forum

Performance Points feedback and suggestions (Standard)

posted
Total Posts
2,750
show more
Rewben2

Tess wrote:

Sure. I already admitted to not having been very clear about what I meant, and I do tend to get ahead of conversation sometimes, so that wasn't really great on my behalf. I also meant to imply that I take your opinion less seriously (not that it's less valid) because your lack of experience in said subject - meaning that I believe that, had you had more experience with it, you would've thought differently, which is enough of a basis to not take an opinion as seriously, to me. I don't entirely disregard it though - I did try to see your point, and I do see what you mean, I simply disagree.

You still didn't really answer my question directly, but it does turn out that what you think makes a map more or less difficult is the patterns used + the speed at which the patterns are played. Though this is true, OD and map length are still a part of it. For map length - your argument is that, if you play a 3-minute 5 star map twice in a row, that's the same as playing a 6-minute 5 star map. This couldn't be less true. First of all, the 6 minute map (provided that there is twice as much break time as in the 3-minute map) would be a lot harder to FC than it would be to FC the 3-minute map twice in a row. The combo is shorter, the amount of focus and stamina required is less and you don't have as much time to fuck up. Osu is about trying to get as close to SS'ing a map as you can, so that you can get the most points (be it score or pp) out of it. This means that both combo and accuracy are incredibly important, and it's also why they're weighed so heavily. Secondly, it would also be a lot harder to get good accuracy on, because the lack of breaks can make it easier to run out of stamina and start losing consistency. The longer a map is, the harder it is to SS it, this shouldn't be news. I think people should stop separating all aspects of difficulty and cherrypicking one of them instead of combining them all to form a more global idea of difficulty.

However, if you were to say that star difficulty is only an indicator of pattern difficulty, then I'm willing to see your point. I haven't heard of this though - "star difficulty" is known to be an indicator of a map's entire difficulty, not part of it. If it's not going to be a global indicator (like most people think it is), then perhaps it should be renamed to something like "pattern difficulty" or whatever else works. Right now it's just confusing, I know of a lot of people who don't understand that a 4.8 star map gave them more pp than a 5.3 star map. If it was depicted as only a part of the map's difficulty (just like OD, CS, etc.) then perhaps it would be easier for people to form a global idea of the map's difficulty for themselves.
I've said this before - the map length idea is supported by the map not being any harder at any given time. It's just longer, not harder. That's how I see it. Sure, it's harder to SS/fc/etc. but, as I've stated above, that's not because the maps hard. It's because it's long. As your last paragraph says, the pattern difficulty is a separate idea and when I think of how hard a map is, I really only think about it in that sense.

For simplicity, I don't disagree with star difficulty including od, ar and map length because it is mostly used as an indicator of pp and should probably reflect this.
Genki1000
I think map length makes a map more difficult in regards to stamina though, especially on maps which don't have breaks/slow parts. I mean, you wouldn't say a 40 note stream is just as difficult as a 20 note stream but just longer, right? (not sure if that example applies here)
silmarilen
including ar and od in star rating is a bad idea imo.
star rating is supposed to indicate how hard a map is to fc, not how much pp it gives. od is completely unrelated to how hard it is to fc, and ar is only a mental thing. map length should be included tho, because it makes the map harder to fc.
if you start including od into star rating you can get a super easy map with od9 having the same star rating as a super hard map with od6 (say https://osu.ppy.sh/s/89810 and https://osu.ppy.sh/s/39368) while the second one is way way more difficult to fc.
this would totally screw with how high star rating you can play.
B1rd

silmarilen wrote:

