Tasha wrote:

http://osu.ppy.sh/b/90616?m=1 this map being on my top ranks despite a miss gives me the feeling that the high SV section at the very end is skewing the maps rating a bit, especially when a much more difficult to FC song like http://osu.ppy.sh/b/284037 isn't on my top 10 yet Taiko Time is.

Combo isn't weighted as heavily in Taiko as it is in standard at the moment. Not exactly sure how much I should change that.

deleted

Tom94 wrote:

Tasha wrote:

http://osu.ppy.sh/b/90616?m=1 this map being on my top ranks despite a miss gives me the feeling that the high SV section at the very end is skewing the maps rating a bit, especially when a much more difficult to FC song like http://osu.ppy.sh/b/284037 isn't on my top 10 yet Taiko Time is.

Combo isn't weighted as heavily in Taiko as it is in standard at the moment. Not exactly sure how much I should change that.

imo depends on amount of misses

obviously fc is better than none, but if there's only 1-3 misses (one combo break) it should be way above 20-30 misses (probably multiple combo break) even though that's impossible to tell without watching replays or knowing exact scores or whatever. maybe each miss penalized harder than the last yea penalize me so hard

EBAWER123 wrote:

kawhy had a guide how to do it but I cba to look for it.

yes pls

edit: as far as I'm concerned, the top 12 are entirely strong DTers? o.o

You said this in the standard pp thread :

Tom94 wrote:

pp are called "performance points". They are rating your performance, not your skill. Those 2 things are very similar but not exactly the same. Let's look at a piano player for instance. Would you rate a perfect interpretation of an easier piece lower than a horrible play through some ridiculously hard piece? I wouldn't.

The first requirement for getting pp is delivering a decent performance. Map difficulty is also factored in and can make up for worse performances up to a certain degree, but barely passing won't get you anywhere. As a guideline, you should try to get below 10 misses and above 90% accuracy if you want your play to be rated well. A higher combo also does help. Of course if the map indeed is ridiculously hard, then you can get away even with a low accuracy, as long as you keep your misses low and your combo high.

I couldn't agree more and that's why I think you need to adjust the algorithm to avoid situation like that :



Like, having 'bad' performance on heavily rated maps because they are full of spam. That means, it's easy to get FC during 95% of the map, and then you have 5% full of 1/8 spam so the algorithm might think it's still a great performance because you have a relative high combo.

i really dont get how the new system works, playing some converted normals/hard for chill and i got pp, fcing specific taiko maps(muzu/oni) and got nothing, same goes with easier difficulties giving more pp than harder.

Nashmun wrote:

You said this in the standard pp thread :

Tom94 wrote:

pp are called "performance points". They are rating your performance, not your skill. Those 2 things are very similar but not exactly the same. Let's look at a piano player for instance. Would you rate a perfect interpretation of an easier piece lower than a horrible play through some ridiculously hard piece? I wouldn't.

The first requirement for getting pp is delivering a decent performance. Map difficulty is also factored in and can make up for worse performances up to a certain degree, but barely passing won't get you anywhere. As a guideline, you should try to get below 10 misses and above 90% accuracy if you want your play to be rated well. A higher combo also does help. Of course if the map indeed is ridiculously hard, then you can get away even with a low accuracy, as long as you keep your misses low and your combo high.

I couldn't agree more and that's why I think you need to adjust the algorithm to avoid situation like that :



Like, having 'bad' performance on heavily rated maps because they are full of spam. That means, it's easy to get FC during 95% of the map, and then you have 5% full of 1/8 spam so the algorithm might think it's still a great performance because you have a relative high combo.

I'll address it this weekend. While it's valid for some players to have such plays in their best performance list (if they're not good at other things), it shouldnt be able to give so much pp in general.

WemadeFOX-solo wrote:

i really dont get how the new system works, playing some converted normals/hard for chill and i got pp, fcing specific taiko maps(muzu/oni) and got nothing, same goes with easier difficulties giving more pp than harder.

