD was actually rewarding and now it's not. Basically you have no incentive to learn HD anymore because the lack of performance points you get.
GhostFrog
The way things are currently, HD is very rewarding if you learn to play it well and likely to be harmful if you don't play it well and just use it on random plays. It gives a 20% boost to aim pp, which can be huge, and a 2% (I think) boost to accuracy pp, which is basically nothing. If you can use it on aim-heavy maps without dropping accuracy for it, it's very rewarding. However, if you use it on less aim-heavy maps with high OD and lose accuracy for it, you very well might find yourself losing pp for using it. And don't even try to play hard-to-read, easy-to-aim maps with HD. Not worth. There are fewer of those after the recent HD change though.

It basically suffers from the same issue as flashlight. Making it feel rewarding for individual maps would make it insanely OP overall. It's a bigger issue for flashlight because being able to play HD on most maps isn't a problem for most people who try to learn it (everyone except me b/c i suck at reading). The system could definitely be better, but I don't know how tom would make this particular aspect of the system smoother within the current framework.
show more
Please sign in to reply.

New reply