including ar and od in star rating is a bad idea imo.
star rating is supposed to indicate how hard a map is to fc, not how much pp it gives. od is completely unrelated to how hard it is to fc, and ar is only a mental thing. map length should be included tho, because it makes the map harder to fc.
if you start including od into star rating you can get a super easy map with od9 having the same star rating as a super hard map with od6 (say https://osu.ppy.sh/s/89810 and https://osu.ppy.sh/s/39368) while the second one is way way more difficult to fc.
this would totally screw with how high star rating you can play.
I don't think length should be included in star rating, you can't calculate it properly. People can figure out on their own how difficulty a song will be to FC based on its star rating and length.
AR isn't just mental, it legitimately makes songs harder to read
Vuelo Eluko
OD does affect how hard a map is to FC. a 50 on od8 could be a miss on od9, etc.
silmarilen
if you miss because of hitting too early/late then you first need to learn accuracy before you complain about a map being hard to fc. star rating assumes you can acc.
Nyxa

silmarilen wrote:

including ar and od in star rating is a bad idea imo.
star rating is supposed to indicate how hard a map is to fc, not how much pp it gives. od is completely unrelated to how hard it is to fc, and ar is only a mental thing. map length should be included tho, because it makes the map harder to fc.
if you start including od into star rating you can get a super easy map with od9 having the same star rating as a super hard map with od6 (say https://osu.ppy.sh/s/89810 and https://osu.ppy.sh/s/39368) while the second one is way way more difficult to fc.
this would totally screw with how high star rating you can play.
This is how I started feeling after GhostFrog made his point about the OD. I do wish that there was a way to name star rating that indicated its meaning better. Also, I do think that AR should be counted into star rating, but in reverse. AR9+ doesn't affect the rating, and the lower the AR gets, the higher the rating gets, though it shouldn't be too large a number (Like, if AR9 were 5 stars, AR4 on the same map would be 5.15 stars or something). Higher AR means that you need more focus to read the map consistently and thus makes it harder to FC.

Still think that the star rating should be called "Focus rating" or "Stamina rating" or something, to prevent confusion, but I'm sure that not many will agree with that.
silmarilen
it's called star rating, what it actually indicates is up for debate, but i think right now it tries to indicate how hard it would be to fc a map.
ofcourse there are maps that are too hard/easy to fc for their star rating but those are a small number.
Vuelo Eluko

silmarilen wrote:

if you miss because of hitting too early/late then you first need to learn accuracy before you complain about a map being hard to fc. star rating assumes you can acc.
a 50 is a less common type of accuracy mistake and usually a result of an aiming mistake rather than a rhythm error, so having less room to work with in that regard DOES make it harder to FC. may as well take star rating out entirely if its just going to assume you can combo everything.
Drezi
Instead of star diff a separate aim diff and speed diff would be better I think. Acc value is OD+number of circles, it could get a separate combined acc diff indicator too, and star diff could be a fourth one, a simple SS-pp/40 value used for the actual star numbers on the map list aswell. Maps could be sorted by aim diff, speed diff, acc diff and overall pp potential (the new stardiff).

It would be more straightforward at least, I think.
Nyxa
Having four different difficulty indicators might be a bit much, but adding an Aim and Speed value would most certainly help a lot. Maybe replace "Stars" with "Level", and remove the decimal so it looks like a whole number in the hundreds. So, a 5.32 star map would be a level 532 map, or something. But if you're already showing aim and speed difficulty separately, you may as well include OD in the map's level so that you can see each individual aspect of difficulty, + the total average difficulty for the entire map (including OD and map length).

Basically, how hard it is to SS a map. Osu is about SSing maps after all, even though most of us (that aren't like this guy) don't focus on getting an SS all the time because that is so damn hard on most Insane+ maps. So, instead, we try to get as close to an SS as possible. I don't think it would be all that weird for the difficulty indicator to talk about how hard a map is to SS rather than to FC. In fact, doing it the other way around sounds illogical to me.
Topic Starter
Tom94
To clarify: Right now the star rating indicates the difficulty of the hardest patterns in the map in their respective categories (speed and aim) in a way similar to how best performances of a player are weighted. The hardest pattern weights the most, the second hardest less and so on. Of course the "pattern" thing is just an abstract way of looking at it.

