forum

osu! ScoreV2 Discussion

posted
Total Posts
171
Topic Starter
smoogipoo
Hey all,

We're just over two months away from OWC and I want to know what the community likes/dislikes about the current ScoreV2 system so we can get it perfected before OWC comes a-knocking.

THIS IS NOT FINAL
Please, do not discuss Star Rating and PP here.

Here are a few facts/talking points about the current osu! ScoreV2 system:
  1. 70% of score comes from combo, the remaining 30% comes from accuracy.
  2. Spinners award 500 points per tick.
  3. The mod multipliers are as follows: HR - 1.10x, DT - 1.20x, FL - 1.12x, HD - 1.06x
Basically, I'm willing to experiment trying different stuff so throw out any ideas you have. I'll be checking this thread periodically but don't expect me to reply to every comment, and please _please_ don't spam me with PMs telling me to read a comment in here ;___;.
Nakano Itsuki
is slider acc still gonna be part of scorev2?
I'd rather it being removed tbh, just too much stress when this is added in tournaments
-Visceral-
Are there any ideas that are being thrown around that could become a possibility soon for ScoreV2? I really like where it's at currently and it feels much more of a representation of skill than ScoreV1 matches, so that's awesome :D
Spaghetti
i love it, seems like its going in the right direction
hehe
similar to pp, miss count could be incorporated into the final score. like a final multiplier. it's a good indicator of aim apart from combo count. just a thought.
Loctav
StarrStyx: Define "slider acc". You mean hitting the slider start with correct precision? While most maps are not designed around it specifically, I feel like it actually should be a thing, given this is a rhythm game (and it is enforced in many other games of the same genre in a comparable fashion as well)

Smoothie World: we are open to any kind of suggestions. It has been a year since OWC passed and in order to fresh up things a little, maybe you or others have ideas in how to balance it any further to be a proper scoring for competitive environements.
Topic Starter
smoogipoo
I agree with Loctav - slider start circle accuracy is intended and I don't see it being changed.
Nakano Itsuki
Can I clarify one thing? For slider acc I remember that if the slider is hit within the 50 hitwindow, even if the player aims it properly it would still break your combo.
Is this true or not?

If its true I just dont understand why its there, cuz why is it combo breaking on sliders but not on circles?
Asonate
Would it be possible to somehow "nerfe"/remove Slideraccuracy on certain sliders?
This would make it possible to let players get through BPM changes and could also be used to adapt old maps that use sliders to lead in bpm changes to scoreV2 without making them too hard too acc well (and also make them worth less so you dont have mappers spamming these to keep ScoreV1/have mappers justify every single one during modding process)
Topic Starter
smoogipoo
@StarrStyx: That sounds false.
@Asonate: Consistency is a key. It would not be easy/feasible/intuitive to nerf/remove slider acc on certain sliders.

Can we just keep it at "slider acc won't be changed"?
Kite
I am split on the slider accuracy, on non techincal maps it's beneficial and a fun change but on strict and ever changing rythmns that use sliders for simplification it's a disaster.
Not a fan of the overly weighed combo either, I consider accuracy to be just as important if not more important.

_________________________________________________________________________________________

This was written some time ago and never knew what to do with it, maybe you can get some ideas out of it.
http://puu.sh/qPhYM/cd5b043393.txt

It goes more into the direction of score/pp interaction though.
ac8129464363

Kite wrote:

but on strict and ever changing rythmns that use sliders for simplification it's a disaster.
(this is regarding main game more than tourney play but)

basically my thoughts on this as well. if a mapper doesn't want stuff like that to be the focus of the difficulty in his map he shouldn't be forced to. over the past 9 years it's just become another tool in the mapper's toolbox to make certain things not pains in the ass and I don't think it's fair to throw away 9 years of development in mapping. i realize this argument comes up every time but i've yet to see a compelling defense to it that's not something along the lines of "we need to move forward"
Kaifin
as both a player who might be playing in the tournament and a mapper, i have to agree that slider acc is something that feels out of place and out of touch to throw randomly into the biggest tournament of the entire game

