Tanaka wrote:
Seems like the standard converts still give more pp than taiko maps.
Noticed this : http://puu.sh/7DIdO.png
Please fix.
Taiko maps should weight much more than standard converts
forum
Performance Points feedback and suggestions (Taiko)
posted
Total Posts
645
just a guess but those taikos being muzukashii might have something to do with itDolphin wrote:
Tanaka wrote:
Seems like the standard converts still give more pp than taiko maps.
Noticed this : http://puu.sh/7DIdO.png
Please fix.
Taiko maps should weight much more than standard converts
this. Some [Hyper]s are equal (or only slightly different) than general [Oni]s.AnFace wrote:
just a guess but those taikos being muzukashii might have something to do with it
i think hypers are a little bit more repetitive than oni but yeah, someone are very similar
Topic Starter
Yeah, but all the other performances that are then below your better performance will have a lower weight, so it won't make _that_ big of a difference. It also is questionable whether it'd even be possible to beat your best-ever performance with that particular map.WemadeFOX-solo wrote:
so if a performance says 50% weighted means i could get the double amount of pp if i do a better performance and get 100% weighted?
I'm wondering how much combo, misses and score have an affect on the pp (do they even have an effect on it?) Full combo should warrant something at least.
My top play has -0.23% acc compared to the one below it, and it's FC as well, but I lost 1pp. Does accuracy count that much?
My top play has -0.23% acc compared to the one below it, and it's FC as well, but I lost 1pp. Does accuracy count that much?
imo combo is pretty irrelevant. It's just a number that tells you how many objects you're able to hit without missing. Sure thats somewhat impressive, but if you look at most other rhythm games, combo is just a number, they focus more on accuracy.
Talking about games like the BEMANI series mostly. On those games, they don't even have combo multipliers, unlike osu!
You can get full combo but have godawful accuracy or you can get a few misses and have amazing accuracy, and in this case, I feel the latter is the more impressive. It's a rhythm game after all, isn't accuracy and timing what we should focus the most on?
But for reals that image you posted and losing 1pp is kinda bullshit too lol
Talking about games like the BEMANI series mostly. On those games, they don't even have combo multipliers, unlike osu!
You can get full combo but have godawful accuracy or you can get a few misses and have amazing accuracy, and in this case, I feel the latter is the more impressive. It's a rhythm game after all, isn't accuracy and timing what we should focus the most on?
But for reals that image you posted and losing 1pp is kinda bullshit too lol
This. Reason why, our match against Japan says it all. We've had a lot of FCs but due to much lower accuracy, they could catch up even by not having high combo. It's said that you can lose about 44,000 points by combobreaking.Dolphin wrote:
imo combo is pretty irrelevant. It's just a number that tells you how many objects you're able to hit without missing. Sure thats somewhat impressive, but if you look at most other rhythm games, combo is just a number, they focus more on accuracy.
I understand now, thanks for the clarification!
Topic Starter
Regarding the latest accuracy changes: Yes, I am aware, that "spam"-scores are values higher again, which I hope to address in some difficulty algorithm changes in the near future.
Now I find that accuracy has little to no value towards pp, my top performance being a 97.95% - the map isn't hard in patterns at all.
https://osu.ppy.sh/b/130115?m=1 => Gives me 264pp, despite it being OD10 and acc being extremely high. The BPM after getting to DT is lower than 190.
https://osu.ppy.sh/b/97949?m=1 => 99.90 on first try, nomod, 207bpm, 272pp
https://osu.ppy.sh/b/89355?m=1 => Low OD, easy map, 282pp. BPM after speedchange is 195.
The best thing is to weigh a bit more accuracy, because what makes a player A better than player B isn't only speed, it's a way to stay consistent with the speed - translated by accuracy difference between two scores. I didn't perform well on Mystic Oriental compared to most scores (it's one of the worst FC DT S of the rankings) but did pretty well on Shuriken or Si te vas - which are both HRDTs.
There's also the HD weighing problem, which I'd like to see how much changes they would give.
https://osu.ppy.sh/b/130115?m=1 => Gives me 264pp, despite it being OD10 and acc being extremely high. The BPM after getting to DT is lower than 190.
https://osu.ppy.sh/b/97949?m=1 => 99.90 on first try, nomod, 207bpm, 272pp
https://osu.ppy.sh/b/89355?m=1 => Low OD, easy map, 282pp. BPM after speedchange is 195.
The best thing is to weigh a bit more accuracy, because what makes a player A better than player B isn't only speed, it's a way to stay consistent with the speed - translated by accuracy difference between two scores. I didn't perform well on Mystic Oriental compared to most scores (it's one of the worst FC DT S of the rankings) but did pretty well on Shuriken or Si te vas - which are both HRDTs.
