No, that is the entire reason for t/220119 this thread's existence.Sword wrote:
Does anyone know why I lost performance points?
I had FC'd TwoThirds - Breathe with really high accuracy, and later FC'd it with a much lower accuracy with HD, but beat my old no-mod score. I lost 2pp in the process, but I thought that only the highest pp giving score counts?
forum
Performance Points feedback and suggestions (Standard)
posted
Total Posts
2,750
I don't think the scoring system should be removed for the system to count only your highest pp score. I think the highest pp score should count only in your pp list; I don't want the scoring system to be removed. Maybe edited, but not removed.Tess wrote:
No, that is the entire reason for t/220119 this thread's existence.Sword wrote:
Does anyone know why I lost performance points?
I had FC'd TwoThirds - Breathe with really high accuracy, and later FC'd it with a much lower accuracy with HD, but beat my old no-mod score. I lost 2pp in the process, but I thought that only the highest pp giving score counts?
Nobody said the scoring system would or should be removed, though.
Hasty assumptions are usually wrong.
Hasty assumptions are usually wrong.
Uhh what exactly am I assuming? Is the thread not about the pp system replacing the scoring system?Tess wrote:
Nobody said the scoring system would or should be removed, though.
Hasty assumptions are usually wrong.
it's not about replacing the score system, it is about adding next to the score system
Well as long as the score system never leaves, and the pp ranking system always selects your highest pp score regardless of what your highest score is, I'll be happy. The way that scores can easily be taken by a score of less PP is a big problem with the system and causes too much confusion and too many threads of people asking why they lost PP.
Those maps are ar9. HD is a lot easier when the AR is higher(9.6+). I really can't find any AR11 scores that don't have HD attached onto them. Meanwhile you will also rarely ever see HD attached to EZ. I suggest the aim bonus from HD be higher when the AR is lower than 9.6. Vice versa for AR higher than 9.6. It doesn't solve the problem with HD only being harder with strange overlaps/long streams in higher ARs though. That could be solved with pattern/reading difficulty + every hit object data.-GN wrote:
HD is not a playstyle. HD is a difficulty increasing mod. If you need factual examples of HD actually increasing difficulty, I've got a couple.
There is also the another issue I have with HD. It doesn't give enough points when using FL+HD. After you get used to playing FL it's hard but it's still nowhere near as hard as HD+FL which is almost impossible to sightread. It's similar to HR+DT jump in difficulty but HD+FL doesn't give nearly as much points as HR+DT. Buffing HD+FL right now would give some maps too much potential for points. That's just an issue with FL being a flat bonus though.
tl;dr
1) maps with low AR need higher bonus + vice versa
2) HD+FL doesn't give enough points
I'd like to sign that.pielak wrote:
SPOILERtl;drThose maps are ar9. HD is a lot easier when the AR is higher(9.6+). I really can't find any AR11 scores that don't have HD attached onto them. Meanwhile you will also rarely ever see HD attached to EZ. I suggest the aim bonus from HD be higher when the AR is lower than 9.6. Vice versa for AR higher than 9.6. It doesn't solve the problem with HD only being harder with strange overlaps/long streams in higher ARs though. That could be solved with pattern/reading difficulty + every hit object data.-GN wrote:
HD is not a playstyle. HD is a difficulty increasing mod. If you need factual examples of HD actually increasing difficulty, I've got a couple.
There is also the another issue I have with HD. It doesn't give enough points when using FL+HD. After you get used to playing FL it's hard but it's still nowhere near as hard as HD+FL which is almost impossible to sightread. It's similar to HR+DT jump in difficulty but HD+FL doesn't give nearly as much points as HR+DT. Buffing HD+FL right now would give some maps too much potential for points. That's just an issue with FL being a flat bonus though.
1) maps with low AR need higher bonus + vice versa
2) HD+FL doesn't give enough points
people attach HD to AR11 scores because once you can """read""" a map on AR11 you have memorized it to a point that HD doesn't matter anymore, this doesn't change the fact that HD increases the reading difficulty.
HDHR is hell for me.
HDHR is hell for me.