Dry details
In reality the map is divided into sections of equal length (between 250ms and 1s, I don't remember exactly) and for each section the highest difficulty of the hitobjects within is chosen. These sections are then sorted decreasingly by their difficulty and a similar weighting algorithm as the one used for player's best performances is used on these.

A consequence of this system is that map length is considered in star rating only up to a certain amount of hard patterns. This leads to the nice property that when taking a part of a map, then the star difficulty of that part is always smaller or equal to the difficulty of the entire map, which makes sense intuitively. The star difficulty thus focuses on the hard parts of a map and shall be interpreted as upper physical difficulty to FC. Note the physical - OD and AR are not taken into account. Keep in mind that this doesn't mean, that accuracy is completely unaffected by star rating. It is inherently harder to get a good accuracy on physically harder maps since you can't entirely focus your body on doing so.

It seems to me that it's best to leave AR, OD and map length as external parameters since this way it is easier for people to judge which maps they can play and which they can't play. If everything was condensed into a single number players would have to try out maps a lot more to see whether the difficulty is of the kind they want to have. The next step is to show aim and speed difficulty seperately as well, to further help with that notion and potentially even make a radar chart showing up in the game.
haha5957
it has been forever since scores that yield "lower pp" are replacing the higher onces. There are some issues that are pretty complicated to fix, but this is relatively easy fix and i have no idea why this isnt getting fixed..

it is extremely frustrating when i know i can DT a map but it has reasonably high accuracy HDHR and i can't make pp out of it.

same refers to FL, and in very rare case, HT.


this also happenes even without mods. I exactly don't know why but



FYI, both of those were yielding 113 pp for me when i had lower accuracy and lower combo. this might not be bug since i have more x and more 50 but still, if it is less performed, the scores shouldn't have been replaced.
mcdoomfrag

haha5957 wrote:

FYI, both of those were yielding 113 pp for me when i had lower accuracy and lower combo. this might not be bug since i have more x and more 50 but still, if it is less performed, the scores shouldn't have been replaced.
The reason this happens is because the PP system is not the same as the official score system.

haha5957 wrote:

but this is relatively easy fix and i have no idea why this isnt getting fixed..
Because it's not that easy. For a fix like this to happen, Tom and peppy would have to work hand in hand to restructure the way scores are saved. Example, what happens if Tom changes the PP formula, and suddenly the scores you had before are now worth more? It's a complicated issue, which would require saving more than 1 score for each player on all maps.
Drezi
Any opinions on the viability of this Tom?

GhostFrog wrote:

How difficult a map is rhythmically isn't taken into consideration at all right now. Doubles are treated the same as any other notes and contribute to the strain values in the same way any other notes would based on their position and timing.

Full Tablet wrote:

Drezi wrote:

That's a shame, when it comes to rhythm the less repetitive it is, the harder.

I mean it's like anyone can hit a constant beat on a drum, but even a repeating pattern is harder to pull off..
Something like the algorithm here in tom94's ask.fm could be used http://pastebin.com/cFGUJdGa

It is for taiko, but could be used for standard too if the only variable of the objects is the time between hits, with only one color present, considering both circles and slider starts as the same kind of object. Sliders might be considered a little different (probably making sliders of a certain duration have a "partial" match with circles or sliders of different duration if both share the same time between key presses, where the partial match reduces the rhythm complexity strain less than a full match).

Using a weighting of the strains of 0.9975 (So the maximum value is 400):

"Rhythm Complexity"
xi - FREEDOM DiVE [FOUR DIMENSIONS]: 348.488
https://osu.ppy.sh/b/297463&m=0 351.973
https://osu.ppy.sh/b/312959&m=1 324.277
https://osu.ppy.sh/b/443272&m=0 271.207
https://osu.ppy.sh/b/323875&m=0 256.527
https://osu.ppy.sh/b/152078&m=1 369.495
https://osu.ppy.sh/b/58063&m=0 328.276
It really is much harder to acc a map with mixed doubles singles triplets etc, than a full 1/2 map like Setting Sail, Pony maps and some others, it's no wonder everyone around my rank has those in their top ranks - they are easy to acc, and rewarded just as high as maps with harder rhythmic patterns. It's part of the reason why DT feels more rewarding - easier maps sped up are usually simpler rhythm wise aswell, thus easier to tap into their potential acc pp.

