Guest
home
beatmaps
rankings
community
store
help

Sign In To Proceed 435v5c

I've forgotten my details

Don't have an ? 473g1v

osu! to create your own !
forum
  1. Forums
  2. osu!
  3. Development

osu!Skills 2u4d3c

posted
Total Posts
1,764
Kert
2,079 posts
ed June 2009
Topic Starter

Hey there!
Once upon a time I got an interesting idea of a system parallel/modification to Performance Points.
So I made a website for this too

https://osuskills.com

More screenshots











Skill explanations for clarification:

Stamina - Tapping stamina
Tenacity - Streaming
Agility - Raw Aim
Precision - Ability to hit small circles
Reaction - Sight-reading
Reading - Aim related map reading
Memory - Flashlight performance
Accuracy - Rhythmic accuracy

The player also receives a"title" (or multiple titles) as a fun factor depending on the set of his best skills.

Future plans
  1. Reading skill
  2. Reaction skill - added!
  3. Memory skill - added!
  4. Player title badges - added!
  5. Overall skill-based rankings - added!
  6. Night mode - added!
  7. Mobile / tablet - added!
  8. Personal bars
Notice that only osu!standard is ed right now.

If you want to help us you can check and participate in some of the related discussion threads
Best FL plays

Or you can improve code and calculations at:
https://github.com/Kert/osuSkills - calculator code
https://github.com/Kert/osuSkillsGUI - GUI for easier development

Developers: Rust_omg
Last edited by Kert , edited 42 times in total.
Halogen-
Former osu!mania 4K Champion: 2015
574 posts
ed September 2009
While this is a really cool idea in concept, the execution of this is really difficult and much like performance points, highly subjective.
Sync
osu! Alumni
1,575 posts
ed April 2011
Interesting concept. Is it usable or testable yet?
TicClick
Global
4,442 posts
ed February 2011
сделай тоггл, чтобы делил на двадцать
Diony
86 posts
ed April 2014
Would be awesome if executed properly.
Kert
2,079 posts
ed June 2009
Topic Starter

Sync wrote: b3o4g

Interesting concept. Is it usable or testable yet?
I'll try to make a test app by using existing top50 scores or something like that.
It should be available next week I hope.
Reiko
59 posts
ed December 2011
gl on this! If its done well it can become quite usefull for me and others as well. :)
Setsuna
216 posts
ed December 2013
Cool concept! Quick question, do the graphs scale in proportionate to the player's rank? For example, what if I were to FC a stream map that is only regarded as 5.1 stars while my friend streams a map that is regarded as 6.4 stars. If we were to both have 99%~ish accuracy and a full combo, would we both have (I'm just putting in a random number here) 700 tenacity/stamina or will he have a higher score since it was a harder map?
Kert
2,079 posts
ed June 2009
Topic Starter
Really depends on the streams. For longer/faster streams you get a higher score.
In your example if he'd had misses his score could be lower than yours (obviously)

Meanwhile we're getting a website soon~~~
Multtari
88 posts
ed September 2011
Really interested in seeing what your system thinks about different players skill sets. Keep it up!
makemecool123
150 posts
ed November 2013
Nice i want to test it
PatZar
3,079 posts
ed March 2013
very nice concept, let's see for future
Isla-
77 posts
ed March 2014
Can't wait!
Poku
63 posts
ed September 2013
If it will be executed correctly, would be really nice and I hope we can see it someday, thanks for sharing~! :D
Kert
2,079 posts
ed June 2009
Topic Starter
Good news! Everything for the first release is ready finally!
Just need a bit of time getting and ing to some hosting service.
SPOILER

I didn't expect to spend a whole week porting stuff to *nix D:
Kert
2,079 posts
ed June 2009
Topic Starter
And here's our first release!

http://osuskills.tk

I also updated the main post, don't forget to check it out~
Loneliness
18 posts
ed October 2013
Классно сделали, жду доработки :^)
Nelly
Beatmap Nominator
1,536 posts
ed August 2014
So it will calculate if you reach top 50?
Он только обновит мой результат если я попадут в топ 50?
Gumpy
2,418 posts
ed June 2014

Nelly wrote: 215p57

So it will calculate if you reach top 50?
No it uses your top 50 scores.

wow I can't read
Last edited by Gumpy , edited 1 time in total.
KRZY
Former osu! Champion: OWC#3
4,848 posts
ed May 2009
Interesting, perhaps add a "compare two s" function?
Kert
2,079 posts
ed June 2009
Topic Starter
That's totally not a bad idea hehe
0120
Former osu! Champion: 2015
82 posts
ed September 2012
Interesting concept, there is room for improvement but its fun to look at all these things.
Btw, you might wanna fix this top map cause im pretty sure its a bug of some sort, and i fear if there is one like this, theres probs more...
Keep it up!
IamBaum
322 posts
ed August 2013
This looks really nice, do you also have any plans to include more game modes in the future?
Sieg
3,981 posts
ed February 2012
So, it calculates only maps where I'm in top 50?
Multtari
88 posts
ed September 2011

Sieg wrote: 565dp

So, it calculates only maps where I'm in top 50?
Apparently since you can't just poop out correct values without API info from the plays which is limited to top 50.


