forum

Performance Points feedback and suggestions (Standard)

posted
Total Posts
2,750
show more
jesse1412

NixXSkate wrote:

Wall
Sorry if it seemed like I was saying "DT should be more rewarded than FL", that wasn't my intention. I mean that if I put equal amounts of effort into something that displays raw skill (where the amount of effort isn't seen as I do not have FL on) as I do into something that displays commitment only (where the effort is obviously displayed via the fact that FL is used) then why should the effort of one be rewarded while the other isn't. I COMPLETELY understand that FL can be done in a manor that involves skill rather than mindlessly replaying a map until you learn it, but the issue is that you do not need to posses the skill nor can you prove that you have it when you play fl. I can sit and mindlessly retry a map with FL for days and set a score that someone else can produce in hours; it should NOT be assumed that I have the skill to set FL scores efficiently unless the skill is 100% displayed. Considering that the skills required for efficient FL plays are unmeasured currently (as someone could literally brute force their way through the map), the pp for "possibly" displaying those skills should not be rewarded.

Also if you think that a 50% aim bonus PLUS an ADDITIONAL bonus for map length (beyond the already overly strong one) isn't enough for playing FL then we've met a MASSIVE disagreement.

It also seems that you believe that I think FL will become "farmable", this isn't the case. I think that FL would become a way for people to obtain scores with pp above their physical play ability by no braining a map that's comfortably within their limits. pp reflects PERFORMANCE; if you're not PERFORMING well enough then you shouldn't be rewarded, putting a lot of time into a single map is an AWFUL way to get pp in the long term but it most certainly could allow for people to "perform" above their level in the short term and skew their rank if FL was buffed.

Nobody FLs koigokoro because: It's BORING AS FUCK.
It takes FUCKING ages.
They could do 10 other pp scores in the time it takes them to FL it.

The map is quite clearly not hard as hell as an immense amount of people have FC'd it. You talk about all these "fast" maps that need buffing but I've yet to see one example of a score. Dungeons 5.3 star DT score isn't a "fast" score in the slightest, it pales in comparison to scores like rrtyuis Neuronecia score (which is genuinely quite tough for FL but sadly underrated nomod).

I think the best resolution here would be to lower the 50% aim boost that FL currently gets and increase a more generic aspect of it to try and roughly balance it out while buffing gneral FL plays, this way non aim intensive scores would receive a little bit more from FL (i.e, every FL score I can think of) and the aim intensive scores (when they exist) will still have that massive juicy pp pool that people have yet to discover.
NixXSkate

jesus1412 wrote:

The map is quite clearly not hard as hell as an immense amount of people have FC'd it. You talk about all these "fast" maps that need buffing but I've yet to see one example of a score. Dungeons 5.3 star DT score isn't a "fast" score in the slightest, it pales in comparison to scores like rrtyuis Neuronecia score (which is genuinely quite tough for FL but sadly underrated nomod).

I think the best resolution here would be to lower the 50% aim boost that FL currently gets and increase a more generic aspect of it to try and roughly balance it out while buffing gneral FL plays, this way non aim intensive scores would receive a little bit more from FL (i.e, every FL score I can think of) and the aim intensive scores (when they exist) will still have that massive juicy pp pool that people have yet to discover.
It's hard to give an example of a fast FL score because they're such a rarity (for a reason). A player can't get a score with FL unless it's really below their max ability. That doesn't mean it's "comfortably in their limits" though, because FL, like other mods, is making it harder for the player, even if the map doesn't physically change. If they can get a score that's close to their max ability, then they're a talented FL player that deserves reward. Of course rrtyui would get the best FL score if he tried, he's the best player playing a map way below his level that's above most others. It's also on a complicated slider filled map which makes it low pp and harder to FL. The awkward mapstyle itself is one of his known specialties as well. We can also add that he would have probably gotten more pp if he just got a score with only HD instead despite it being the "best FL score", but whatever (don't know how much pp his score gives, but I'm assuming based on what I've seen). Anyway, Dungeon's score may be slow for say, an HD+DT score, but with DT+FL it's a totally different story, just look at how hard it is to follow after 230 combo until the end because of its speed. Sure faster bpm FL scores exist, such as Mesita's score on this, a sloppy but nice score, but it's an easy insane, especially in comparison to World End. Despite the mapping being very vanilla, you can tell how hard it is to follow at certain parts, especially around the climax, because of its high bpm. Something like Suklaapallit may be more aim intensive (without FL) and have a higher star rating, but I wouldn't consider it as hard to FC with DT+FL as World End because of the challenge the speed difference adds to following it with FL. I'm focusing on what makes an FL score harder to execute, not harder to memorize, here. Overall I think we somewhat agree that the harder aim maps should be where the pp is coming from with FL, but the problem I have is that what's easier and harder to aim with FL is different than how aim pp is determined, which makes it complicated. You think FL shouldn't reward more for aim, and I think it's wrong for aim pp to be the only factor involved in this calculation.