The problem is, similar to what you were saying to GN, is how much hidden is affected by low AR. If the AR is low in comparison to the bpm/jumps in a map, only then does hidden can make a big difference with FL (feels like hitting invisible notes). However, unless a map is short, you're going to be memorizing a map for FL anyway. Usually if I'm practicing FL for a map, I don't use hidden until I'm able to FC it, and once I add hidden, there isn't really any big difference. I think it's more based on the person and the confidence, perhaps? (I know people like Mephix can't play FL with HD) Basically, not everyone that plays FL is highly affected by adding hidden as well, but some people can't think clearly with it on. All more proof that FL is indeed a skill and not pure memorization ability.pielak wrote:
After you get used to playing FL it's hard but it's still nowhere near as hard as HD+FL which is almost impossible to sightread. It's similar to HR+DT jump in difficulty but HD+FL doesn't give nearly as much points as HR+DT. Buffing HD+FL right now would give some maps too much potential for points.
It's a shame that PP is only broken into physical abilities of aim, speed, and accuracy, because the mental ability of reading should be a section of itself, where things like hidden, easy, really low or high AR, flashlight, and maybe weird timing/spacing (like seibei jumps) could go. But God knows how that could possibly be measured accurately.
Difference between FL and HDFL is basically that you can still see the note when you hit it, and do some "last second fine-tuning" if necessary, as well as having an easier time aiming correctly since you'll see the approach circle before reaching the note and can adjust accordingly.
I am not a fan of this ranking system, I recently have only gone down rank even when getting an S on a hard 5 star songs (I know the star system is not accurate) and even getting high combos, I got a 1100 combo on abstract nonsense witch can be improved but I got -rank for it. this has been happening to me a lot, I'm sure its because I do not understand the ranking system fully but any good or bad ranking system should NOT -rank when your learning a song that in my opinion is just stupid.
You're not losing pp for bad plays, your rank is falling due to other people passing you up. It doesn't have anything to do with the ranking system.[ Ian ] wrote:
I am not a fan of this ranking system, I recently have only gone down rank even when getting an S on a hard 5 star songs (I know the star system is not accurate) and even getting high combos, I got a 1100 combo on abstract nonsense witch can be improved but I got -rank for it. this has been happening to me a lot, I'm sure its because I do not understand the ranking system fully but any good or bad ranking system should NOT -rank when your learning a song that in my opinion is just stupid.
You could get a 10 thousand combo on a map, if it's 11 thousand in total, you won't get much pp. FC or nothing basically.[ Ian ] wrote:
I am not a fan of this ranking system, I recently have only gone down rank even when getting an S on a hard 5 star songs (I know the star system is not accurate) and even getting high combos, I got a 1100 combo on abstract nonsense witch can be improved but I got -rank for it. this has been happening to me a lot, I'm sure its because I do not understand the ranking system fully but any good or bad ranking system should NOT -rank when your learning a song that in my opinion is just stupid.
that's not exactly true, 10k/11k would give quite a big portion of the map's pp. possible 80-90% of it.
i got 2300/2600 combo on a map today and i got more pp from it than someone who fc'd it because i had better acc
i got 2300/2600 combo on a map today and i got more pp from it than someone who fc'd it because i had better acc
I'm like 99% sure the scalar for combo pp is (Your Max Combo^0.8) / (Maximum Combo^0.8). So getting 90/100 will actually grant you 92% of maximum pp if you had fc'ed it. (This is excluding other factors such as accuracy, # of misses). Someone can confirm.
The system is set to be more lenient the lower your combo. 90/100 will grant you 92% of pp, but getting say 50/100 would still grant you 57% of the pp. Of course, that's not ideal, but just for comparison's sake.
The system is set to be more lenient the lower your combo. 90/100 will grant you 92% of pp, but getting say 50/100 would still grant you 57% of the pp. Of course, that's not ideal, but just for comparison's sake.
I got over a 900 combo on a 1000 combo map with 1 100 and 2 misses in a row and got 214pp. An S with 1 100 gives 251pp...silmarilen wrote:
that's not exactly true, 10k/11k would give quite a big portion of the map's pp. possible 80-90% of it.
i got 2300/2600 combo on a map today and i got more pp from it than someone who fc'd it because i had better acc
Not complaining or anything, it just shows getting max combo is a pretty big deal if you want any pp.
i think that may have been because of the 2 misses aswell (misses drop your pp just for the sake of being misses, not just because they lower your acc and combo), i only had a sliderbreak myself.NixXSkate wrote:
I got over a 900 combo on a 1000 combo map with 1 100 and 2 misses in a row and got 214pp. An S with 1 100 gives 251pp...silmarilen wrote:
that's not exactly true, 10k/11k would give quite a big portion of the map's pp. possible 80-90% of it.
i got 2300/2600 combo on a map today and i got more pp from it than someone who fc'd it because i had better acc
Not complaining or anything, it just shows getting max combo is a pretty big deal if you want any pp.
