Precision is broken though - it pretty much only takes into account CS. It makes no sense for airman HR to give the same aim as airman nomod, and airman HR to give the same precision as some random TV size with HR.
forum
osu!Skills
posted
Total Posts
1,732
Why is HD highly underweighted, when it goes to Reaction?
Comparing mine and other people FCs' I noticed that the same AR10.87 scores what are worth about 884 reaction points with HR,DT, are worth only 816 reaction points with HD,HR,DT.
The same with playing AR11: the very top people in Reaction ranking have +HR,DT plays and avoid HD on purpose, to farm more points.
Could someone explain why playing without Hidden requires much more reaction than while playing with HD?
Comparing mine and other people FCs' I noticed that the same AR10.87 scores what are worth about 884 reaction points with HR,DT, are worth only 816 reaction points with HD,HR,DT.
The same with playing AR11: the very top people in Reaction ranking have +HR,DT plays and avoid HD on purpose, to farm more points.
Could someone explain why playing without Hidden requires much more reaction than while playing with HD?
In HD the note fades in much faster than no mod, so you see it sooner. Same with mods. This makes HRDT harder to react to than HDHRDTScarlet Evans wrote:
Could someone explain why playing without Hidden requires much more reaction than while playing with HD?
With HD enabled, objects are fully faded in after 40% of the AR window (so on ar10 with 450ms, it's faded in after 180ms). Without HD enabled, fade-in time is a static 400ms.
But in both cases approach time, the real time you have to react, is 450 ms, right? Isn't taking fading into account by player more like memorisation and predicting when you should click the circle, while the time you have to react is the same? Or I don't understand the subject properly? I mean, even if it fades faster, it appears in the same time, so you have the same time to react, right?
Nope.Scarlet Evans wrote:
But in both cases approach time, the real time you have to react, is 450 ms, right? Isn't taking fading into account by player more like memorisation and predicting when you should click the circle, while the time you have to react is the same? Or I don't understand the subject properly? I mean, even if it fades faster, it appears in the same time, so you have the same time to react, right?
In the example provided:
so on ar10 with 450ms, it's faded in after 180msIt's fully faded in at 180ms, leaving you with 270 ms reaction time requirement on HD assuming you suddenly see the object at 100%. Code uses somewhere between 40-60% iirc (am on mobile so will need to check, so will use 40% for this example). That means that the object is 40% faded in at 72ms giving us with a 378ms reaction time requirement. On no mod, the object is fully faded in at 450ms, meaning it's 40% faded in at 180ms, giving us a 270ms reaction time requirement.
40% fade-in @ NM: 270ms reaction time
40% fade-in @ HD: 378 ms reaction time
The sooner you see the circle the sooner you have to realize where it is located, the more time you have to react and move the cursor there. While both start fading in at the same time, HD fades in faster, allowing you to percieve the note sooner.
Interesting.. I didn't know that... I always thought there was another reason for HDHR having rewarded less than HR...
http://osuskills.tk/user/DavidStylez this guys' hacked scores need to be cleared or smth
What do you mean by NoMod circles fading away? I never noticed anything like that :Oabraker wrote:
Nope.Scarlet Evans wrote:
But in both cases approach time, the real time you have to react, is 450 ms, right? Isn't taking fading into account by player more like memorisation and predicting when you should click the circle, while the time you have to react is the same? Or I don't understand the subject properly? I mean, even if it fades faster, it appears in the same time, so you have the same time to react, right?
In the example provided:so on ar10 with 450ms, it's faded in after 180msIt's fully faded in at 180ms, leaving you with 270 ms reaction time requirement on HD assuming you suddenly see the object at 100%. Code uses somewhere between 40-60% iirc (am on mobile so will need to check, so will use 40% for this example). That means that the object is 40% faded in at 72ms giving us with a 378ms reaction time requirement. On no mod, the object is fully faded in at 450ms, meaning it's 40% faded in at 180ms, giving us a 270ms reaction time requirement.
40% fade-in @ NM: 270ms reaction time
40% fade-in @ HD: 378 ms reaction time
The sooner you see the circle the sooner you have to realize where it is located, the more time you have to react and move the cursor there. While both start fading in at the same time, HD fades in faster, allowing you to percieve the note sooner.
I just tried to compare some AR0 + HT plays, using CE's [5] frame-by-frame speed key, with HD and without it and I fail to notice anything other than nomod circle disappearing... after it's clicked or missed, which doesn't affect reaction at all (because that happened in past).
For me, from what I see, the circle and approach circle both appear exactly at the same time on screen and the circle persists, until it's missed or clicked. Nothing fades away. At least on CuttingEdge.