Please name the maps and show the scores, so I can check it.

Of course if the player doesn't have a lot of scores they should appear in his bp. But it's giving pp to everyone and the one I screened is in the top #50.

AnFace wrote:

yea penalize me so hard

ily

XK2238 wrote:

EBAWER123 wrote:

kawhy had a guide how to do it but I cba to look for it.

yes pls

edit: as far as I'm concerned, the top 12 are entirely strong DTers? o.o

It seems accurate, as those players possess enough accuracy and precision in reading notes to hit them at faster speed, and more speed = more BPM, and at that point playing well consistently at those speeds is reason enough to be placed fairly high. Though it makes noobita OP :3

Nashmun wrote:

You said this in the standard pp thread :

Tom94 wrote:

pp are called "performance points". They are rating your performance, not your skill. Those 2 things are very similar but not exactly the same. Let's look at a piano player for instance. Would you rate a perfect interpretation of an easier piece lower than a horrible play through some ridiculously hard piece? I wouldn't.

yesssssss

ozzyyyyyyyyyyyy i thought u ded

OzzyOzrock wrote:

Tom94 wrote:

pp are called "performance points". They are rating your performance, not your skill. Those 2 things are very similar but not exactly the same. Let's look at a piano player for instance. Would you rate a perfect interpretation of an easier piece lower than a horrible play through some ridiculously hard piece? I wouldn't.

yesssssss

Now does that "yes" agree with me or not? D:

(Can't embed 3 quotes within each other, so I removed Nashmun's enclosing one.)

so the new system give your pp based on high acc, high bpm and dt is the best mod right now? doing fc in onis with 95%acc is worse than some normal with dt 99 acc

WemadeFOX-solo wrote:

so the new system give your pp based on high acc, high bpm and dt is the best mod right now? doing fc in onis with 95%acc is worse than some normal with dt 99 acc

Yes, because 95% acc is considered garbage in taiko. DT is obviously the best mod being the hardest one after FL and HR, isn't it obvious?

and u think a player who does 10 converted normals with dt is better than a player who does 10 oni with 95 acc?

depends on what you call oni and normal

@Tom why do these 2 scores are valued equally 1st have better score, better acc and less misses 2 vs 15 yet no pp increase:?



Map: https://osu.ppy.sh/b/58061&m=1

Is it because the difference is so little?

I earned 1 pp with that DT play and immediately lost it after that DTHR play just because of accuracy, altough the DTHR play is a lot harder to achieve. I could 99 DT this map anytime but it's hard to keep a consistent 96+ with DTHR. The pp system as it is now favors way too much DT. If you want pp, then don't bother HR/HD or DT + mod because you will most likely earn less pp than with DT only because your accuracy will be lower. IMO that's something that should be fixed.

so the new system is, 98+acc or forget about your pp no matter what difficulty are u playing

WemadeFOX-solo wrote:

so the new system is, 98+acc or forget about your pp no matter what difficulty are u playing

Basically

Not rly there are some garbage records with C and B in my top ranks, so u can have bad acc as long as the map is some kind of hard

talking about regular maps, no that retarded 1/16 or 1/8 streams which are almost impossible to do even with HT

WemadeFOX-solo wrote:

talking about regular maps, no that retarded 1/16 or 1/8 streams which are almost impossible to do even with HT

?

So how come I had 19 100's on https://osu.ppy.sh/b/147656&m=1
Then I did it with HD which was hard as F***, and lost like 12 pp? I had pretty good accuracy too... 44 100's --> 98.45% accuracy
Also I DT a song, 11 100's +3 pp cool... DTHD 18 100's -3pp...

What's going on? Am I just better off getting high accuracy with no mods than slightly lower accuracy with 1 mod?

you went from 19x 100 to 44x 100, there is no way ever that the bonus from hd can outweigh that.
same with the DT score, you went from 11x 100 to 18x 100.

protip: 19x 100 -> 44x 100 is not a slight decrease, it is a huge one.