Probably, not until hit object data or whatever it's called is available.Ivan wrote:
I don't think PP system is going to get a tweak or a change anytime soon ;(
i do this so often its hilarioussilmarilen wrote:
i got more pp from it than someone who fc'd it because i had better acc
https://osu.ppy.sh/b/297411 Is this accurate to what you planned for pp to be?
Rank 1 is 210pp, rank 2 is 450pp
Rank 1 is 210pp, rank 2 is 450pp
It's more accurate than your score being #1 on that map over a HD+DT play.those wrote:
https://osu.ppy.sh/b/297411 Is this accurate to what you planned for pp to be?
Rank 1 is 210pp, rank 2 is 450pp
mod score bonuses were decided a long time ago by a man who really doesnt even play osu so of course there are some issues
I actually, seriously wonder whether this was supposed to be a troll post or a jokethose wrote:
https://osu.ppy.sh/b/297411 Is this accurate to what you planned for pp to be?
Rank 1 is 210pp, rank 2 is 450pp
Yes.those wrote:
https://osu.ppy.sh/b/297411 Is this accurate to what you planned for pp to be?
Rank 1 is 210pp, rank 2 is 450pp
Even my own score on that map gives 240 pp - HR and DT change the map gameplay so drastically for the harder, it's no wonder a HDFL score gives much less since it's a nomod score with a decent bonus to aim.
pp takes actual map difficulty and gives points based on that, as you probably know.
pp takes actual map difficulty and gives points based on that, as you probably know.
I always seem to wonder about pp and there's something I can never quite shake. I don't think it exists but I always seem to end up wondering why, if the peak parts of a map are where the most reward comes from, don't you get rewarded pp if you can guarantee that the person FCed that part (or all of those parts) by looking at their maximum combo achieved. Say the map maximum combo is 1000 and the majority of the map is 2 star tier apart from one sections that is 200 combo into the song which plays like a 10 star map, you can make a little check to find the minimum combo needed to have FCed this part of the map. For this map, let's assume that the extreme hard part is 50 combo long, to ensure that the person FCed that section the player would need AT LEAST 800 combo. Obviously add an extra 10 or so combo to either side as entering and exiting hard sections can be hard in itself and viola, you have some VERY basic per hit object analysis.
Another example: map has a maximum of 800 combo, hard part is at 700 combo and lasts for 50 combo. The player would need a 700 + 50 + 20 combo minimum to ensure that they hit the hard section. 700 is the amount of combo required MAXIMUM for the player to have reached the start of the hard section, 50 is the note length of the hard section and 20 is a little insurance combo to make sure they managed to get in and out of it themselves. Obviously there would be a little more complexity to it when maps have multiple peak areas but this is just the basic idea.
Not sure how crazy this would be to implement but it doesn't sound easy.
Another example: map has a maximum of 800 combo, hard part is at 700 combo and lasts for 50 combo. The player would need a 700 + 50 + 20 combo minimum to ensure that they hit the hard section. 700 is the amount of combo required MAXIMUM for the player to have reached the start of the hard section, 50 is the note length of the hard section and 20 is a little insurance combo to make sure they managed to get in and out of it themselves. Obviously there would be a little more complexity to it when maps have multiple peak areas but this is just the basic idea.
Not sure how crazy this would be to implement but it doesn't sound easy.
just one of the many wonderous things that could happen when hit object data becomes available i guess
did you even read his post?
what jesus suggested means that checking the song from both ends to find combo values which guarantee that certain higher difficulty parts have been FC-d could be used to somewhat work around not having per hit object data, so that the weight of combo could be more accurate for different maps.