There's also the HD weighing problem, which I'd like to see how much changes they would give.
yes pls ;w;_Gezo_ wrote:
There's also the HD weighing problem, which I'd like to see how much changes they would give.
now acc is underrated imo, i would just reward a little bit more acc
WemadeFOX-solo wrote:
now acc is underrated imo, i would just reward a little bit more acc
lolnopes
kinda agree with the accuracy thing; Remote Control is by far my most accurate HDT (99.52%) and yet it's still #9. Night of Fire is easier than that, if you ask me; it's 98.88% HDT, yet it's weighted a lot more.
Or is it because SD got its own weight too? Also, I think Distorted Lovesong and Because Maybe! should swap places, as the latter is the same speed, but 10 minutes lengthier. Acc difference is .03% (99.27 and 99.24, respectively), but still. :b
also lol Carnival at #1
Or is it because SD got its own weight too? Also, I think Distorted Lovesong and Because Maybe! should swap places, as the latter is the same speed, but 10 minutes lengthier. Acc difference is .03% (99.27 and 99.24, respectively), but still. :b
also lol Carnival at #1
Well, that case should be treated as well. I can get 10 misses and no 100, I'll get more accuracy than a guy who FCs the same map with 21x 100 - The latter has performed better. It makes no sense to have more performance when not getting FC !Didou wrote:
WemadeFOX-solo wrote:
now acc is underrated imo, i would just reward a little bit more acc
lolnopes
Topic Starter
You wouldn't get the same pp. You'd get less._Gezo_ wrote:
Well, that case should be treated as well. I can get 10 misses and no 100, I'll get more accuracy than a guy who FCs the same map with 21x 100 - The latter has performed better. It makes no sense to have more performance when not getting FC !
The only reason the above 2 scores are that close is, because 1500 combo vs 1600 combo is not such a big difference and, because a difference between 97.8% vs 98.6% is not close at all.
Agree with Tom, and tbh the acc weighing is about just right. (maybe slightly low)Tom94 wrote:
You wouldn't get the same pp. You'd get less._Gezo_ wrote:
Well, that case should be treated as well. I can get 10 misses and no 100, I'll get more accuracy than a guy who FCs the same map with 21x 100 - The latter has performed better. It makes no sense to have more performance when not getting FC !
The only reason the above 2 scores are that close is, because 1500 combo vs 1600 combo is not such a big difference and, because a difference between 97.8% vs 98.6% is not close at all.
Only problem now I think is the weighing of HD mod.
Should be interesting when it comes out...
^ this, is frustrating when u do a map with misses and 98.50 acc and then fc the map with 98.00 acc and you lose pp, feels so wrong_Gezo_ wrote:
Well, that case should be treated as well. I can get 10 misses and no 100, I'll get more accuracy than a guy who FCs the same map with 21x 100 - The latter has performed better. It makes no sense to have more performance when not getting FC !
well I can't say this is wrong, it's better as it doesn't overrate acc, but now it is weighed a bit wrong due to missing barely taken into account (see didou's screen, 7pp difference despite having more acc yet misses)abrian159 wrote:
Agree with Tom, and tbh the acc weighing is about just right. (maybe slightly low)
Only problem now I think is the weighing of HD mod.
Should be interesting when it comes out...
Acc weighting doesn't depend on OD at all. This is my feeling - see 2nd previous post where I show that my OD4 HRDT gets above tons of OD5/6/7+ HRDT
He He He He He.... pay-time *rubs hands*abrian159 wrote:
Only problem now I think is the weighing of HD mod.
Should be interesting when it comes out...
At first glance, I think I like the new weighting for accuracy a lot more. I'm currently ranked around the mid-high 2,000s, and the change in my "top performance" actually gives me a little bit of hope of improving my rank at a more "reasonable" pace, so to speak.
Ever since the reintroduction of Taiko pp, my Kare Kano Kanon score was ranked rather adamantly #1 in Top Performance, being the best performance I have ever had when specifically considering accuracy:
But all of those A ranks are floating around high 96% accuracies, and most of those songs feature much more complex or faster-paced gameplay (Sa'eed, for example, containing my highest Taiko-diff combo to date, being 1,202). Before, it felt a little bad because all of the emphasis did seem to be on getting top-tier accuracy, which at my skill level isn't so consistent. Feels a lot better to see something more recent at #1 rather than something incidental from 2 months ago.
Ever since the reintroduction of Taiko pp, my Kare Kano Kanon score was ranked rather adamantly #1 in Top Performance, being the best performance I have ever had when specifically considering accuracy:
But all of those A ranks are floating around high 96% accuracies, and most of those songs feature much more complex or faster-paced gameplay (Sa'eed, for example, containing my highest Taiko-diff combo to date, being 1,202). Before, it felt a little bad because all of the emphasis did seem to be on getting top-tier accuracy, which at my skill level isn't so consistent. Feels a lot better to see something more recent at #1 rather than something incidental from 2 months ago.