jesus1412 wrote:

It also seems that you believe that I think FL will become "farmable", this isn't the case. I think that FL would become a way for people to obtain scores with pp above their physical play ability by no braining a map that's comfortably within their limits. pp reflects PERFORMANCE; if you're not PERFORMING well enough then you shouldn't be rewarded, putting a lot of time into a single map is an AWFUL way to get pp in the long term but it most certainly could allow for people to "perform" above their level in the short term and skew their rank if FL was buffed.
As it stands now, there is no FL player (no matter how pro they are with FL) with top ranks full of FL scores, unless they're holding themselves back from getting higher pp scores from non-FL scores or just don't care about pp, so I don't know why you're worrying. I was trying to make it clear that it's only fast aim intensive maps that need to reward more, aka the ones that aren't farmed, barely farmable by skilled players, and barely even exist. Basically its bonus being not just aim, but factors like, the bpm/the amount of time between each jump, the chaos, etc. (It would probably be difficult to do so it probably wouldn't happen, though.) As it is now, I feel that jumpy maps around ~180bpm or spacey streamy maps are benefiting most from the current system since they're generally at a speed where you can process what you're doing with FL well while still having a map with some aim pp in it.

Also this might sound weird and sacrilegious, but I think AR11 bonuses could be halved if FL was used since the map has to be memorized.
-GN

NixXSkate wrote:

this might sound weird and sacrilegious, but I think AR11 bonuses could be halved if FL was used since the map has to be memorized.
sounds a bit weird to me. playing AR11, even with FL, really cuts down the little time you have for adjusting your aim and making sure you hit the next circle, which makes consistency extremely hard to gain. it probably doesn't matter too much on maps with slower fullscreen jumps, but no one has done anything like that, so eh...

i think it balances out as a whole.
Woobowiz
This may be a stupid suggestion, but is it within reason to "buff" lower OD by weighing the accuracy values of 100's and 50's?

So for OD 6, 100's would be worth 2/5 or 1/2 of a 300 rather than the usual 1/3rd of a 300. Low OD is weak enough for giving less pp even for an SS, why not give it a little push
Vuelo Eluko
the issue with low OD isn't in 100's/50's.. it's in that its low OD. The map is at fault, you can SS wi-fi bridge which is pretty impressive since its 5.55 star map, but it's only worth 190pp because mappers are scared to use high od, most old hard modless maps arent played anymore because od8 is worth crap for the effort at that difficulty.

for this reason i think a better buff to low OD has always been to count unstable rate/error and consider the score to be worth more if it could have been high accuracy on higher OD.

too bad this would only retroactively impact scores on the top 50 with replay data, 99.9% of players get screwed by this.
Woobowiz
I'm a little iffy on the whole "Factor in Unstable Rate" thing, on the bright side it provides more maps to farm pp, on the other hand, players could FC easy maps with a really low unstable rate and get stupid amounts of pp for the difficulty of the maps they play.
Vuelo Eluko