Is HD even weighted? Let alone correctly.

How complete is the algorithm? I know it's still WIP, but can you at least clear up some values so people don't get the wrong idea.

HD is different to standard. As opposed to requiring aim, taiko does not. You also have to differentiate between colours and big notes in less time, at the same time. With no aim component, how do you weight the difficulty of reading and its effects on accuracy?

There are not a lot of good HD taiko players from what I can tell, whereas nearly all scoreboards in osu!standard are filled with HD scores, making it important to learn HD. As a result, it's often referred to as "free points." Learning HD is not a requirement in taiko though, probably for the reason that it is genuinely difficult. It's definitely not "free" unless you are using double-screen.

>HD
>DS

nnnnnope. It's definitely readable (I mean, what is that area for?), but it's the default skin (because people who don't play taiko as the primary tend to not have custom Taiko skin files, imo) which bends a heck lot of perspectives about Taiko. I do know that there are some who can't HD Taiko even without the default skin, but DSing HD is just obviously dumb.

tl;dr: it's obviously readable, you just need some time to get used to it.

XK2238 wrote:

tl;dr: it's obviously readable, you just need some time to get used to it.

Is it "free?"

You don't play standard that much, so maybe you don't know what I mean by "free," because that's where I draw the comparison from. It seems to me that the 19x100 vs 44x100 (-12pp) is something that was borrowed directly from standard, because that's pretty much the same scale of effect it has there, where it's bonus to accuracy is negligible (so getting worse accuracy with HD will always be worse, discounting aim)... because literally... its effect on accuracy is actually negligible.

Do you agree with this for taiko?

If so, why are there so few HD scores? What does HD actually make harder, if anything?

RaneFire wrote:

XK2238 wrote:

tl;dr: it's obviously readable, you just need some time to get used to it.

Is it "free?"

You don't play standard that much, so maybe you don't know what I mean by "free," because that's where I draw the comparison from. It seems to me that the 19x100 vs 44x100 (-12pp) is something that was borrowed directly from standard, because that's pretty much the same scale of effect it has there, where it's bonus to accuracy is negligible (so getting worse accuracy with HD will always be worse, discounting aim)... because literally... its effect on accuracy is actually negligible.

Do you agree with this for taiko?

If so, why are there so few HD scores? What does HD actually make harder, if anything?

You don't play taiko, do you? =)

silmarilen wrote:

you went from 19x 100 to 44x 100, there is no way ever that the bonus from hd can outweigh that.
same with the DT score, you went from 11x 100 to 18x 100.

protip: 19x 100 -> 44x 100 is not a slight decrease, it is a huge one.

So if I add a mod to make it more difficult, I'm expected to get the same accuracy or better than with the previous 1 mod or no mods to get pp :c?
I cannot do that. Adding HD to a song that is fast is very difficult and you have to play it over and over, so I'd expect at least 1 pp maybe u.u not a subtraction of 12 D:
Sure 19x 100 to 44x 100 is a big difference but 98.45 for 44 100s with HD is good for Cruel clocks in my opinion. Also, 11x 100 to 18x 100 is not a big difference. DT 11x 100 is better than DTHD 18x 100 I guess.

EBAWER123 wrote:

You don't play taiko, do you? =)

Apparently not. I just hit some funny buttons now and then.

If you're fine with the system as is, then please... don't enlighten me. This is not for my sake anyway, although it is for my curiosity, and maybe others'. I ask the question from a third-person perspective because I do not play taiko at the level which counts for considering mod difficulty weightings. Remember, Tom also doesn't play taiko. He also needs your input, so you better keep posting if you want improvements, because there's no way he's going to be able to deduce them all by himself.

And pretty much as SeasonS said... HD is a difficulty increase. The question is to where it should apply. If you don't want to answer that, and just dodge the question, then what good is debating the ranking system.

Elitism does not help anyone.