Yes_Gezo_ wrote:
Acc weighting doesn't depend on OD at all. This is my feeling - see 2nd previous post where I show that my OD4 HRDT gets above tons of OD5/6/7+ HRDT
I need some help understanding how pp calculation works because I'm really confused right now.
For an example, I'll use two maps from my top performance list and compare them:
DragonForce - Through The Fire And Flames [Taiko] vs Demetori - Emotional Skyscraper ~ World's End [mekadon's Taiko]
Mods used: HR = HR (no difference)
Star Rating: 3.56 > 3.31 (TTFAF wins)
Accuracy: 99.57% ~ 98.35% (should probably be a win for TTFAF, or be about even at worst; both accuracies are mediocre, but it's OD7 vs OD10)
Misses: 0 < 5 (TTFAF wins)
Combo: 3121 (FC) > 747 (out of 2121) (TTFAF wins)
BPM (in case high SV gives bonuses with HR): 200 > 178 (TTFAF wins)
pp Value: 282 < 296 (Emotional Skyscraper inexplicably wins)
I got the Emotional Skyscraper score in multiplayer almost half a year ago, it's terrible and should not be my #5 performance. It's not a hard map and according to these stats it should give way less pp than TTFAF. Yet somehow it doesn't.
What am I missing?
For an example, I'll use two maps from my top performance list and compare them:
DragonForce - Through The Fire And Flames [Taiko] vs Demetori - Emotional Skyscraper ~ World's End [mekadon's Taiko]
Mods used: HR = HR (no difference)
Star Rating: 3.56 > 3.31 (TTFAF wins)
Accuracy: 99.57% ~ 98.35% (should probably be a win for TTFAF, or be about even at worst; both accuracies are mediocre, but it's OD7 vs OD10)
Misses: 0 < 5 (TTFAF wins)
Combo: 3121 (FC) > 747 (out of 2121) (TTFAF wins)
BPM (in case high SV gives bonuses with HR): 200 > 178 (TTFAF wins)
pp Value: 282 < 296 (Emotional Skyscraper inexplicably wins)
I got the Emotional Skyscraper score in multiplayer almost half a year ago, it's terrible and should not be my #5 performance. It's not a hard map and according to these stats it should give way less pp than TTFAF. Yet somehow it doesn't.
What am I missing?
Topic Starter
Seems like the recent accuracy update got them closer (294 vs 292). The reason really is the OD. Getting 98.35% at OD10 is considered better than 99.57% at OD7 (by about 2 accuracy pp, 125 vs 145 to be exact). That means that the rest of the scores also get a difference of ~20 pp in the other direction, making up for that.Luna wrote:
I need some help understanding how pp calculation works because I'm really confused right now.
For an example, I'll use two maps from my top performance list and compare them:
DragonForce - Through The Fire And Flames [Taiko] vs Demetori - Emotional Skyscraper ~ World's End [mekadon's Taiko]
Mods used: HR = HR (no difference)
Star Rating: 3.56 > 3.31 (TTFAF wins)
Accuracy: 99.57% ~ 98.35% (should probably be a win for TTFAF, or be about even at worst; both accuracies are mediocre, but it's OD7 vs OD10)
Misses: 0 < 5 (TTFAF wins)
Combo: 3121 (FC) > 747 (out of 2121) (TTFAF wins)
BPM (in case high SV gives bonuses with HR): 200 > 178 (TTFAF wins)
pp Value: 282 < 296 (Emotional Skyscraper inexplicably wins)
I got the Emotional Skyscraper score in multiplayer almost half a year ago, it's terrible and should not be my #5 performance. It's not a hard map and according to these stats it should give way less pp than TTFAF. Yet somehow it doesn't.
What am I missing?
Should misses / combo breaks be penalized even more? It should be noted, that you only had 5 misses, so they didn't pull the pp of emotional skyscraper down by much.
Oh, my top performance list had apparently not updated fully until just now, it was different just a few hours ago (with the stats I posted).
But it still feels weird that a score on a 3.56 diff map gets beaten so easily by a shitty non-FC score on a 3.31 diff map.
And 98.35 OD10 isn't even good, it should definitely not give 20 additional pp over the TTFAF accuracy.
So yes, I think misses should be penalized a bit more, map difficulty should probably be weighted higher (although the algorithm still needs a lot of work) and the accuracy weighting still makes little sense to me.
But it still feels weird that a score on a 3.56 diff map gets beaten so easily by a shitty non-FC score on a 3.31 diff map.
And 98.35 OD10 isn't even good, it should definitely not give 20 additional pp over the TTFAF accuracy.