Woobowiz wrote:

players could FC easy maps with a really low unstable rate and get stupid amounts of pp for the difficulty of the maps they play.
Yuudachi-kun
HR pp is related to the OD/CS and not UR?
Vuelo Eluko
point is people get ridiculous pp with it relative to the difficulty just by being accurate.. i mean i dont see any 80-90% HR fcs worth 300+pp unlike DT...

thats what PERFORMANCE is though

my suggestion just kinda makes nomod similar to hardrock just in terms of od not circle size difficulty modifiers..
Yuudachi-kun

Riince wrote:

point is people get ridiculous pp relative to the difficulty just by being accurate.. i mean i dont see any 80-90% HR fcs worth 300+pp unlike DT...
The people who play HR are also the ones who have good acc in the first place.

e: Can you even pass a HR song with that acc?
Vuelo Eluko
what is your point? i was just letting woob know that "getting ridiculous pp for the difficulty of the map" is already a thing that happens, oftentimes with DT too, and that it shouldn't be a bad thing to take nomod along for the ride so that people actually start playing it again, not let it sit and rot because mappers all made them too hard for mods but too low od to be rewarding in a system that came out much later.

Kheldragar wrote:

e: Can you even pass a HR song with that acc?
depends on the map, doesnt even matter about the HP if its 9.8 or 10, drain is mostly determined by how many long spinners or long low density drain parts there are, a map can be hp10 but in reality you lose hp as if its hp6.

you can pass catastrophe with a C with hard rock and thats 9.8.
silmarilen

Woobowiz wrote:

I'm a little iffy on the whole "Factor in Unstable Rate" thing, on the bright side it provides more maps to farm pp, on the other hand, players could FC easy maps with a really low unstable rate and get stupid amounts of pp for the difficulty of the maps they play.
getting <100ur on a 2star map is near impossible, give it a try if you want.
B1rd
That is a terrible idea, UR should never be factored into pp. I don't mind the thought of low OD getting a small buff, but the reality is the current game meta is high accuracy and not amazing scores (rrtyui), traditional nomod isn't going to be the meta as long as this is the case. But you won't see much od8 in the future, soon nomod will be much more like mods, hard maps will probably be mostly ar/od9.x.

I don't really care that much about OD, FL etc., what really needs buffing is sliders.
Barusamikosu
I don't really like the idea of factoring in UR either. I think nomod should be made more appealing somehow. For example, more OD9 maps in the 4-5 star range that your average 4/5 digit player could farm. (Though it doesn't really solve the current low OD issue with existing maps)
E m i
simple
separate aim, speed and accuracy
jesse1412

B1rd wrote:

That is a terrible idea, UR should never be factored into pp. I don't mind the thought of low OD getting a small buff, but the reality is the current game meta is high accuracy and not amazing scores (rrtyui), traditional nomod isn't going to be the meta as long as this is the case. But you won't see much od8 in the future, soon nomod will be much more like mods, hard maps will probably be mostly ar/od9.x.

I don't really care that much about OD, FL etc., what really needs buffing is sliders.
A lot of people say this, got any example maps?
Karuta-_old_1

jesus1412 wrote:

A lot of people say this, got any example maps?
There is probably a whole bunch of them but here is one
https://osu.ppy.sh/s/148092
303 pp for first place with only 3 hdhr player

but I am staying neutral since there is so many kinds of beatmaps out there
uberpancake

jesus1412 wrote:

B1rd wrote:

That is a terrible idea, UR should never be factored into pp. I don't mind the thought of low OD getting a small buff, but the reality is the current game meta is high accuracy and not amazing scores (rrtyui), traditional nomod isn't going to be the meta as long as this is the case. But you won't see much od8 in the future, soon nomod will be much more like mods, hard maps will probably be mostly ar/od9.x.