In taiko, HD is pretty hard to judge difficulty-wise. It mostly depends on how the notes are scrolling. A song where the notes scroll by really fast is going to be way harder with HD, where as using HD on a song like http://osu.ppy.sh/s/13451 doesn't make it too much harder. Then, you have things like extremely slow SVs making HD harder and songs where the SV changes a lot. Also, long, complicated streams ( http://osu.ppy.sh/s/24634 ) make HD way harder as well. There's probably some other factors, but that's all I can think of right now. As a whole, I wouldn't say HD is worth more than HR, but HD does make a notable increase in difficulty.

RaneFire wrote:

There are not a lot of good HD taiko players from what I can tell

*sobs quietly in a corner*

I personally find HD to be easier than HR. (not by much really) Mostly because I feel I can play a larger asset of maps well with it, opposed to HR where I can only play below 180 BPM (on SV 1.40), and I usually screw up accuracy a lot on OD6+HR. (For some reason I have the same accuracy on OD5 with or without lol)

But there are plenty of good HD taiko players imo, but most players find HR to be easier (for some reason)
A lot of the Japanese players are good with HD, for instance.
You usually see HR scores on maps below ~190 BPM because its easier to read and play with for most people, whilst on maps with higher BPM (like 190+) you'll start seeing more HD scores, since its rather difficult to read such a high BPM/SV with HR.

Not that any of this matters, just in case you wanted to know.

Dolphin wrote:

You usually see HR scores on maps below ~190 BPM because its easier to read and play with for most people, whilst on maps with higher BPM (like 190+) you'll start seeing more HD scores, since its rather difficult to read such a high BPM/SV with HR.

Just pointing something. It could be simply a matter of score, since HR is more rewarding than HD if you have at least one spinner in the whole map.

Dolphin wrote:

You usually see HR scores on maps below ~190 BPM because its easier to read and play with for most people, whilst on maps with higher BPM (like 190+) you'll start seeing more HD scores, since its rather difficult to read such a high BPM/SV with HR.

It really starts to abruptly change once you hit 220+. Past 230, they barely exist. Reach 240, only a few select can even do it. 250, you're sure that there are at most 2. Reach 280, there are none. At 200 or 210 you see some at non-negligible amounts. But it starts to really break past 220. 190 if SV is higher.

For HD, you can pretty much expect scores on a range from 120-130 to 300-320 BPM (I can read up to 300-ish with no problem despite my unfamiliarity with said mod) -- anything below is known to be hell difficult even by the HD players themselves - point proven in TWC where barely any HD map was picked.

Whoever legitimately thinks that HD and DS are correlated need their head checked. The area is pretty large to read on, so people can play it correctly.

_Gezo_ wrote:

Whoever legitimately thinks that HD and DS are correlated need their head checked. The area is pretty large to read on, so people can play it correctly.

but if i can't do it then they must be cheating

_Gezo_ wrote:

Whoever legitimately thinks that HD and DS are correlated need their head checked. The area is pretty large to read on, so people can play it correctly.

Taiko players so sensitive. I didn't make any accusations, I was just using it as an example in my "free points" analogy. Maybe I shouldn't have said that, but I didn't know how else to put it, or what example I could have used instead. I did not imply that people were legitimately using DS.

Dolphin wrote:

Not that any of this matters, just in case you wanted to know.

I did. Thanks for replies everyone.

Must not assume that, because this is an old game mode, what's obvious to you is obvious to everyone.

RaneFire wrote:

XK2238 wrote:

tl;dr: it's obviously readable, you just need some time to get used to it.

Is it "free?"

You don't play standard that much, so maybe you don't know what I mean by "free," because that's where I draw the comparison from. It seems to me that the 19x100 vs 44x100 (-12pp) is something that was borrowed directly from standard, because that's pretty much the same scale of effect it has there, where it's bonus to accuracy is negligible (so getting worse accuracy with HD will always be worse, discounting aim)... because literally... its effect on accuracy is actually negligible.

Do you agree with this for taiko?