its like if you randomly changed a vital aspect of the game for the most important competition of said game, if you're gonna make a change like score v2 it shouldn't be trialed in the biggest tourney of the entire game

as for the other aspects of score v2, i think that the acc vs combo weighting is just about right, especially if slider acc is removed, because its actually slightly more than 70/30 when you think about it because 100s/50s also detract from your score

if the 50 slider break thing is still in score v2, please remove it, it doesnt make any sense, hitting a 50 should give you a 50, not a miss (that is if you even keep slider acc)

the mod balancing is good! i think its very good that HR is worth more than HD and adds an interesting dynamic to freemod

also, the score system is kinda lame from both a player and spectator perspective, is it more exciting to see 100k point totals on both sides during a tense long tiebreaker, or 4 million? i understand the difficulties and limitations that come with implementing the old score system with the new 70/30 acc weighting, but in my opinion that is the most imperative thing that needs to change

to clarify, i think score v2 is a good idea and a step in the right direction for osu, but needs a couple tweaks before its fully tested on the world stage in like THE most important tournament for standard players

i like score v1 more personally, but i know that this is the future and there's no real stopping score v2, but i think that score v2 takes away from the unique feeling and niche of osu as a rhythm game. right now, osu is much more of an aim game than a rhythm game, but that's because its pretty faithful to its source material (Osu! Tatakae! Ouendan)! Now, i know that you've all heard this a million times, but the entire basis of mapping rhythm has centered about using this system that allows sliders to be a more lenient means of hitting notes

now, you may say: mapping have evolved to that point! just let it evolve again! to which i will pose the question: what exactly are the BENEFITS of score v2 that couldn't just be added to score v1? what does slider acc bring to the game exactly? because as of right now, it not only ruins an entire genre of technical mapping (as someone who mainly plays technical maps this is a huge huge huge concern) such as slider heavy wub maps, but also DETRACTS from most, if not all current maps

and to gain what? a sense of superiority when stepmania players say "LOL osu thats not a real rhythm game!1!!!11!"

it feels as if in each staff statement regarding score v2 and slider acc, the argument always boils down to: "its a rhythm game so it should be as rhythmic as possible." but to that, i'd point out that osu has a component that no other "REAL rhythm game" has, an aim component

like gee i wonder why the hit judges aren't brutal like in other rhythm games, probably because you have to aim complex patterns while tapping to the rhythm of the song! it's what makes osu unique, and taking away the lenient nature of sliders by adding accuracy goes against the inherent design of the game and how its played

thats just my two cents about the issue, i hope you consider these points before implementing it for OWC and more importantly the game as a whole

again i think its wrong to use the biggest osu tourney to test out your new score system experiment but that happened last year so :)
Kite
I'd like to add that scorev2 should be kept for tournament purposes only.
In the current state of the game it makes no sense to introduce a new score system to the main game as everything revolves around PP.
Scoreboards currently only matter very little.
Endaris
I would cite my most recent suggestion from the old thread but apparently you moved it to the wastelands or some other non-public area so I will just rewrite it here with a higher degree of elaboration:

Change the hitwindows of circles for the sake of reducing notelock.
In OWC we are very likely to see maps with spaced streams that heavily punish the occurence of notelock.
At the same time it can be expected from our elite players that they will only hit something that is currently a 50 when they committed a rather grave mistake. I feel that missing instead due to a lowered hit area is a legitimate punishment for entirely mistimed hits and the benefits of reduced notelock weigh more than that.
Personally I can just say that I feel cheated by the game when I hit a single as a 50 (on OD8) because I obviously messed up so hard that I do not deserve to keep combo. While I can't guarantee it, I think many good players share this experience.

So we got this sweet OD-table on the wiki:

As easy to determine the hitarea of 50s will go down about 1.66 times the value of the 300 hitarea while it is 1.25 times for the 100s area.
For casual OD this isn't that grave but especially at peak-OD (above 8) it really starts to show as the 50/100 area becomes bigger and bigger relative to the 300 hitarea.