I don't really care that much about OD, FL etc., what really needs buffing is sliders.
A lot of people say this, got any example maps?
Maybe maps like scary rose, big black, adult's toy and talent shredder? I'm sure there are better examples though
jesse1412

uberpancake wrote:

jesus1412 wrote:

A lot of people say this, got any example maps?
Maybe maps like scary rose, big black, adult's toy and talent shredder? I'm sure there are better examples though
All underrated because of the 1key sections, not the sliders afaik.
Vuelo Eluko
of those examples, the only ones with streams are big black and adult's toy and they are extremely easy streams, so the fact that most of the maps difficulty comes from 1key parts is a big factor, i think jesse is right

well except maybe towards the end of bb, but id say that map is still 6.5 stars so..
Woobowiz
Re-posting my original suggestion because people keep seeing the UR thing Riince talk about after this.

Woobowiz wrote:

This may be a stupid suggestion, but is it within reason to "buff" lower OD by weighing the accuracy values of 100's and 50's?

So for OD 6, 100's would be worth 2/5 or 1/2 of a 300 rather than the usual 1/3rd of a 300. Low OD is weak enough for giving less pp even for an SS, why not give it a little push
E m i
that makes no sense since it basically rewards poor performance (low OD low acc) and would make low OD high acc further underrated.
B1rd
No one replied because it was a stupid idea, don't just keep posting what you said until someone acknowledges it.

jesus1412 wrote:

uberpancake wrote:

Maybe maps like scary rose, big black, adult's toy and talent shredder? I'm sure there are better examples though
All underrated because of the 1key sections, not the sliders afaik.
What do you mean by '1key section'? I think it's safe to say that those maps would be a lot easier if the sliders were circles, I actually tried changing some sliders to circles on Scarlet Rose and the star rating actually increased.

As for examples, I think you mentioned Neuroncia as one, this would be another example. But I think it's already pretty evident that sliders are underrated, I don't think there is one slider map that gives good pp, every good pp map is almost all circles. They don't give anywhere enough pp for the extra aim and reading you have to do, and of course they give no accuracy pp which is a big deal.
jesse1412

B1rd wrote:

No one replied because it was a stupid idea, don't just keep posting what you said until someone acknowledges it.

jesus1412 wrote:

All underrated because of the 1key sections, not the sliders afaik.
What do you mean by '1key section'? I think it's safe to say that those maps would be a lot easier if the sliders were circles, I actually tried changing some sliders to circles on Scarlet Rose and the star rating actually increased.

As for examples, I think you mentioned Neuroncia as one, this would be another example. But I think it's already pretty evident that sliders are underrated, I don't think there is one slider map that gives good pp, every good pp map is almost all circles. They don't give anywhere enough pp for the extra aim and reading you have to do, and of course they give no accuracy pp which is a big deal.
1key sections as in... the entirety of the maps (if you 1key). Very spaced slow streams if that's how you prefer to see them.

There are only two ways that I can figure this out, one possibility is that jump distance is taken from the ends of slider, in this case, when a player doesn't follow a short slider (as in, they treat it like a normal hit circle) the effective jump distance is calculated with the wrong assumptions. The player presses the slider start and then moves to the next note/slider as if it was a hitcircle but the star algorithm assumes that they follow the slider to the end then treats the jump as the shorter distance between the slider end and the next note.

The other possibility that could possibly mitigate the above scenario is that the jump distance is taken from the point furthest from the end of a slider that would yield a 300. In this scenario the previously mentioned scenario can still occur but it should be dampened unless the sliders get especially long. This solution however has it's own issues, shorter sliders will be assumed to work as standard hitcircles.

The one thing I can't understand is how long slider maps are underrated, in my eyes it just doesn't make any sense and I'd have to guess that it's underrated for other reasons.