Even if the proposed algorhythm by Full Tablat for weighting acc values is imperfect or flawed, it is not a problem, since we're not breaking anything here - on the contrary, we'd be making progress towards reflecting actual difficulty. Not having rhythmic complexity being accounted for is a flawed state in itself, having at least SOME maps being evaluated better, having at least some degree of differentiation is preferable I think.
Nyxa
I think that the reason people feel that DT is overrated is this rhythm complexity issue, and not DT itself. Koigokoro, Setting Sail, Raise This Barn, Apples to the Core, etc. - these maps aren't /easy/ to FC, but they really aren't harder to get high accuracy on than, say, https://osu.ppy.sh/b/297463?m=0 this map, which gives less than 216pp for an SS while Koigokoro gives 300pp for that. Now, I understand that Koigokoro has the hard jumps and all that, but is it really that so many people are so talented at DT jumps that all of them have a good Koigokoro score?

I think that if rhythm complexity were taken into account (which is something that should've happened long ago, this is a rhythm game after all) that a lot of maps would be evened out in terms of pp and there'd be much less complaints about over or underrated maps.
haha5957

mcdoomfrag wrote:

haha5957 wrote:

FYI, both of those were yielding 113 pp for me when i had lower accuracy and lower combo. this might not be bug since i have more x and more 50 but still, if it is less performed, the scores shouldn't have been replaced.
The reason this happens is because the PP system is not the same as the official score system.

haha5957 wrote:

but this is relatively easy fix and i have no idea why this isnt getting fixed..
Because it's not that easy. For a fix like this to happen, Tom and peppy would have to work hand in hand to restructure the way scores are saved. Example, what happens if Tom changes the PP formula, and suddenly the scores you had before are now worth more? It's a complicated issue, which would require saving more than 1 score for each player on all maps.


At least i'm pretty sure that FL and HDHR won't worth more than DT with same accuracy in any future timeline
and they obviously do not have to change the way to save the score since they already have the current pp earned from the song, and all they have to do is to compare your achieved score to previous best score, not with the scorepoint, but with pp.
Nyxa
I think you're portraying the solution to this problem as something much simpler than it could possibly be.
haha5957
Can this guys read and think or something? dear lord, Start thinking seperately.

1) Best scores not being the best performance achieved IS a problem, and it's coming from using old score system to determine the best score. This makes no sense and obviously decreases the true meaning of pp rank system that tries to measure true stkill and not very-combo centered, but fortunatly is extremely easy fix. Even if the algorithm changes in future, I'm very sure that the current pp system will look much alike the algorithm of future, not the score system.

A fix not being "perfect" does not mean it is not "better" than before. Do i really have to explain this kind of stuff? Being a no no man just because it is not perfect seems extremely immature and does not make any sense to me.

2) Now, the star diff being little off from what it really feels like, especially with high SV maps and complex rhythm maps, comes from different issue, most likely being pp not taking rhythm complexity and slider paths(I guess). This is where you start talking about how algorithm can get better to tell the true difficulty and is relatively hard fix compared to the first one.
Keeby
Has the idea of keeping ARs separate been considered? (in terms of pp). I think it would be interesting. Perhaps it takes less skill to master lower ARs, but I think a better osu! player should be able to play a wide range of ARs. I honestly can't do AR8.5 or lower without some practice. Same with AR10...