It only seems to have calculated my very old top 50 scores but not newer ones. Bug or are the scores still being processed?
Yousei
70 posts
ed October 2013
What if it could calculate other mode...
Ciuriya
61 posts
ed September 2014
This is a nice concept, but why aren't you using /api/get__best from peppy's api?
It can get a player's top 100 best plays which, imo, is a much better indication of skill than top 50 scores.
Kert
2,079 posts
ed June 2009
Topic Starter

Multtari wrote: 263o32

It only seems to have calculated my very old top 50 scores but not newer ones. Bug or are the scores still being processed?
Currently we're using a bit dated Top 50 scores database, that is why. I'll make an updater soon

Smc wrote: 1h2m21

This is a nice concept, but why aren't you using /api/get__best from peppy's api?
It can get a player's top 100 best plays which, imo, is a much better indication of skill than top 50 scores.
Because this is not based on PP at all
I'll think about adding these results there too though
qoot8123
1,500 posts
ed March 2011

IamBaum wrote: 6j5u3m

This looks really nice, do you also have any plans to include more game modes in the future?
same here. hope it will have other game modes
Flanster
5,938 posts
ed August 2010
>.tk

Good stuff you got goin but... how accurate can this be?
[224]Reol
Former osu!catch Champion: 2015
64 posts
ed September 2013
how think about CTB?
im ctb
pls call me if you want advice haha

i waiting ctb OwO
[224]Reol
Former osu!catch Champion: 2015
64 posts
ed September 2013

how about change interface? :D
A... pentagon... line... color... *&%#
i think that... this interface seems more appropriatede
Trent
osu!catch Champion
58 posts
ed October 2013
i like 224's idea for interface.
Nelly
Beatmap Nominator
1,536 posts
ed August 2014

224dj wrote: 146668


how about change interface? :D
A... pentagon... line... color... *&%#
i think that... this interface seems more appropriatede
!
Jing Yuan
1,500 posts
ed September 2013
having a space in your name generates an error and _ must be used.

edit: im not sure if it occurs only on first time loading of the name, but there was an error screen that showed up when i typed in my name with a space the first time.


It's working fine now with a space in it.
Endaris
5,282 posts
ed June 2010

224dj wrote: 146668

A... pentagon... line... color... *&%#
i think that... this interface seems more appropriatede
I think this is a stupid idea. This kind of graphic approach only makes sense if you can assume correlations between neighbored skills and exclude the existence of correlations for non-neighbored skills.
This isn't some Pokemon-game where you can artificially push the stat for your Pokemon's coolness and where it doesn't matter that neighboring stats don't have a correlation.
And if we don't want to assume easy correlations like this we should either make it more simple or more complicated than this.

Oh no, abraker is on this *runs* :D
abraker
Global
8,355 posts
ed July 2014

Endaris wrote: 6j5f7

Oh no, abraker is on this *runs* :D
I work behind the scenes, advising Kert so that everything makes sense as well as fix bugs here and there. It's up to Kert's final decision to use my ideas or not.
Granger
835 posts
ed April 2014
Well, it seems like Agility is about every top ranked player's best skill... Not sure what i expected.
B1rd
2,974 posts
ed December 2013
Rucker is better than Hvick in every attribute but aim.
Minhtam
550 posts
ed September 2014
Well, gee. Either you need to change up some of your variables or I am the most imbalanced player in the game.

Kert
2,079 posts
ed June 2009
Topic Starter
Don't worry, this will be tweaked appropriately
The Gambler
1,307 posts
ed September 2013
Can we include people that are not in the top 50?

For some reason this is what mine looks like...
Endaris
5,282 posts
ed June 2010

The Gambler wrote: 1483t

Can we include people that are not in the top 50?

For some reason this is what mine looks like...
Get good
It's about Top50 scores on a map, not about top50 players. Seemingly you never got on a scoreboard. http://osustats.ppy.sh/
Kert
2,079 posts
ed June 2009
Topic Starter
Top 100 scores are ed now!
Endaris
5,282 posts
ed June 2010
Nice.
The Accuracy stuff looks really weird though.