I'll admit this is an issue and that it's probably one of the more fixable ones at this point in time (compared to heavy 1key sections/slow spaced streams).
silmarilen
to give you a better example of underrated sliders: https://osu.ppy.sh/s/128645 or https://osu.ppy.sh/s/102282
i personally think pp doesnt look enough at technical difficulty, it's pretty much only physical ability. https://osu.ppy.sh/s/290040 are easily 5+ stars in terms of difficulty but only 4.32 stars because it's a slow map with pretty much no spacing
DeletedUser_4329079

silmarilen wrote:

to give you a better example of underrated sliders: https://osu.ppy.sh/s/128645 or https://osu.ppy.sh/s/102282
i personally think pp doesnt look enough at technical difficulty, it's pretty much only physical ability. https://osu.ppy.sh/s/290040 are easily 5+ stars in terms of difficulty but only 4.32 stars because it's a slow map with pretty much no spacing


lol I didn't know Terminal was so underrated
dung eater
Not sure if this has been suggested before.

Maps could analyzed for difficulty on a combo timeline to determine what combo guarantees you fcd the hard parts. Find the lowest difficulty section, that x combo passes for some amount of combos (25%, 50%, 75%, 90%, 100% combo for example).

You could make a simple table for every map to weight the pp/combo% they give and buff nonfcs on maps with hard parts in the middle, lower the pp nonfcs in maps with hard parts in the beginning/end give.
jesse1412

jaaakb wrote:

Not sure if this has been suggested before.

Maps could analyzed for difficulty on a combo timeline to determine what combo guarantees you fcd the hard parts. Find the lowest difficulty section, that x combo passes for some amount of combos (25%, 50%, 75%, 90%, 100% combo for example).

You could make a simple table for every map to weight the pp/combo% they give and buff nonfcs on maps with hard parts in the middle, lower the pp nonfcs in maps with hard parts in the beginning/end give.
Suggested but I still agree with the idea.
Endaris
factoring in UR doesnt sound reasonable
first of all you'd have to replace it with average aberration and then again:
If you can get a bonus to a 300 why wouldn't you give extra pp for hitting circles closer to the middle?
And if there's already a function to calculate boni why wouldnt we replace every circle with a target like this?
http://www.clipartbest.com/cliparts/pc5 ... e9pkxi.png
would kill any DT-players with non-perfect aim and move the focus to smaller circle sizes when ppl git gud at it
slight irony because it would make the game weird imo but didn't want to keep it for myself
silmarilen
UR isnt about where you hit the notes, it's about how accurate you hit them. it's a better version of accuracy
Drezi
I think he knows that, but it measures your absolute accuracy, whereas hitting notes somewhere within the hitarea does not, only whether it's a hit or not based on CS, while this aim aspect could also be measured in an absolute way (regardless of CS, just like UR doesn't care for OD) if we wanted to go down that direction: hit distance from 1x1 centre of hitcircle, averaged similarly to unstable rate (and similarly to hit timing, more than X aim error = miss, where X depends on CS/OD).

It seems fair just like basing acc on UR, but I don't think it'd "feel good" to play like this (even though it'd be good to have mods that allow players to push their "aim accuracy" too by allowing CS7 and stuff like that on ranked maps). Probably measuring absolute error instead of categories wouldn't feel that harsh when applied to accuracy, maybe because that's already divided into 4 outcomes vs hit/miss of aim and we already have OD10-11 that require you to be rather precise with your acc, while CS7< is pretty rare.
DeletedUser_4329079
UR weighting would be really unfair for those who have cheap setups, I don't think it's a good idea.
Yuudachi-kun
Introduce rainbow 300's to standard that give acc weightings of 150%.

huaehaueahu
Endaris

Drezi wrote:

I think he knows that
^
Additionally UR is defined as:

osu! Wiki wrote:

This value represents how consistently you time your hits, with lower numbers being better (top players often score below 150). Note that this measures consistency, and not accuracy, so if you're consistent in hitting 15ms early, you'll get similar results to if you're consistent in hitting on time. The formula is essentially the standard deviation of your hit errors (in milliseconds) multiplied by 10.
Which the part with "average aberration" referred to as UR doesnt measure accuracy as mentioned in the definition.
You'd need a different value that calculates the deviation to the perfect timing in a sensible way.