I'm only saying this because AR8 is basically outdated now. I think it's another form of reading, lol. Lower AR seems discouraged with high star difficulty (all the AR8s are pretty much 3 stars, but the AR8s that have high star difficulty are crazy) and we haven't generally accepted many AR10s yet, but I'm expecting there to be more in the future. (In which case, there is a huge difference between reading AR9 and 10).
Drezi

haha5957 wrote:

1) Best scores not being the best performance achieved IS a problem, and it's coming from using old score system to determine the best score. This makes no sense and obviously decreases the true meaning of pp rank system that tries to measure true stkill and not very-combo centered, but fortunatly is extremely easy fix.
2) Now, the star diff being little off from what it really feels like, especially with high SV maps and complex rhythm maps, comes from different issue, most likely being pp not taking rhythm complexity and slider paths(I guess).
The reason the first issue hasn't been solved yet is because plain logic doesn't always apply to programming challenges. It might seem like a trivial task on paper, but there might be several issues stopping them from fixing it easily. You can be sure that if it was all that simple Tom would have already fixed it that way (storing two performances, both highest score and highest pp, or the highest pp one only).

Right now what's in the works is storing one performance per mod combination, and that looks like a less efficient way, and doesn't solve the 'better accuracy/sliderbreak plays being overwrited' issue either, still I doubt he'd be working on that if a simpler and better alternative was feasible.

The second issue has many aspects to it, rhythm complexity being one, and we're discussing possible improvements in that regard right now.
PepsiCat
littlederp
koromo

PepsiCat wrote:

why did my pp just disappeared i don't get it and my rank got up of no reason. the thing is like when i checked at osu.ppy.sh its i have 2pp but when i start osu! it says i have 215pp. Can someone please explain why this happend.
You were looking at your taiko stats.
EroPudding
Can someone explain to me how the PP system work. like i did this song http://puu.sh/ctlUn/00c27abbfe.jpg and i didn't get any PP and the song didn't even get in my top ranks as it should have because its a 4.8 stars song. am i missing out on someting important here or. i'm a bit confused ATM
Nyxa

MisakaCandy wrote:

Can someone explain to me how the PP system work.
t/181850
https://osu.ppy.sh/wiki/Performance_Points
Full Tablet

MisakaCandy wrote:

Can someone explain to me how the PP system work. like i did this song http://puu.sh/ctlUn/00c27abbfe.jpg and i didn't get any PP and the song didn't even get in my top ranks as it should have because its a 4.8 stars song. am i missing out on someting important here or. i'm a bit confused ATM
That play seems to be pretty average for someone in your rank. If that map had higher OD and you had that same accuracy percentage or higher, then you would have got more pp (most easy DT maps that have a star rating similar to that map are OD7, so they have timing windows similar to OD9 no-mod).
Nyxa
You also shouldn't really associate stars with the amount of pp gained. If you read back 15 pages there should be a post of mine complaining about star ratings not reflecting pp gain at all, and then Tom explaining that it's to do with the upper physical difficulty of the map - excluding shit like AR and OD. Really, just read the thread before asking questions.
silmarilen

MisakaCandy wrote:

Can someone explain to me how the PP system work. like i did this song http://puu.sh/ctlUn/00c27abbfe.jpg and i didn't get any PP and the song didn't even get in my top ranks as it should have because its a 4.8 stars song. am i missing out on someting important here or. i'm a bit confused ATM


excuse me?
Nyxa
So what's happening to all the rhythm complexity issues and suggestions? Are they just gonna be ignored?
Fanty
So I just improved one of my top scores, the total pp it's worth went up from 108 to 115 (Sakuzyo - Neurotoxin on my userpage), I improved the accuracy and combo. But somehow I ended up losing 14 of my total pp. I'd like to know how. I didn't use any mods and yes this is standard osu!.
silmarilen
most likely a qualified map you played got unranked
Fanty
"Got unranked"? 0_o
Vuelo Eluko

Fantyful wrote:

"Got unranked"? 0_o
Got unranked.
Drezi
But this isn't a "why did I lose 4 pp, please halp" thread, it's feedback and suggestions..
Fraudo
I made a thread here on the topic of complexity and the pp and star rating systems.