I made this little table to elaborate the issue (picked "common" OD-values in OWC and lower for comparison):
collapsed image

(the units on the relative values are obviously nonsense)


As you can see easily, the hitarea of 300s relative to the total hitarea shrinks by a whole 9,4% from OD7 to OD10 (more than it does from OD7 to OD3).
On top of it the 300 area almost halves when comparing OD10 with OD7 while there is only a moderate change for the 100/50 hitarea.

Now this is a very big discrepancy imo. The system is very lenient regarding hitting stuff offtime while it becomes very strict for hitting on time.
The problem with this is that there is no value in having a great leniency for somehow hitting stuff when the actual sweet area is that small (20%).
On the opposite, the great leniency causes problems on certain maps with notelock and timeshifting on streams
timeshifting
timeshifting is when you misaim in a stream but can't click the next note due to notelock and have to wait for the misaimed note to register as a miss so you can hit the next one without notelock occuring. However there is no way to not hit it off-time (depending on how much the notelock overlaps the next note) and you will often time end up hitting perfectly on time but being shifted by one note against the stream, causing you to hit only 50s unless you purposely overstream (which feels stupid because you were actually streaming perfectly on time).

The result:
When you missed one note in a stream you get heavily punished and can only soften the punishment by hitting off-time and faster to nullify the shift - basically it's everyone's favorite way to die on maps such as Freedom Dive where the notes in streams are 67ms apart from each other. Once you missed a note in the stream you can't hit the next one before being at least 53ms late which means that the minimum time difference to the next note's sweetspot is only 14ms. In the usual case you will fall one note behind and as notelock likes to induces further notelock you will most likely die.

Especially in cases such as Freedom Dive (OWC'15 finals) where the overlap is so big that you're forced to hit 50s in order to not notelock more and die it is extremely difficult to compensate for something that is only ONE miss. Now it's obvious that even with a reduction of the hitareas you can't remove the possibility of notelock on 222bpm streams but it is possible to get it down to a degree where you're not bound to automatically hit 50s and/or die and most of all it significantly reduces the mental effort you have to do in order to not trigger a sequence of notelocks off of one notelock.
tl;dr
Notelock occurs less frequently and is easier to deal with if 50/100 area is reduced.

For reference on notelock also read [Guide] Avoiding "notelock" at high BPM.
According to the values here pretty much everything at 150bpm and above is prone to notelock within streams at OD10.

I don't see any reason to not introduce a new OD-table that reduces the size of the lower judgements in order to raise this notelockborder to a bpm where a significant range of the relevant bpm is unlikely to cause notelock.
Like this for example (I just cut the areas of 100/50 by 10ms on each side):
collapsed image

(the units on the relative values are obviously nonsense)


With the 50s area being down to 80ms you are already very unlikely to notelock on 170bpm and below making misses actual misses and not timeshifting notelock-bullshit. Reducing it even further to 70ms would cover for streams up to 190bpm.
I think having such timewindows is very fair to the quality of our top 5k players.
It can significantly reduce the occurence of relatively random deaths on maps like PUPA+HR (OWC '15 Finals Freemod) where notelocking on one specific stream will cost you the map regardless of your performance in the other 110 seconds of playtime. (that being said, a 80ms window would still cause notelocking with perfect timing while 70ms wouldn't but notelocking is still likely to occur/being randomly avoided on both).
It will slightly raise the significancy of the accuracy-score as 50s will be more likely to occur and also make a performance that consistently strays far from optimal accuracy more likely for death from drain as you're more likely to miss out on the bonus-health from Katus as well.

In such a context the bonus on HR might have to get increased. As a fun spice you could keep the old judgementareas for sliders as people won't like the sliders that are delicate to hit on time being even harder.

If I messed up the sheets somehow, tell me where because I threw them together rather quickly. Also feel free to come up with different values. I just wanted to outline this problem and propose a solution, details adjustable.


-----

Another thing I'm not sure of is the drain.
With slider acc you lose out on free health from properly aiming sliders and makes the occurence of Geki more rare as well.
As far as I can see the health lost from missing sliders/sliderbreaks or rather the health gained from 100s wasn't ever adjusted to keep the maps about as hard to pass as without slideraccuracy.
Since the health is set by the mapper with v1 in mind this seems only fair for me and would also make absolutely stupid deaths on maps such as Clumsy Thoughts+HR (OWC '15 Group Stage) a less significant factor. Playing for "who survives?" just isn't that fun when you could look at a more nuanced performance of each player to determine the result instead.
The reasoning here is obviously similar to the one for notelock+PUPA. I don't want to see people not having their score counted when their performance on the map was actually good.