Both on NoMod and HD circle appears not immediately, but a little faded, however it's really a tiny amount of time and I don't see a time difference in how long it takes for NoMod and Hidden circles to reach their normal state and stay on screen for these XXX milliseconds.
In other words, you start seeing HD and NoMod circle at the same time. The difference is that you see NoMod all the time, while HD starts to fade away, but it doesn't really affect your reaction. You already registered that there is a circle, both with HD and without it, within the exact time brackets. Now, with HD, it's more like a memory that "there is a circle here" from now on, rather than an actual change in the reaction time needed, because the Approach Time is not affected by HD in any way.
In other words, there's "an animation time" and the difference is that with HD it still happens and takes the same time, but it's just faded away and invisible for us. Nevertheless, it was no less or more visible for us for a certain extend of time and it appeared at the same moment at which NoMod circle would appear, so there's no really any difference in reaction time needed.
(1) Circle appears on screen, at the same time on HD and NM;
(2) Now, you perceive the circle and your time to react starts here;
(3) If it's Hidden, then circle starts to fade-in after some timespan;
(4) You reach the Perfect Click moment and your Reaction Time ends here. After that you can still get 300,100 or 50, but you clicked too late.
(5) Time you had to react is the one that lapses between (2) and (4), which isn't really affected by HD.
This is how I see this. But of course, I can be terribly wrong, as with anything I ever say
So, I'd really love you to explain more what is this NoMod fading-in that you say about
Fade-in: The circle first appears transparent, then it's opacity gradually increases over time.Scarlet Evans wrote:
What do you mean by NoMod circles fading away? I never noticed anything like that :O
I just tried to compare some AR0 + HT plays, using CE's [5] frame-by-frame speed key, with HD and without it and I fail to notice anything other than nomod circle disappearing... after it's clicked or missed, which doesn't affect reaction at all (because that happened in past).
For me, from what I see, the circle and approach circle both appear exactly at the same time on screen and the circle persists, until it's missed or clicked. Nothing fades away. At least on CuttingEdge.
Both on NoMod and HD circle appears not immediately, but a little faded, however it's really a tiny amount of time and I don't see a time difference in how long it takes for NoMod and Hidden circles to reach their normal state and stay on screen for these XXX milliseconds.
In other words, you start seeing HD and NoMod circle at the same time. The difference is that you see NoMod all the time, while HD starts to fade away, but it doesn't really affect your reaction. You already registered that there is a circle, both with HD and without it, within the exact time brackets. Now, with HD, it's more like a memory that "there is a circle here" from now on, rather than an actual change in the reaction time needed, because the Approach Time is not affected by HD in any way.
In other words, there's "an animation time" and the difference is that with HD it still happens and takes the same time, but it's just faded away and invisible for us. Nevertheless, it was no less or more visible for us for a certain extend of time and it appeared at the same moment at which NoMod circle would appear, so there's no really any difference in reaction time needed.
(1) Circle appears on screen, at the same time on HD and NM;
(2) Now, you perceive the circle and your time to react starts here;
(3) If it's Hidden, then circle starts to fade-in after some timespan;
(4) You reach the Perfect Click moment and your Reaction Time ends here. After that you can still get 300,100 or 50, but you clicked too late.
(5) Time you had to react is the one that lapses between (2) and (4), which isn't really affected by HD.
This is how I see this. But of course, I can be terribly wrong, as with anything I ever say
So, I'd really love you to explain more what is this NoMod fading-in that you say about
Fade-out: The circle, after some time, starts losing opacity over time. This only happens with HD.
Comparing a recording of the game screen at 120fps, on the same map (Cheatreal, AR10+DT), both HD and no-mod appear at the same time with the same opacity (barely visible), but the speed it gains opacity during the fade-in with HD is higher than no-mod.
I think I've got it, thanks!
But how do you judge at which moment of fading in player notices the circle? People can actually perceive it a little faster, going for 100% of visibility can be not fully fair.
edit: I didn't noticed that what an online dictionary showed me in summary was definition of just fading, but under the name of fading-in, not fading-in, my bad ;_;
But how do you judge at which moment of fading in player notices the circle? People can actually perceive it a little faster, going for 100% of visibility can be not fully fair.