500 combo total, hard part from 200 to 250, that means you can't place a 300 combo chain on the map without including the hard part, thus achieving a 300 combo on that map could be handled by the algorythm as such (knowing it for granted, that you FC-ed the hard part) and valued accordingly.
what jesus suggested means that checking the song from both ends to find combo values which guarantee that certain higher difficulty parts have been FC-d could be used to somewhat work around not having per hit object data, so that the weight of combo could be more accurate for different maps.
500 combo total, hard part from 200 to 250, that means you can't place a 300 combo chain on the map without including the hard part, thus achieving a 300 combo on that map could be handled by the algorythm as such (knowing it for granted, that you FC-ed the hard part) and valued accordingly.
I asked Tom about this on ask.fm, how you can have a part of a song be really difficult while the rest is easy which ups the song difficulty while having the benefit of map length and able to get higher accuracy on the easier parts. The current method in place to try and fix this is high penalty for misses, which seems to work quite well.jesus1412 wrote:
I always seem to wonder about pp and there's something I can never quite shake. I don't think it exists but I always seem to end up wondering why, if the peak parts of a map are where the most reward comes from, don't you get rewarded pp if you can guarantee that the person FCed that part (or all of those parts) by looking at their maximum combo achieved. Say the map maximum combo is 1000 and the majority of the map is 2 star tier apart from one sections that is 200 combo into the song which plays like a 10 star map, you can make a little check to find the minimum combo needed to have FCed this part of the map. For this map, let's assume that the extreme hard part is 50 combo long, to ensure that the person FCed that section the player would need AT LEAST 800 combo. Obviously add an extra 10 or so combo to either side as entering and exiting hard sections can be hard in itself and viola, you have some VERY basic per hit object analysis.
Another example: map has a maximum of 800 combo, hard part is at 700 combo and lasts for 50 combo. The player would need a 700 + 50 + 20 combo minimum to ensure that they hit the hard section. 700 is the amount of combo required MAXIMUM for the player to have reached the start of the hard section, 50 is the note length of the hard section and 20 is a little insurance combo to make sure they managed to get in and out of it themselves. Obviously there would be a little more complexity to it when maps have multiple peak areas but this is just the basic idea.
Not sure how crazy this would be to implement but it doesn't sound easy.
Anyways, your idea still seems to be good in an ideal situation. It's a way of checking if the hard part was fc'd. It gets messy when there's multiple hard parts. Also, it's a bit sketchy if the hard part is near the start. If the song had 1000 combo and hard part was at 300 and went for 50, you'd need 350 combo to ensure that the hard part was done right but you can get that from the remaining 650 notes after the hard part.
I'm interested to see if there's a way to implement your idea with the current system
without hit object data there's no way to make this work well though because where the hard part is can't be accurately determinedDrezi wrote:
did you even read his post?
what jesus suggested means that checking the song from both ends to find combo values which guarantee that certain higher difficulty parts have been FC-d could be used to somewhat work around not having per hit object data, so that the weight of combo could be more accurate for different maps.
500 combo total, hard part from 200 to 250, that means you can't place a 300 combo chain on the map without including the hard part, thus achieving a 300 combo on that map could be handled by the algorythm as such (knowing it for granted, that you FC-ed the hard part) and valued accordingly.
there could potentially, MAYBE, be a way to determine what part of the map in terms of minutes/seconds contributes the most to the star difficulty but which hit objects it is and where in the max combo it occurs probably can't happen without the necessary information but that's just an educated assumption on my part feel free to shit on me tom.
really you just see red when you see my posts and sometimes you let it cloud your logic
Actually, in that case, you would need 700 combo in order to be absolutely sure that the hard part was done right, not 350 (calculated from the end towards the beginning). That way, you avoid the possibility of gettingRewben2 wrote:
... If the song had 1000 combo and hard part was at 300 and went for 50, you'd need 350 combo to ensure that the hard part was done right but you can get that from the remaining 650 notes after the hard part.
Rewben2 wrote:
that from the remaining 650 notes after the hard part.
Oh right, I'm dumb. I had a feeling there was something flawed with what I was writing >.>Gigo wrote:
Actually, in that case, you would need 700 combo in order to be absolutely sure that the hard part was done right, not 350 (calculated from the end towards the beginning). That way, you avoid the possibility of gettingRewben2 wrote:
... If the song had 1000 combo and hard part was at 300 and went for 50, you'd need 350 combo to ensure that the hard part was done right but you can get that from the remaining 650 notes after the hard part.Rewben2 wrote:
that from the remaining 650 notes after the hard part.
