forum

Performance Points feedback and suggestions (Taiko)

posted
Total Posts
645
show more
Fudgy

lolcubes wrote:

I agree, but every time removal was suggested, it was declined.
Since TAG4 maps have been removed from pp in standard, converts should be removed too, it's a sound logic if you take into account that there are plenty of converts which are easy to A with shit accuracy and high combo to get insane pp, however since all suggestions have been declined, a flat pp penalty is the 2nd best thing.
Is there somewhere I can find the reasons why all suggestions have been declined and if they still apply to now?

Also, the current problem is not how much pp converts give, but the fact that they give ANY pp. I don't understand the point of a pp penalty. What does that even fix? Why would it be accepted to reduce pp but not remove it? What reasoning is behind that?

Are we trying to reduce pp farm or encourage new taiko players (and active) to play taiko maps? Those were specially made for them in the first place.
OnosakiHito
Let's say that this is currently under some kind of observation. Whether current examples of converts are enough or not I don't know, but it doesn't hurt to provide more for the moment to have a more refined and fasten statement.
toara_fict
I agree

Sashimio
I think HR is difficult so we should be given more pp.
RNG
I want you to when add EZ, almost without pp. And I want you to reduce std map pp.
abrian
https://osu.ppy.sh/b/157233&m=1

98.20% S with NC only gave 371pp :>
seems legit
Prophecy
I think SV change can be the factor too.
Seems SV change in High BPM song is hard.
Yuzeyun
@ Prophecy:

Tom94 wrote:

Unfortunately the infrastructure of taiko pp still doesn't really have a measure for scrolling speed, so I can't easily add a HR or HD bonus relative to scroll speed. The general consensus seems to be, that HD is worth too much right now, so I'll move it downwards a bit.
OzzyOzrock
Rankings would be so much cuter with taiko maps only (⌯˃̶᷄ ﹏ ˂̶᷄⌯)

and kill EZ mod wtf, people get punished for playing a song with no mods because then they can't get a score with EZ? lmfao
OnosakiHito

OzzyOzrock wrote:

and kill EZ mod wtf, people get punished for playing a song with no mods because then they can't get a score with EZ? lmfao
Oh, that's something new to me. What do you mean?
OzzyOzrock

OnosakiHito wrote:

OzzyOzrock wrote:

and kill EZ mod wtf, people get punished for playing a song with no mods because then they can't get a score with EZ? lmfao
Oh, that's something new to me. What do you mean?
Well, if you notice a lot of top players' top scores, you'll see that they often pair EZ with DT for the pp increase that comes with higher accuracy given by EZ.

I don't know the numbers exactly, but by the looks of it EZ doesn't seem to give any penalty whatsoever for being used despite lowering SV and OD. As in, it seems that the pp system reads "Player X player X song at X BPM with X accuracy", but ignores that EZ mod is used. My score can be 50% of the top score, but pp system doesn't notice, and gives me full pp.

This also means that if I am to play, say, The End of a Century with no mod, there go my chances of playing it with EZ DT and instead I have to work for my pp playing it with DT, giving me a direct disadvantage than if I would have played with EZ DT.
Yuzeyun

OzzyOzrock wrote:

Well, if you notice a lot of top players' top scores, you'll see that they often pair EZ with DT for the pp increase that comes with higher accuracy given by EZ.

I don't know the numbers exactly, but by the looks of it EZ doesn't seem to give any penalty whatsoever for being used despite lowering SV and OD. As in, it seems that the pp system reads "Player X player X song at X BPM with X accuracy", but ignores that EZ mod is used. My score can be 50% of the top score, but pp system doesn't notice, and gives me full pp.