I figured out:
The number provided when clicking on the detailed overview for your accuracy-rating is the maximum amount of points assigned to the map. This number seems to cap at 795 for OD10.
There is some sort of weightage for the scores as my final points are lower than my best result(I know it cause SS).
A maximum of 100 maps is shown in the detailed overview.
OD has a heavy influence on Accuracy points(Up to ~20% increase) while vision-impairing mods (HDFL) both don't.
Qualified maps aren't included in the search.
Actual pattern difficulty isn't as important for the assigned number compared to OD:
OD -- maximum number assigned
OD10 795
OD9.8 750(?)
OD8.4 720(?)
OD8 712(?)
OD7 685(?)

Actual pattern difficulty doesn't even matter(X-Encounter is confirmed to be sub-770).
Roxy Lalonde
100 posts
ed June 2014
Can we get formal definitions on what these mean?



I can't make out anything of how these are calculated when I have the same scores for each attribute.

While this isn't supposed to be based on PP I'd like to point out that pp is far more accurate than looking at a select top map rankings (because not everyone has top map rankings, and certain maps are far harder to get top rankings than others just by popularity rather than actual difficulty).

edit:

and also...

neko cookies
31 posts
ed August 2015
Ahhhhh this is looking pretty good so far :3c itll be good to use after the OWC is through though :>
Kert
2,079 posts
ed June 2009
Topic Starter

Proph Nobster wrote: 652b57

Can we get formal definitions on what these mean?
Check main page
IOException_old
IOException_old
Is this open source?
abraker
Global
8,355 posts
ed July 2014

IOException wrote: 4f2u6w

Is this open source?
Not at the moment
Remyria
363 posts
ed June 2012
Will you add the 3 others modes (Even if for them, it would be different)?
Riana
1,286 posts
ed October 2012
does it take some time update top 100 rank info?
i don't have any top rank record.
so i made some with latest ranked maps, but if i search my nickname, i see 'no scores in top 100'.
Kert
2,079 posts
ed June 2009
Topic Starter

Remyria wrote: 236526

Will you add the 3 others modes (Even if for them, it would be different)?
I am not proficient at other mods at all. Can't promise anything

Zisa wrote: fo6z

does it take some time update top 100 rank info?
i don't have any top rank record.
so i made some with latest ranked maps, but if i search my nickname, i see 'no scores in top 100'.
It's not automated yet. Please, wait for the next update to come
Professor Prinny
280 posts
ed July 2012

Kert wrote: 6o3b5i

Remyria wrote: 236526

Will you add the 3 others modes (Even if for them, it would be different)?

I am not proficient at other mods at all. Can't promise anything

Even if you aren't proficient in the other mods, couldn't you discuss it with people who are?
Kert
2,079 posts
ed June 2009
Topic Starter

Professor Prinny wrote: 69351u

Even if you aren't proficient in the other mods, couldn't you discuss it with people who are?
Everyone can suggest stuff here in this thread
Endaris
5,282 posts
ed June 2010
Based on what I wrote in my earlier post I suggest the following:
Right now it looks like you're checking OD first and rhythmical complexity after(if at all). Worst case would be if you just check amount of circles vs. amount of sliders.
In my opinion you should try and figure out some difficulty for rhythmic patterns and factor in speed and OD based on how difficult the pattern actually is.
This would fix the issue of rhythmically very easy maps with high OD(esp Easy and Normaldiffs with DTHR) being overweight in of accuracy while it would push actually difficult maps up as the OD would be worth more.
Kert
2,079 posts
ed June 2009
Topic Starter

Endaris wrote: 6j5f7

Based on what I wrote in my earlier post I suggest the following:
Right now it looks like you're checking OD first and rhythmical complexity after(if at all). Worst case would be if you just check amount of circles vs. amount of sliders.
In my opinion you should try and figure out some difficulty for rhythmic patterns and factor in speed and OD based on how difficult the pattern actually is.
This would fix the issue of rhythmically very easy maps with high OD(esp Easy and Normaldiffs with DTHR) being overweight in of accuracy while it would push actually difficult maps up as the OD would be worth more.
Don't worry we're updating accuracy hopefully at the end of this week
Kyozoru
334 posts
ed August 2015
Does it Really calculate your profile?
Kert
2,079 posts
ed June 2009
Topic Starter
Yes.

Meanwhile we got a new update!