If so, why are there so few HD scores? What does HD actually make harder, if anything?

js, I played HD more than you can think out of my Std playcount, and yes, it's "free" in std, but why did you even compare these 2 clearly contrast game modes in such terms from the first place anyways?
-------------
Difficulty-wise:
1. HD in slow SV maps adds (a lot) more challenge (the slower, the harder), while HR kills the challenge itself.
2. HD in very fast SV maps also adds about the same amount of challenge as above (the faster, the harder), unlike Std which is totally unaffected by SV, only AR (AR8/9 HD is very common anyways, even I can read it). As for HR, people won't even think about playing such maps with it. The case is different for Taiko-specific maps, where the SV is almost always the same multiplier, making only the BPM count.
3. Talk about long, complex streams... HD will surely give you a hard time reading such streams, while HR simply eases it off (although depending on the BPM, as said in No. 2).
4. Different beat snaps in one pattern/stream? Pretty similar as above.

About accuracy, it's pretty much NOT that negligible in Taiko, whose timing window is around half as narrow than Std, from what I've seen. On top of that, I know a few people who can easily score SS HR in already insane ODs (or even with DT added) but have a rough time trying to achieve the same SS in HD (which has obviously less OD than HR) for the sake of it, so no, I can't agree with you.

As a side note, the OD increase caused by HR adds more spinner hits required to clear it in Taiko, earning you more score than in HD, which is why HR SSes have more score than HD SSes in maps with spinners, attracting players to go HR in such maps (especially if the spinner is like in AugoEiDes or such).

I felt like I still have few more things to add, but I can't come up to it. Threw my three cents.

RaneFire wrote:

_Gezo_ wrote:

Whoever legitimately thinks that HD and DS are correlated need their head checked. The area is pretty large to read on, so people can play it correctly.

Taiko players so sensitive. I didn't make any accusations, I was just using it as an example in my "free points" analogy. Maybe I shouldn't have said that, but I didn't know how else to put it, or what example I could have used instead. I did not imply that people were legitimately using DS.

And this, people, is why I hate generalizations.

</post>

can we know if this current system is the last one or is going to have some modifications since is pretty broken

Tom94 wrote:

I'll address it this weekend. While it's valid for some players to have such plays in their best performance list (if they're not good at other things), it shouldnt be able to give so much pp in general.

I don't mean to be impatient or rush anything, but two (2) weekends have passed since this was posted and absolutely nothing has been changed.
I'd like to give further feedback on the system, but I want to wait until after the announced update at least. Changing too many things at a time would only make it harder to pinpoint why things improved/went downhill.
Btw, by now eoe has farmed his way to #1 through spam maps. Yes, he is a very good player. But he does not deserve #1, and especially not because of those maps.

Luna wrote:

Tom94 wrote:

I'll address it this weekend. While it's valid for some players to have such plays in their best performance list (if they're not good at other things), it shouldnt be able to give so much pp in general.

I don't mean to be impatient or rush anything, but two (2) weekends have passed since this was posted and absolutely nothing has been changed.
I'd like to give further feedback on the system, but I want to wait until after the announced update at least. Changing too many things at a time would only make it harder to pinpoint why things improved/went downhill.
Btw, by now eoe has farmed his way to #1 through spam maps. Yes, he is a very good player. But he does not deserve #1, and especially not because of those maps.

You're completely right, I apologize for missing this deadline. I had experimented with these things and had a modified version running locally and I wonder how it slipped getting pushed to the servers. Sorry again. :/
eoe has quite good accuracy on some of these maps, thouh, so I can't really say what exactly will happen. He will likely drop from #1 but I don't know how much. Might require a change to the difficulty calculation part, reducing the impact of speed even further compared to rhythm and color complexity. Would that make sense?

If you are going to drop the speed weighting so heavily that spam maps become fairly weighted, truly difficult speed maps would likely be underrated.
I still prefer just cutting the pp gain from converted maps by a decent margin, as was requested initially.