edit: I didn't noticed that what an online dictionary showed me in summary was definition of just fading, but under the name of fading-in, not fading-in, my bad ;_;
just wondering, now that Hatsune Miku - Kouon Chu Oniki Test is loved-- once adding maps of the category to osu!skills would the 0108 diff be worth tons of stamina due to having an infinity (or 2,949,075? not sure if it's 1/65535 or 1/∞) BPM quad in it? or is it not long enough to count
I have ran that map through the algorithm, and it's fine.unko wrote:
just wondering, now that Hatsune Miku - Kouon Chu Oniki Test is loved-- once adding maps of the category to osu!skills would the 0108 diff be worth tons of stamina due to having an infinity (or 2,949,075? not sure if it's 1/65535 or 1/∞) BPM quad in it? or is it not long enough to count
omg do isogu 290bpmabraker wrote:
I have ran that map through the algorithm, and it's fine.unko wrote:
just wondering, now that Hatsune Miku - Kouon Chu Oniki Test is loved-- once adding maps of the category to osu!skills would the 0108 diff be worth tons of stamina due to having an infinity (or 2,949,075? not sure if it's 1/65535 or 1/∞) BPM quad in it? or is it not long enough to count
Can you also make a nifty tool so people can run unranked maps through the aalgorithms
Such tool already does exist, and it's what we use in the back end. It's up to Kert if he wants to release it though.Dawnsday wrote:
Can you also make a nifty tool so people can run unranked maps through the aalgorithms
Right,
So I haven't been on osu!skills in a while, and I wanted to check my user stats
I went onto the site and typed in "Karma", expecting to see my profile, I didn't I saw some Italian with the username KaRmA.
Odd.
So I try my old username I changed a few months ago which was "SparKticas"
TL;DR - If i try search my name Karma, I dont find me, I have to search my old username.
Does this mean that it's not updated?
So I haven't been on osu!skills in a while, and I wanted to check my user stats
I went onto the site and typed in "Karma", expecting to see my profile, I didn't I saw some Italian with the username KaRmA.
Odd.
So I try my old username I changed a few months ago which was "SparKticas"
TL;DR - If i try search my name Karma, I dont find me, I have to search my old username.
Does this mean that it's not updated?
I'll tell kert to update itKarma wrote:
TL;DR - If i try search my name Karma, I dont find me, I have to search my old username.
Does this mean that it's not updated?
https://osu.ppy.sh/ss/6639910
so whilst this map was qualified i set the highest score ( the 88% one) and when it got ranked i decided to submit it to osuskills seeing as i'd thought it would count, it didn't. i then make a few other scores (as seen in the screenshot), submitting all of them. none have yet to submit? what's going on
so whilst this map was qualified i set the highest score ( the 88% one) and when it got ranked i decided to submit it to osuskills seeing as i'd thought it would count, it didn't. i then make a few other scores (as seen in the screenshot), submitting all of them. none have yet to submit? what's going on
Have you tried manually adding them?Chemistry wrote:
https://osu.ppy.sh/ss/6639910
so whilst this map was qualified i set the highest score ( the 88% one) and when it got ranked i decided to submit it to osuskills seeing as i'd thought it would count, it didn't. i then make a few other scores (as seen in the screenshot), submitting all of them. none have yet to submit? what's going on
yesabraker wrote:
Have you tried manually adding them?Chemistry wrote:
https://osu.ppy.sh/ss/6639910
so whilst this map was qualified i set the highest score ( the 88% one) and when it got ranked i decided to submit it to osuskills seeing as i'd thought it would count, it didn't. i then make a few other scores (as seen in the screenshot), submitting all of them. none have yet to submit? what's going on
Did you use ctrl+F to search? This is on your agility:Chemistry wrote:
yesabraker wrote:
Have you tried manually adding them?
So you are saying the value should be lower?unko wrote:
um, this map with DT is ridiculously overweighted for stamina, so much so that i can only put it to some sort of error/mischief in calculation
it's ranking with 300bpm deathstream maps (804 points for 4xmiss), yet it's only 234... is it perhaps the doubles?
very, it's valued higher than reunion (15 second 276bpm stream section, 802 points for 0xmiss), fascination maxx (long 300bpm stream section, 846 for 1xmiss), and about the same as gekishou (65 note 300bpm stream, 897 for 0xmiss)abraker wrote:
So you are saying the value should be lower?unko wrote:
um, this map with DT is ridiculously overweighted for stamina, so much so that i can only put it to some sort of error/mischief in calculation
it's ranking with 300bpm deathstream maps (804 points for 4xmiss), yet it's only 234... is it perhaps the doubles?
native faith's deathstream is about 40 notes longer and 25bpm higher, yet it's worth over 100 points less for an fc if i'm right
Yeah, but they don't have 1/4 stream jumps. I'm not sure how that factors in though.
it updated cause i got a better score on it, https://osu.ppy.sh/ss/6655821
my stamina hasn't gone up since though, 542 before (this score is 600 stamina) still 542
my stamina hasn't gone up since though, 542 before (this score is 600 stamina) still 542
I think it's because the circles weren't hidden enough by the darkness-Jukke- wrote:
there is definitely something wrong with the memory stat
in my room FL is WAY harder than any of those and took me way more retries than all the others combined
(Sight reading)
Even if it was ar11 with 800bpm a stream with fl doesn't count much as long you can sight read
I found out about many useful things and some game mechanics perks here, so forgive me that I will ask one more thing that keeps me wonder =-) [I am HD player, just in case this matters]
Playing very high AR, I find it much easier to hit the sliders. I don't really mean slider's leniency, which don't really punish you for bad acc on score v1, but I have an impression that they are literally kind of easier to react to.