- Added top PP scores
- Versus mode
- New Accuracy algorithm
- Some scaling tweaks
- Cosmetic changes
Houtarou Oreki
303 posts
ed March 2014
You're awesome
Sieg
3,981 posts
ed February 2012

Houtarou Oreki wrote: 2i2y5s

You're awesome
Niioxce
11 posts
ed June 2013
I love this idea, keep it up!
Endaris
5,282 posts
ed June 2010
Already looks way better. Quite a lot to do still.
My precision-rank though:

SPOILER
Precision

QLOCKS - Aozora (v2b) [x265 Difficulty] +HDHR (99.05%)
452
Hitomi Kuroishi - Over the Sky (Card N'FoRcE) [Hard] +HDHR (96.32%)
445
Hyadain - CRASH!! (Let's do it) (mtmcl) [Burai's Insane] (94.41%)
375
Wada Kouji - Biggest Dreamer (Ekaru) [Kasada's Insane] (94.89%)
363
Chata - Papapapanda (Nymph) [Insane] +HR (98.68%)
361
Koji Kondo - Dragon Roost Island (HolyBoy) [Normal] +HDHR (98.33%)
357
Rabpit - Saika (tsuka) [Expert] (98.7%)
348
Amane - Melt in the sky (Gabi) [Hard] +HR (97.9%)
336
Nightcore - You Got Me Dancing (osuplayer111) [Insane] (96.5%)
327
Kajiura Yuki - The Dreamers (Lissette) [Hard] +HR (97.2%)
316
DJ Satomi - Castles in the Sky (Hikato) [Insane] (98.41%)
313
Masami Ueda - Shinshuu Plains (Card N'FoRcE) [God] (96.66%)
308
Drop - Granat (Kaguya Hourain) [Hard] +HDHR (99%)
299
Enya - Trains And Winter Rains (Blue Dragon) [Hard] (99.49%)
289

This should try to include spacing and speed somehow, as Aozora+HR is pretty easy to aim compared to stuff like Saika. As the "ability to hit small circles" is related to Agility there should be some noticeable influence.
abraker
Global
8,355 posts
ed July 2014

Endaris wrote: 6j5f7

As the "ability to hit small circles" is related to Agility there should be some noticeable influence.
Ability to hit small circles should be related to precision, not agility.
Endaris
5,282 posts
ed June 2010
You can't hit a small circle without moving your cursor using your agility... ?
And if you can, you don't have to move your cursor and it isn't impressive at all if you hit it?
CelegaS
484 posts
ed April 2012
Should be ppv3
otoed1
639 posts
ed May 2014

CelegaS wrote: b3r4f

Should be ppv3
If this is developed well it could actually do just that.
Poncho
68 posts
ed February 2014
is the website down right now?
Kert
2,079 posts
ed June 2009
Topic Starter
http://downforeveryoneorjustme.com/osuskills.tk
It works
Endaris
5,282 posts
ed June 2010
Works for me.
Streaming/stamina can need some love too, it looks like deathstreams are correctly identified but that apart it's not looking so good:

This is for Tenacity

daniwellP - Nyanyanyanyanyanyanya! (theowest) [Collab] (97.87%)
326
(one hell of a long stream-map with spaced streams(up to 16 notes?) but no deathstreams)

DJ Genericname - Dear You (Rue) [Dear Rue] (98.32%)
393
Master Blaster - Come Clean (Nightcore Mix) (Asphyxia) [Collab] (97.44%)
390
Ceui - Asphodelus (terametis) [Insane] (99.05%)
382
^ all way easier in of streaming
The 15bpm difference to Asphodelus shouldn't matter at all if you look at the density of streams and the spacing.
Minhtam
550 posts
ed September 2014
Wait, so osu!Skills counts the players top 100 performances instead of the top 100 scores now?
Endaris
5,282 posts
ed June 2010

Minhtam wrote: h445w

Wait, so osu!Skills counts the players top 100 performances instead of the top 100 scores now?
both, but top 100 scores aren't updated as regularly
Minhtam
550 posts
ed September 2014

Endaris wrote: 6j5f7

Minhtam wrote: h445w

Wait, so osu!Skills counts the players top 100 performances instead of the top 100 scores now?
both, but top 100 scores aren't updated as regularly
It needs to be able to count more than that. I have plenty of scores that would have given more than 200 precision (thanks +HR), but they don't count because they fall outside my top 100 plays, hence my low precision rating. If there was a way for osu!Skills to take into every single recorded score, the program would be absolutely perfect.
-Sandy Corzeta-
75 posts
ed September 2013
Are you guys do have some API implementation on the future development?

if you do, please make some wiki or documentation of it like osu! did.
-Sandy Corzeta-
75 posts
ed September 2013
And sorry, a little bit some of bug issues. Why does most of the performances on the list taken on normal mode. Can you implement it on mania mode too?


*its cause i'm a mania player.
show more
Please sign in to reply.

New reply 1b384i

0
1 / 1,764