So yes, I think misses should be penalized a bit more, map difficulty should probably be weighted higher (although the algorithm still needs a lot of work) and the accuracy weighting still makes little sense to me.
This is what I wanted to tell with Didou vs Nashmun case, getting very few misses can still give you an enormous reward (even by missing halfway through the map)Tom94 wrote:
Should misses / combo breaks be penalized even more? It should be noted, that you only had 5 misses, so they didn't pull the pp of emotional skyscraper down by much.
i just lost about 70pp but not rank at all, there is a new change?
I wonder how is your overall accuracy calculated?
It kept changing even thought I didn't rank any maps,
does it only counts your top 100 plays or so in the pp list..?
It kept changing even thought I didn't rank any maps,
does it only counts your top 100 plays or so in the pp list..?
If it's anything like ppv1 accuracy, it's weighted by your top performances (so your #1 top performance score has the highest impact on your accuracy and so on)
Topic Starter
Yup, that's exactly how it's done. Your accuracy is the weighted average of your scores' accuracies by the respective "Weighting" shown in your Top Ranks profile section.Luna wrote:
If it's anything like ppv1 accuracy, it's weighted by your top performances (so your #1 top performance score has the highest impact on your accuracy and so on)
Losing 2 pp cuz 0.09 of acc? imo fc need to weight more than acc without combo
Its a rhythm games. Putting more focus on accuracy makes perfect sense. Combo is just a number.WemadeFOX-solo wrote:
Losing 2 pp cuz 0.09 of acc? imo fc need to weight more than acc without combo
But I do agree that it is currently a BIT strict still.
it shouldn't when the highest combo you got, the highest score you receive ... that's somewhat rewarding your combo with score, shouldn't be the same with pp?Dolphin wrote:
Its a rhythm games. Putting more focus on accuracy makes perfect sense. Combo is just a number.WemadeFOX-solo wrote:
Losing 2 pp cuz 0.09 of acc? imo fc need to weight more than acc without combo
But I do agree that it is currently a BIT strict still.
runs
It's not quite an absolute truth, but I do mostly agree with this. Accuracy isn't worth more than combo until at least 97%, especially in this mode where, while our combo multiplier isn't multiplicative unto infinity, our timing is pretty strict and hitting 100x on some patterns can actually be the same as missing. Accuracy being extremely weighted like that kind of caters it to the top players who are doing SSes, while being non-intuitive to the middle players who can get way better accuracy but a lower score, and as a result doesn't count for pp. Losing pp because you did better is very disheartening.seicHmsc wrote:
it shouldn't when the highest combo you got, the highest score you receive ... that's somewhat rewarding your combo with score, shouldn't be the same with pp?Dolphin wrote:
Its a rhythm games. Putting more focus on accuracy makes perfect sense. Combo is just a number.
But I do agree that it is currently a BIT strict still.
runs
Would there be any way to make the score that is counted for pp be the one the system considers you performed best on, rather than just being the one you make the best score on? I don't think this would affect the top of the ladder much, but would help tons around the middle where it's easily possible to invalidate a good performance.
i feel like missing is the worst you can do in a rhythm game, worse than pressing later or earlier
Yes, imo it's better to press a note late rather than miss it completely, in terms of performance (like for example, playing a piano), the latter mistake is really obvious.WemadeFOX-solo wrote:
i feel like missing is the worst you can do in a rhythm game, worse than pressing later or earlier
I agree with you when it's just about a 1:1 ratio (1 miss is worse than one mistimed note), but that's how the system already works.
The example you gave is (if I calculated correctly) 1 miss vs. 3 mistimed notes on OD5 (so they are off by quite a lot). I'd still agree that the miss is worse in this case, but it's getting a lot closer.
If you read through the last couple of posts again, you will see that we already asked for an increased miss punishment; so once Tom has time to actually implement that, your example scores should probably look better (FC giving more than the other score). Just don't expect to beat scores with way superior accuracy just because you didn't miss.
The example you gave is (if I calculated correctly) 1 miss vs. 3 mistimed notes on OD5 (so they are off by quite a lot). I'd still agree that the miss is worse in this case, but it's getting a lot closer.
If you read through the last couple of posts again, you will see that we already asked for an increased miss punishment; so once Tom has time to actually implement that, your example scores should probably look better (FC giving more than the other score). Just don't expect to beat scores with way superior accuracy just because you didn't miss.
no, 1 miss worth 0% and 1 mistimed worth 50% of acc, so 1 miss is equal 2 mistimed notes. if u hit 4 x300 and 1 miss is the same if u hit 3x300 and 2x100.
I dunno, missing is completely losing the melody and the rhythm, i think missed need to be more punished than x100 but i dont know if its is possible
I dunno, missing is completely losing the melody and the rhythm, i think missed need to be more punished than x100 but i dont know if its is possible