This also means that if I am to play, say, The End of a Century with no mod, there go my chances of playing it with EZ DT and instead I have to work for my pp playing it with DT, giving me a direct disadvantage than if I would have played with EZ DT.
There actually is a penalty, but only because of lower OD (I probably could estimate it at only 10% penalty). As quoted earlier in this thread (come on how can you not consider it), SV can't be taken into account because there is nothing that can measure it.
TimmyAkmed

OzzyOzrock wrote:

This also means that if I am to play, say, The End of a Century with no mod, there go my chances of playing it with EZ DT and instead I have to work for my pp playing it with DT, giving me a direct disadvantage than if I would have played with EZ DT.
Here goes the same problem with HD/HR. Since HD gives more pp than HR on most maps due to ranked maps being mapped with OD5. How do I beat my HR SS scores with HD? It should display in the scoreboards the maximum score play. but count the best pp play in your profile. Or just sort the scoreboard so the highest pp play = top score of the map.

This part isn't related entirely to Taiko but imo this idea will be the best when the Taiko pp system will be alright.
Fudgy

TimmyAkmed wrote:

OzzyOzrock wrote:

This also means that if I am to play, say, The End of a Century with no mod, there go my chances of playing it with EZ DT and instead I have to work for my pp playing it with DT, giving me a direct disadvantage than if I would have played with EZ DT.
Here goes the same problem with HD/HR. Since HD gives more pp than HR on most maps due to ranked maps being mapped with OD5. How do I beat my HR SS scores with HD? It should display in the scoreboards the maximum score play. but count the best pp play in your profile. Or just sort the scoreboard so the highest pp play = top score of the map.
This will be the case in the future. This multiple score leaderboards feature is ready to be implemented, but they can't implement it for the moment.

From ppy's blog :

@ppy Do you have any plans on allowing multiple scores per map? If you are, when about will it be implemented? –zargotlewl

There’s a few areas that need some overhauling before this can go live, even though most of the infrastructure already supports this.
Tom's answer from 5 days ago : http://ask.fm/Tom94/answer/131223922638
Yuzeyun
Another concern, which isn't directly related to pp itself but regards the star rating, is that with certain patterns, SR tends to horribly skyrocket. This has been probably brought up in the converts discussion with 1/8 doubles (which is common since you have 1/8 sliders now, ggnore), but this does not stop here.

In Kantan difficulty, if you don't keep a constant rhythm with 1/1 notes, the SR skyrockets a lot. That is, take a map which is 1/1 only, and a map which is objectively easier but has 3/2 notes:
The difference is small on that example, but on larger note amounts, this can increase the rating greatly.

Futsuu difficulty: Double notes is a factor of massive star rating increase, and that carries over to every possible snapping (with even more dramatic increase as the snapping is tinier, to the point it reaches ridiculous ratings, thinking about the converts we've talked about earlier in this thread) - Proof:

When spacings between notes are regular (2/1, 1/1, 1/2 only)
When 3/2 spacings and 1/2 doubles happen:

Another concern is the 1/6 + 1/4 streams, which end up giving ridiculous ratings:

Comparing to maps that are purely 1/4, this has to be a huge problem:

Another problem is streams. Depending on how you construct a stream, the map might be overrated a bit or ridiculously underrated.
https://osu.ppy.sh/s/28609 Kyoai - 5.06 stars for fairly simple streams
https://www.dropbox.com/s/dwcrd3gbk3hhl ... 4.osz?dl=0 Starstream 7.14 - 5.04 stars for painful streams.
I've also noticed that sometimes, long streams containing kkkdk somehow get rated lower than streams having only {{kk}*{dd}*}*.
[Ayase Eli]
That's some pretty impressive findings... I always wondered if certain patterns were responsible for that one song I can't get an A on even though it's an easy SR for me. This should definitely be helpful for whoever might decide to work on the Taiko pp system, and a fresh reprieve from the old auto-convert witch hunt :D
karterfreak
So is this being looked at anymore? I feel like this thread has been largely ignored... (and Gezo brought up some really good points too...)

That said, system really needs some work, some maps are vastly overrated while others are vastly undrerrated (not going to provide further examples, plenty have been provided in this thread). This is amplified when you consider DT, as some maps give ridiculous pp despite being far easier than an equal SR nomod counterpart.