The part with the target thing was meant as a fitting analogy that shows how awkward it would feel if something like weighted accuracy outside of OD would be added to the pp-calculation. Because what it essentially would be is that you're able to modify the OD of the map at will(to exactly what you're capable of) and the map still counting for your pp afterwards.
Coming from that, you could claim that possibility for any of the metadata: O look, I can SS this map at Hp4 and Hp9, gimme more pp for the latter.

and

Default wrote:

UR weighting would be really unfair for those who have cheap setups, I don't think it's a good idea.
My personal UR went down by around 30 INSTANTLY just because i switched to good input-devices before I even got used to them. When cheap setups already have a harder time it'd be unfair to give them less reward for something they already have to put more effort in.

PS: I feel seriously trolled by scoring system:
DeletedUser_4329079
https://osu.ppy.sh/s/13768 this map has insanely hard sliders too and gives ~180pp (WWW is the only one who has an S on it) I'm unsure if it's possible to fix it with the current system though.
blahpy
pp: [Tom94] Reduce value of the Easy mod to counterbalance difficulty calculation bugfix.
May I ask, what exactly happened in this update? It sounds like Easy got nerfed, but neither me or any of my friends that have Easy scores in their top ranks seem to have lost any pp.

Itsudemoo
Fix fast sliders and fast streams not worth anything atm
Skyanide
This has probably already been mentioned (and is probably a bad idea for one reason or another), but oh well, prepare for opinions.

I think it's a bit odd that max combo% has such a large impact on the amount of PP that a non-FC play is worth; I have had multiple cases where a play with multiple misses at the end of the song is worth considerably more PP than a play with one miss in the middle of the song.

In my opinion, the number of misses should have a bigger impact on how much PP a non-FC play is worth, and combo% should have a much smaller impact (or no impact at all). To account for slider breaks, perhaps there could be a PP bonus for getting a full combo (or a penalty for not getting a full combo).

jaaakb wrote:

Not sure if this has been suggested before.

Maps could analyzed for difficulty on a combo timeline to determine what combo guarantees you fcd the hard parts. Find the lowest difficulty section, that x combo passes for some amount of combos (25%, 50%, 75%, 90%, 100% combo for example).

You could make a simple table for every map to weight the pp/combo% they give and buff nonfcs on maps with hard parts in the middle, lower the pp nonfcs in maps with hard parts in the beginning/end give.
I think this is a good idea too.
uberpancake

jaaakb wrote:

Not sure if this has been suggested before.

Maps could analyzed for difficulty on a combo timeline to determine what combo guarantees you fcd the hard parts. Find the lowest difficulty section, that x combo passes for some amount of combos (25%, 50%, 75%, 90%, 100% combo for example).

You could make a simple table for every map to weight the pp/combo% they give and buff nonfcs on maps with hard parts in the middle, lower the pp nonfcs in maps with hard parts in the beginning/end give.
Imo it doesn't make too much sense to reward for a play where you missed on the easy part. Giving less pp for non fcs on maps with difficulty spikes near the end or beginning does sound like a good idea though.
GhostFrog

Skyanide wrote:

In my opinion, the number of misses should have a bigger impact on how much PP a non-FC play is worth, and combo% should have a much smaller impact (or no impact at all). To account for slider breaks, perhaps there could be a PP bonus for getting a full combo (or a penalty for not getting a full combo).
You're probably right about number of misses being weighted too little and % max combo being weighted too much in judging how good a play is, but you also found the problem that prevents a more accurate system from being possible: sliderbreaks. It's not possible to get accurate information about them from past plays and they're pretty much the equivalent of misses. I don't think there's a better way to account for the possibility of sliderbreaks than using max combo achieved, though exactly how much pp should scale with combo is debatable.

Bonus/penalty for FC/non-FC might be a good thing but it can't be too big without being unnecessarily harsh towards a low number of missed slider ends on a lot of maps.
show more
Please sign in to reply.

New reply