Basically, I talked over some issues that have to deal with what makes a map hard aside from the physical aspects like speed and aim, and would appreciate any discussion and feedback over in that thread. Just posting here for visibility and so that it can be noted as a feedback/suggestion related to the current system.
Woobowiz
While Accuracy is important for the core of osu, I feel like some maps are severely underrated because of that importance, namely No-mod and anything between OD 7 and 8. The gap in pp between 99% and 100% is tremendous for a lot of maps. The gap between 98% and 99% is not as huge, but it's a stupid large gap nonetheless.

Examples include :


and many more...

There aren't any good no-mod examples because they all give underwhelming amounts of pp anyways simply because of their lower OD or they're OD 9 or really difficult maps.

Also~ I feel like FC's on maps that are 230+ bpm get underrated too, because speed and aim only really scale linearly while accuracy apparently scales non-linearly.
Laacis
Hi,
this is my first time posting in osu forums so don' t get mad If I posted in the wrong section.

My issue is that I got more pp for better accuracy but no FC and lost pp when I got FC and about -0.5% on the accuracy https://osu.ppy.sh/ss/2249414
The original pp that I got for the S rank with the 354 combo was about 74 and I got about 68 for the one with FC and 98,85 accuracy.

Is the accuracy really accounted that much in pp that slider breaks in the middle don't change the pp? Or is it supposed to be like that because slider breaks don't count as misses?
GhostFrog
Well, there are competing forces at work here...

-Your aim and speed pp for a map get multiplied by (your combo/max combo)^.8. In this case, that meant you got just about 80% of the aim and speed pp for your 345 combo play.
-Accuracy pp increases faster the closer you get to 100%. When you're around 99%, this can be a big deal. Notice that 98.85 is nearly twice the accuracy loss (well, one and two thirds times) as 99.31%.

In this case, with the map being 3.57stars and effectively OD8.33 with DT, the accuracy component of the pp is pretty noticeable. Apparently enough so that losing 20% of your aim and speed pp wasn't enough to compensate for the acc loss. So I guess the short answer is "yes, accuracy is in fact represented that much in pp (in this case, at least)." It's also worth noting that if you'd gotten a miss instead of a slider break it would have multiplied your aim and speed pp by .97, which still wouldn't have made your 345 combo play worth less, though the increased acc loss from a miss vs a 100 might have done it.



I would say this thread desperately needs an FAQ section to handle questions like this, but most people just post new threads instead of checking this one so I guess that wouldn't be so useful...
Laacis

GhostFrog wrote:

Well, there are competing forces at work here...

-Your aim and speed pp for a map get multiplied by (your combo/max combo)^.8. In this case, that meant you got just about 80% of the aim and speed pp for your 345 combo play.
-Accuracy pp increases faster the closer you get to 100%. When you're around 99%, this can be a big deal. Notice that 98.85 is nearly twice the accuracy loss (well, one and two thirds times) as 99.31%.

In this case, with the map being 3.57stars and effectively OD8.33 with DT, the accuracy component of the pp is pretty noticeable. Apparently enough so that losing 20% of your aim and speed pp wasn't enough to compensate for the acc loss. So I guess the short answer is "yes, accuracy is in fact represented that much in pp (in this case, at least)." It's also worth noting that if you'd gotten a miss instead of a slider break it would have multiplied your aim and speed pp by .97, which still wouldn't have made your 345 combo play worth less, though the increased acc loss from a miss vs a 100 might have done it.



I would say this thread desperately needs an FAQ section to handle questions like this, but most people just post new threads instead of checking this one so I guess that wouldn't be so useful...
Thanks for the quick reply, but still 120 less combo in a 460 combo map is really significant. Well anyways guess I have to get my accuracy up.
Thanks again.
Ace3DF
I don't get how (for example) a song with 1000 objects; 990x combo (slider brake) with 40 100s (no misses or 50s) score would get (for example) 200pp while 2 500x combos (slider brake midway) with only 2 100x and no misses/50s get you 150pp.
mad
show more
Please sign in to reply.

New reply