One thing I'd like to see beyond improvement of the algorithm is making the pp difference between a good score and an amazing score larger. (Why is it that taiko still has no pp scores that come even close to top standard ones despite some of them being equally as impressive?) This should come after fixing both the pp algorithm and the SR algorithm. Some of the top scores performed by the best players in the game should be getting 650pp+ without a doubt. This would also much better show the actual gap between some of the top 50 and top 10 players (It's much larger than portrayed in the rankings currently).

Oh and one last thing, I still think autoconverts have no place in taiko and are a large part of why the algorithm doesn't work as it should. If I recall correctly, the algorithm tries to compensate for things that happen in convert maps that don't happen in regular ranked taiko maps (1/8 spam due to slider converts, I'm not sure if this has been changed since then), and therefore limits the points players get from high speed plays, leading to the situation mentioned above (amazing scores not getting adequate pp). I know that this has been argued to the ends of the earth, but I fail to see why we should continue to support converted maps that would never get ranked in taiko when we have tons of ranked maps to choose from for taiko nowadays.
OnosakiHito
Currently developers are having an eye on mania. Later they will maintenance Taiko.
About converts I did some months ago my move and contacted Staff about it. Removing converts from the ranking would lead in lower converted difficulties not compensating the loss of a Futsuu anymore, which would lead to not being allowed getting Muzu+Oni GDs anymore ranked and hence, make ranking for Taiko GDs very hard. Considering our situation with all the ranking, I'm not confident about doing that move yet even though I'm the same opinion like you when it's about converts. Maybe in the future when we have a more stable ranking process we can remove converts and do a poll for it as well.


General Ranking Criteria wrote:

The difficulties in the mapset must be in a consecutive order. Easy or Normal can be skipped if the gap in the star rating spread allows it. The order can be seen in the chart below. If your mapset has two difficulties, one of them cannot be an Insane or Expert
Starry-

OnosakiHito wrote:

Currently developers are having an eye on mania.
Lately I doubt even this. We've heard no news or progress in the past months, and considering http://ask.fm/Tom94/answer/133131856078 I'm really not quite sure what the current plans are in terms of working on pp for mania, let alone Taiko.
dewero
I was under the impression that Tom had stopped working on pp entirely due to his PhD
Sleepteiner
I believe that higher OD levels should make the star difficulty rating higher. Out of all the maps in the the length<100 star<1.4 range, the only ones that are taking me more than 5 tries to SS are the maps in the 4-7 OD range. Some are taking me significantly more than 5 tries. Anything with that low of a star rating shouldn't be that hard in my opinion.
SKSalt
What was changed about Taiko pp system? I just got free 46pp lol

I want to know more details about that :x
Hanjamon

SKSalt wrote:

What was changed about Taiko pp system? I just got free 46pp lol

I want to know more details about that :x
same here
Yuzeyun

Map : https://osu.ppy.sh/p/beatmap?b=776951&m=1

This is fucking stupid. HR is worth nothing. This proves that HR pp is solely based on OD, literally nothing else. And what's even worse is that I personally checked the algorithms and I did find that HR has the least bonus of all three non-time-based difficulty-increasing mods.
dewero
Well I mean aren't the only effects of HR increasing the scroll speed and increasing the OD and HP? That beatmap is already OD10 so HR doesn't change the OD, so the only thing that's different is the scroll speed, which currently isn't checked by the pp algorithm. So it makes sense that the pp doesn't change.

Also I thought it was already known that HR pp is based only off the increased OD
SKSalt

_Gezo_ wrote:


Map : https://osu.ppy.sh/p/beatmap?b=776951&m=1

This is fucking stupid. HR is worth nothing. This proves that HR pp is solely based on OD, literally nothing else. And what's even worse is that I personally checked the algorithms and I did find that HR has the least bonus of all three non-time-based difficulty-increasing mods.
This map is OD10 but pp is HD>>HR=none. okay.

And I'll present my thoughts below.

1) EZ’s OD is not considered.
Now, when we attach EZ, pp decrease only 10%. But I think that still isn't enough. Please look at this map. https://osu.ppy.sh/b/93236?m=1

(pp=385.38)

In this way, when we attach EZHD, It is easy to supply a shortage. I think that is not good. HR=consider OD / EZ=not consider OD? It becomes easy to take precision and read a map. I want you to consider about EZ's OD.

