|
On October 07 2013 00:21 Teliko wrote:
Is that not how the current system works? I noticed that it's pretty much impossible to gain new ranks unless I'm using mods in v1, how does v2 differentiate from that?
I'll just dump this here. This is written by peppy, not me.
-----------------------------
Basic things to note about calculation method:
Scores:
* Accuracy based, not score. The main metric here is comparing your accuracy to the average achieved for that map. If you are above average, you will get pp. How far above average decides how much you get.
* Currently does not consider max combo (will likely be factored in).
* DoubleTime +14% (but caps at 100% accuracy), HardRock +8%, Flashlight +6%, Hidden +5%. Please don't bitch about these; subject to change.
* S/SS +2%, Perfect +1%
* Scores are still normalised in a similar fashion to ppv1 – see the wiki for details on that.
Beatmap Weighting:
* Weighted based on days since rank, where anything over than 1 year is worth ~10% of newer maps. Very recent ranks will get an initial low value until enough scores have been set on them. this is a logarithmic scale, more details to come.
* Weighted based on OD (overall difficulty). 70% for 1, 140% for 10, linear.
* Weighted based on CS (circle size). 80% for 1, 120% for 10, linear.
* Weighted based on ratio of perfect plays (no combo break). Kinda complex math, details coming later.
* Weighted based on average accuracy achieved by all plays, excluding outliers, relative to all other maps. This is kind of the telling factor, and works very well in practice.
I urge any criticism of these calculation methods to be based on very general assumptions, as nothing is in stone, and what I have said about is oversimplifying. Though I am trying to keep things as simple as possible.
---------------------
The most important part of ppv2 is that the calculation algorithm will be open, so we can criticize it on actual points rather than assumptions and guesses.
|
Awesome, thanks. Pretty strange having accuracy altered by mods so significantly, but we'll see how it actually turns out.
|
Not that it's super hard or anything, but I finally passed this song.
+ Show Spoiler +
Next goal: FC it.
|
weird, so it seems u can get pp for scores that aren't your top score?
|
On October 07 2013 01:10 Yuugo wrote:
The most important part of ppv2 is that the calculation algorithm will be open, so we can criticize it on actual points rather than assumptions and guesses.
Which is good, because until we get a concrete algorithm to analyze, we can't REALLY comment on things anyway. Well, we could but the comments would be invalid arguments since things are subject to change. The real important thing here is that the various weighting/bonus/etc. values work well together, and don't inadvertently introduce situations where specific combinations are either underrated or overrated by the system.
When the ppv2 algorithm is fully opened, I'm gonna see how much I can figure out about the interactions of the various factors.
|
My only problem on the given values here is the cap at 100%. The way it is written here it looks like it only affects DT, but it is still silly. This means that an 92% FC would weigh the same as an 86%. A 99% with one slider-100 would only be 2% better than that 86% score when both scores only use DT. That seems pretty unfair to me.
Oh well, the whole subject to change thing makes it feel like a waste of time commenting on it too much right now. We'll have to wait for the formula to be finalized to properly talk about it.
Oh and acie, it will still be counted for your best play on that map.
|
Ah Yuugo, funny to see you here, you spectated me! I wondered who you are!
I'm turning up the hate for ppv2 as well. Not liking it.
gj chimeri, actually took me a long time to do that song (longer than it shoudl have really)
+ Show Spoiler +
|
am i crazy or does it seem that hd+hr is easier than just hr?
|
Imo anything becomes stupid with hidden, the difficulty of hidden approaches the sum of n deltafunctions with n the number of stacked notes. But i think i'm quite extreme in that opinion. So i don't think it matters much. What could be the issue though is that one is better at 'tapping along with music' then 'tapping when the approach circle vanishes'.
Another issue
A few months ago i bought a tablet, and decided to give it a try. The result after a week was okayish, but when i wanted to switch back to mouse, i noticed (or so i thought) that i couldn't do anything with mouse anymore. So i decided to give tablet my all. Now, a few days ago, i for the first time ever, raged at a computer game, which caused me to switch back to mouse. And now i'm nearly back at my old level. I managed to get a A on aqo's 'let's go' (ar10), which is a 7% accuracy increase compared to when i switched to tablet. I can't stop thinking on how much better i would've been if i had just sticked with mouse.