Regarding your suggestion, it might become feasible to code (but probably still not easy) if we use the combo colors. Considering the ranking of maps is getting more standardized, it'll be consistent for calculating these difficult parts.jesus1412 wrote:
Not sure how crazy this would be to implement but it doesn't sound easy.
The most difficult aspect though is, who's gonna sit around figuring out a system that will :
- Calculate the difficulty of each individual combo color segment
- Find the most difficult segment BEFORE you broke your largest combo (not overall in the map)
- (optional) Weight the rest of the segments relative to that segment so all segments sum to a pp value for SS
- Each individual combo color section broken/missed in their respective combo chunks further deducts from your final pp value
- If breaks combo and the combo is less than the number of notes in the particular combo segment, default that section to either 0pp value or some minimum pp
- (optional) pp Deductions become less severe the more your combo exceeds the combo of the most difficult section
- (optional) pp Deductions become more severe the more your combo precedes the combo of the most difficult section
- (optional) pp Deductions become based on the deviation between the most difficult segment and the others
- Curve pp value with your accuracy for only the sections that apply between the combo breaks
- This calculation will be done for each combo chunk, so if I missed/broke slider in the middle of a 1001 combo map, it will do calculations for the first 500 combo chunk and then a calculation for the other 500 combo chunk.
- 2 ways to deal with the final calculation :
- Sort chunks in descending order by value and weight each subsequent chunk after that (just like Top Plays) and the total pp after weighting will be the pp you earn
- Average all the chunks together, then scale the average down by the number of chunks you had
Sorry, I have a problem..
As I've red in FAQ, the number in brackets is how much did you get from this map. Well here it says that I got 29 pp, when actually It gave me only 5 - 6 pp
And I played this map only 2 times and didn't get any pp from it before. Help me, mb I don't understand something тт
As I've red in FAQ, the number in brackets is how much did you get from this map. Well here it says that I got 29 pp, when actually It gave me only 5 - 6 pp
And I played this map only 2 times and didn't get any pp from it before. Help me, mb I don't understand something ттWhen you set a score, your plays below it are pushed down and have lower weightings and contribute less towards your total pp.Apacci-k wrote:
Sorry, I have a problem..
As I've red in FAQ, the number in brackets is how much did you get from this map. Well here it says that I got 29 pp, when actually It gave me only 5 - 6 pp
The reason I suggested this is because star rating already finds the most difficult parts of the map, at least to my knowledge. Toms difficult calculator could generate graphs of the difficulty at certain times, hence why I thought the idea was feasible. Here's a picture of one of the graphs in question:Woobowiz wrote:
Regarding your suggestion, it might become feasible to code (but probably still not easy) if we use the combo colors. Considering the ranking of maps is getting more standardized, it'll be consistent for calculating these difficult parts.jesus1412 wrote:
Not sure how crazy this would be to implement but it doesn't sound easy.
The most difficult aspect though is, who's gonna sit around figuring out a system that will :But I suppose it's only theory
- Calculate the difficulty of each individual combo color segment
- Find the most difficult segment BEFORE you broke your largest combo (not overall in the map)
- (optional) Weight the rest of the segments relative to that segment so all segments sum to a pp value for SS
- Each individual combo color section broken/missed in their respective combo chunks further deducts from your final pp value
- If breaks combo and the combo is less than the number of notes in the particular combo segment, default that section to either 0pp value or some minimum pp
- (optional) pp Deductions become less severe the more your combo exceeds the combo of the most difficult section
- (optional) pp Deductions become more severe the more your combo precedes the combo of the most difficult section
- (optional) pp Deductions become based on the deviation between the most difficult segment and the others
- Curve pp value with your accuracy for only the sections that apply between the combo breaks
- This calculation will be done for each combo chunk, so if I missed/broke slider in the middle of a 1001 combo map, it will do calculations for the first 500 combo chunk and then a calculation for the other 500 combo chunk.
- 2 ways to deal with the final calculation :
- Sort chunks in descending order by value and weight each subsequent chunk after that (just like Top Plays) and the total pp after weighting will be the pp you earn
- Average all the chunks together, then scale the average down by the number of chunks you had