2) A difference of pp is narrow.
For example about this map : https://osu.ppy.sh/b/896390&m=0
Standard : SD=1.74, HDDTHRFL=about 25pp, SD=2.99. HDDT=about 82pp, SD=4.53, HD=about 130pp, SD=6.18, HD=about 301pp

Taiko mapset is from here : https://osu.ppy.sh/s/371857
Taiko : SD=1.51, None=about 65pp, SD=3.0, DT=about 152pp, SD=4.59, None=about 194pp, SD=6.17, None=about 302pp

I mean that way of rising of pp is too flat. Therefore when yu68 achieved super ultra record, he gets pp less than expectation. Also beginner player can easily to get pp than other mode. This is my request. I want to see the ranking that a little clearer difference than currently ranking.

3) About spam overweighted pattern.
It does not justify repetition, but at least, long doublet 1/4, 1/6, 1/8's pp should be fixed.
https://osu.ppy.sh/p/beatmap?b=400238&m=1
https://osu.ppy.sh/b/95733?m=1
https://osu.ppy.sh/p/beatmap?b=655794&m=1
Such 3 maps, the pp system does not work anymore.
dewero
The thing is, all of these issues are already known by most of the community, and continually raising them isn't going to benefit as much. What I think we should start doing is develop a formal algorithm to more accurately determine the start rating and pp curves of a given beatmap and to address the many issues raised by this thread. That way, someone can create a new pp-calculator implementing this algorithm that can potentially replace the current system
SKSalt

dewero wrote:

The thing is, all of these issues are already known by most of the community, and continually raising them isn't going to benefit as much. What I think we should start doing is develop a formal algorithm to more accurately determine the start rating and pp curves of a given beatmap and to address the many issues raised by this thread. That way, someone can create a new pp-calculator implementing this algorithm that can potentially replace the current system
Yes, but I had not yet said own opinion about this. So I just posted :oops: (similar to other opinions though)
Yuzeyun

dewero wrote:

The thing is, all of these issues are already known by most of the community, and continually raising them isn't going to benefit as much. What I think we should start doing is develop a formal algorithm to more accurately determine the start rating and pp curves of a given beatmap and to address the many issues raised by this thread. That way, someone can create a new pp-calculator implementing this algorithm that can potentially replace the current system
thats literally what i'm trying to do but the program just won't run
AAAAAAAAAA
dewero
What are you having trouble with? I can help you out if you need it
Yuzeyun
hmu in pm it's about osu-performance
[Ayase Eli]
Do we like, actually know if anyone is even available to work on these issues? Like I know there's that one guy everyone refers to when it comes to pp, but have we even established there is someone to take any of the 40 something pages worth of complaints here to?

I don't mean to be a downer, but do we even have reasonable hope to expect some kind of change to the pp system in the next, say, year...? At this point I'm just much more inclined to sit back and let pp do its thing. Its just like every other game, you gotta play by its rules to win.
Yuzeyun
there was an alternate star rating scaling that never happened to live outside of this thread
dewero
When my finals are finished (after this week :D ), I'd be willing to spend some time trying to come up with a better difficulty calculator
darkmiz
please buff hr

i need pp so much
karterfreak
HR doesn't need any more buffs, screw that.
Yuzeyun
i would qft if it wasn't underweighted on low OD and OD10

rescaling OD curve is the only option for now
darkmiz
https://osu.ppy.sh/b/67204&m=1

why oni > inner oni

why pp farm with so easy patterns
Samatrain
k
Shinsekai-

darkmiz wrote:

https://osu.ppy.sh/b/67204&m=1

why oni > inner oni

why pp farm with so easy patterns
You need to see the patterns to comment.... the Oni diff have so much color changes and paterns what make the diff more complicated for example we can take the kkk kkkk kkkk kkk kkk part... For a KKDD or DDKK player that part is really hard for a KDDK is not... (AT ALL), What i mean with this? the colors and the position play a important with the SR
Please sign in to reply.

New reply