Seriously, fuck tablets TT (nah, but it's how i feel atm). I guess i learned a lesson the hard way, losing 2 months
Also, i wanted to start anew so i created another account (although osu doesn´t like it), so i won't be on my old one anymore.
@monomo: Good job on nyaten! That map's a bitch. ):
edit2: fartownik noob  :
+ Show Spoiler +![[image loading]](http://osu.ppy.sh/ss/1026521) First time i beat someone top 100 :D
|
Daaaamn lol this map
I switched to tablet 1st of August (started playing 1st of feb) and really I love it - it made ar10 possible for me really, and keeping combo on any map is just impossible past 200 with mouse (really heave fatigue issues) for me
tl;dr
Tablet: huge success
|
On October 09 2013 03:31 Yorbon wrote:
Imo anything becomes stupid with hidden, the difficulty of hidden approaches the sum of n deltafunctions with n the number of stacked notes. But i think i'm quite extreme in that opinion. So i don't think it matters much. What could be the issue though is that one is better at 'tapping along with music' then 'tapping when the approach circle vanishes'.
the skill you need to improve for hidden is the ability to predict circle timing by watching when it appears (or disappears, but that gives you less time to process what to do). The idea is that you're always looking ahead of where you're actually aiming.
Also, get a skin with semi-transparent circles for hidden. Semi-transparent circles make stacked hitcircles a joke compared to standard skins.
It seems to me that many people actually use the approach circles as their main visual timing cue. IMO that is a mistake, since you can get the same timing information by a combination of the approach rate and the time a circle first appears. If you take the approach of always being visually ahead of where you're aiming (basically following the hitcircles as they appear) the difference between hidden and nomod becomes much smaller. I won't go into any more detail, unless someone actually cares enough to ask.
|
Man it's been awhile since I played. But the past few months I started playing again, I think starting from early August. Managed to get quite a speed boost, one pass on Paraoka - Boot in particular that just felt awesome. To pass that song felt like such an improvement considering I've been wanting to pass that song for fucking months. Nuff' talk, more screenshots.
+ Show Spoiler +
I basically only play for speed up(as you can probably see judging by the huge amount C's and D's I'm posting) until I can pass the very very hardest songs in Osu, which is quite a ways to go.
|
Bump! Holy moly got one of those passes that shouldn't have been, check out the health bar, pretty much dead most of the time:D
+ Show Spoiler +
I was also livestreaming at the time so I got a VOD of it too if you wish to check it out, FPS is kind of crappy though.
http://www.twitch.tv/tenoxsc2/c/3073485
|
I hate that map, took me several trys to get it done... almost deleted it if i weren't a vocaloid pervert fanboy
|
On October 11 2013 15:22 Bobbias wrote:Show nested quote +On October 09 2013 03:31 Yorbon wrote:
Imo anything becomes stupid with hidden, the difficulty of hidden approaches the sum of n deltafunctions with n the number of stacked notes. But i think i'm quite extreme in that opinion. So i don't think it matters much. What could be the issue though is that one is better at 'tapping along with music' then 'tapping when the approach circle vanishes'.
the skill you need to improve for hidden is the ability to predict circle timing by watching when it appears (or disappears, but that gives you less time to process what to do). The idea is that you're always looking ahead of where you're actually aiming. Also, get a skin with semi-transparent circles for hidden. Semi-transparent circles make stacked hitcircles a joke compared to standard skins. It seems to me that many people actually use the approach circles as their main visual timing cue. IMO that is a mistake, since you can get the same timing information by a combination of the approach rate and the time a circle first appears. If you take the approach of always being visually ahead of where you're aiming (basically following the hitcircles as they appear) the difference between hidden and nomod becomes much smaller. I won't go into any more detail, unless someone actually cares enough to ask. Well, i guess i just have an inappropriate skin. I don't have any trouble at all with hidden, except for stacked notes (hence the delta functions, although of course that's a bit of an exaggeration). So, (with my skin) it becomes boring very easily, much like how airman becomes boring: 90% of the song is ridiculously easy. The other 10% is too easily combobreakable: you're not playing it for high score because of that, but it's not worth practicing either, because 90% is too easy. There is just no point in playing it. The difference is of course that, once you've attained such a level, airman becomes just another map, while stacked notes with hidden (without an appropriate skin) will never get easy. You just won't see them.