Is this just an impression or there is something behind this, like a slightly more time to react for sliders?
Or maybe I just perceive them easier because they are bigger and more visible, while there's nothing what technically makes them easier?
It really makes me curious
Playing very high AR, I find it much easier to hit the sliders. I don't really mean slider's leniency, which don't really punish you for bad acc on score v1, but I have an impression that they are literally kind of easier to react to.
Is this just an impression or there is something behind this, like a slightly more time to react for sliders?
Or maybe I just perceive them easier because they are bigger and more visible, while there's nothing what technically makes them easier?
It really makes me curious
I noticed this too, but have not really looked much into it. I will see what that is about when I get a chance to.Scarlet Evans wrote:
Is this just an impression or there is something behind this, like a slightly more time to react for sliders?
I have not found any fade in variation between hitcircles and sliders, so the only thing left is to suspect that we better react to moving objects. Having snaking sliders enabled will do this.abraker wrote:
I noticed this too, but have not really looked much into it. I will see what that is about when I get a chance to.Scarlet Evans wrote:
Is this just an impression or there is something behind this, like a slightly more time to react for sliders?
Would be nice if i could submit a score that wasn't my best on a map...
You canLick wrote:
Would be nice if i could submit a score that wasn't my best on a map...
You need to use "Manual score input" on your profile page, it's located on player's profile, little below the scores:http://puu.sh/sOkx9/2f1a3d4f3a.png
Now, you need to write in ID of map or link of it. For example, for
Renard - Love Song [insane]
you will need to paste / write either " https://osu.ppy.sh/b/211776 " or just "211776".
The number after /b/ in link is beatmap ID. Don't mistake it with ID or link to mapset, which in this case is https://osu.ppy.sh/s/74617
Also, make sure that you enter correct ID. It happened to me few times that I entered ID of the wrong difficulty. In this case, you just need to enter it again, but this time a correct one^^.
Personally, I am addicted to this button. I don't know, if it's supposed to refresh only from time to time or just once upon visiting your profile, then the next refresh is after X time, but when I play something and I did good, then even if it's within top #1000 on map or top #100 in my scores, I want to see how it affects my profile quickly.
So, I refresh it few times and upon not seeing any change, I just add a play manually. Sometimes I do this quite many times a day
AFAIK it's a queue. So as soon as its turn comes in the list of requests, it updates.Scarlet Evans wrote:
I don't know, if it's supposed to refresh only from time to time or just once upon visiting your profile, then the next refresh is after X time,
Sometimes it just randomly updates old scores too...abraker wrote:
AFAIK it's a queue. So as soon as its turn comes in the list of requests, it updates.Scarlet Evans wrote:
I don't know, if it's supposed to refresh only from time to time or just once upon visiting your profile, then the next refresh is after X time,
How's that reading skill coming?
I wish I can say it's close to done or at least an estimation to when it might be. All I can say is what I am currently doing, and that is an algorithm for circle and slider overlap calculation.[Toy] wrote:
How's that reading skill coming?
Topic Starter
You should get more scoresGudafild wrote:
How does this work?
It's weighted mean, where the weights are:Gudafild wrote:
How does this work?
1,
0.95,
0.95^2,
0.95^3,
...
0.95^{N-1},
for N being number of your scores and 0.95^x is exponentiation.
Theoretically it works like this: all score values are sorted (descending from the highest one to the lowest one), then they are multiplied by their weights and divided by the sum of weights. We are assuming here that aside of your achieved scores, you also have infinitely many scores that are equal to 0.
In this case, sum of weights is 20, so we can simply say that N-th score is takes as: (value*0.95^{N-1})/20.
You sum up these for all your scores and you have your total skill (✿◠‿◠)
For a little insight about how the first scores impact the total value, check this, it can be useful:
p/5306937
Edit: In your case, you have just a few of the highest scores, which are not enough to overweight the rest. The first top six scores, even though they are much higher than the rest, have a total sum of weights ~26.5%, and give you a total of about ~185.42 Agility, so the total score depends mostly on the lower ones. We can say that not having enough of such high Agility scores drags you down a little.
If all your lower scores were higher, for example like this 601 one, your Agility would be much higher and equal to about 627.