(yes, i like parentheses)
|
On October 13 2013 06:15 Yorbon wrote:
(yes, i like parentheses)
I too like parentheses.
And yeah, you really should try even just making an alternate version of your skin with modified alpha. I was amazed at how much easier it made stacked notes in hidden. Of course, the downside is that the circles blend into the background more, so if you pay without dimming the BG, that can make things harder. I replaced all my backgrounds with a black jpg file before the option to dim BGs was added, because I had such a hard time reading semi-transparent notes on anything but a black BG... But I'm weird so I don't think most people would have such a hard time.
|
In fact, I just noticed you using a non-existant object to describe the difficulty of stacked notes in hidden.
What does that mean, philosophically?
|
On October 14 2013 17:46 monomo wrote:
In fact, I just noticed you using a non-existant object to describe the difficulty of stacked notes in hidden.
What does that mean, philosophically? I doubt the difficulty of stacked notes in hidden does exist (or can be defined consistently), so i don't think there's a problem there.
I do think it makes sense though. Suppose there is a function f(t) that describes the difficulty of a map as a function of time given a certain skill set. This function should be discreet, but i'll just assume some continuous approximation of it. The estimated score (disregarding combo) should be of the form Score1=Z*integral(f(t)dt) with Z some constant so that this integral equals the maximum score when f(t)=1 for all t. Now, a mod will increase (or decrease) the difficulty so the score changes, let's say like this: ScoreMod=Z'*integral(f(t)g(t)dt) with g(t) the effect of a mod on the achieved score (again, given a certain skill set). For example, the mod easy is widely regarded as unplayable, so g(t) is approximately 0 at every t given (un)certain skill sets. As for hidden, i said the difficulty of hidden approaches n deltafunctions with n the number of stacked notes. This means (to me) that the effect of hidden on the score is very large at all stacked notes except for the first in a series, for obvious reasons. At all other times, the effect on the score approaches zero. g(t) = 1- sum(delta_1/a (x-x_i)) with i the summation index, and a an arbitrary number a>>0. delta_b (x) = 1/(b*sqrt(pi))*exp(-x^2/b^2). That's why the word 'approaches' was important in my assessment of hidden difficulty.
Or maybe i could've said that g(t) is 1 for all t except at t=stacked note, where it's 0. Seems a lot shorter and a lot less iffy. But that didn't cross my mind at the time of writing 8)
Aside from that, i feel talking about the existence of mathematical objects in itself is absurd.
/offtopic
edit: btw, i am a mere astronomer, not a mathematician, so my story may be totally off, in mathematical terms. There is a reason mathematicians cringe whenever an astronomer tries to prove his equations used, huehuehue. (at least at our college)
|
/offtopic
sorry, didnt read, just came by to state that "delta functions" don't exist
EDIT: I got lvl 100 just now! I WANT A COOKIE!
|
On October 14 2013 19:34 monomo wrote:
/offtopic
sorry, didnt read, just came by to state that "delta functions" don't exist
EDIT: I got lvl 100 just now! I WANT A COOKIE! TT
Just when I felt this good about writing such a long post.
|
|
|
|
|
|
EPT![[image loading]](http://i.imgur.com/MfjzVYH.jpg)
![[image loading]](http://i.imgur.com/BGkz6xS.jpg?1)
![[image loading]](http://i.imgur.com/AEIMbIU.jpg)
![[image loading]](http://i.imgur.com/pHnOeBf.jpg)
![[image loading]](http://i.imgur.com/FjjjyvE.jpg)
![[image loading]](http://i.imgur.com/f9F1KxS.jpg)
![[image loading]](http://i.imgur.com/XZJABnL.jpg)
![[image loading]](http://i.imgur.com/cr2IJ9p.jpg)
![[image loading]](http://i.imgur.com/TSjVLaq.jpg)