/edit: fixed one typo, one not-so-significant semantic and commented on the nonsense-units in my screenshots
/edit 3years later: reuploaded broken image links
hehe
are there any changes to the multipliers of difficulty-reduction mods?
7ambda
Regarding combo and acc percentages, I honestly think they should be changed so that acc is 60% while combo is 40%.
TakuMii
I think the most important thing that anyone could ask for is to be able to test out the scoreV2 system freely. It's kinda difficult to judge how well a system will work without actually using it first-hand. And it may be just a pipe dream, but I'd be willing to see the new scores calculated alongside the old scores instead of replacing them (a la Beatmania IIDX) once this metric becomes good enough to become the primary score system.

As for an actual suggestion: Maybe implement a score cap that scales off of max combo count rather than simply capping every single map to 1 million? As far as I'm aware, the score cap was the biggest concern with the original scorev2 implementation, so this might be a decent way to address the psychological issue without deviating too much from the original calculations.
Halfslashed
As someone who has played and staffed many tournaments with score v2, team based and not, my biggest concern (slider acc aside) is combo:score scaling.

Currently, it is actually possible for someone with 98% and a 500-600 combo on a 2400 combo map to narrowly beat out someone with a 1200 combo and 90% accuracy. The problem with this is that the score difference between the 500-600 combo and 1200 combo isn't enough to justify the consistency involved in getting the higher combo. It turns out that for a good few people, it's much harder to maintain higher combos than accuracy, even if you're heavily trained in the area of skill the map requires for you to get a higher combo.

I can't discount the fact that players can tank accuracy in a tournament, but I see the accuracy percentage as a factor of how much a player has spent their time developing their skill up to their point, while combo is represented by not only skill, but how stable they can remain during play, how familiar they are with the maps, etc.

Now, this actually isn't as much of a problem for maps in the 1000 combo range or below, since a player will have a much easier time obtaining a higher proportion of the combo, and accuracy will still play a role, just not as big of one.

My suggestion to this is to make the combo portion of score v2 linear, and decrease the scaling each time there is a combo break (preferrably after 20-40 combo or so, since when players miss, they can sometimes do it in rapid succession due to a tricky part). Some math would have to be played around with, but the concept would be something like 1x scaling to start, 0.7x after first break, 0.5x after second, etc. Also, have this scaling apply and change depending on the biggest combo, as to not punish early breaks. Ex (assumes above proportions): On a 1000 combo map, I broke 3 times and got combos of 50x, 850x, and 100x. The 850x would have normal weight, the 100x would have 0.7x weight, and the 50x wold have 0.5 weight towards my total combo score weight.

The other issue is more of a show thing, but one of the other complaints I've heard about score v2 is that it's boring to watch. People miss the big numbers of score v1, and it's not as enjoyable to see someone FC a 2k combo map to get a score less than 1 mil. My suggestion here is to just multiply the 1 mill by the max combo of the map and some constant. Ex (here I used a constant of 0.03): https://osu.ppy.sh/b/221777 would have a score of 71,640,000 for a nomod SS. This may seem silly, but I think it would make the audience as well as maybe even the players happier overall.

That's pretty much it, the weights are fine at lower combo lengths, but the combo:score ratio is messed up at higher ones.
Enon
OWC is osu!standard world cup.

it dosen't need to input a characteristic of other game modes like slider-acc stuff since that is breaking originality of osu!standard mode.

that stuff would be acutally kinda not good if you have been played osu!standard without the stuff in my opinion...

I also have tried that and felt kinda not good from the stuff.....


it's fine system though if there is no slider-acc stuff :)
MiruHong
Circle
If we have the slider acc to be the same as circle acc, that would make the difference between sliders and circles less noticeable. This would impair mappers' ability to create contrast in song components using sliders and circles.

However, I also think that as a rhythm game, it is important for the sliders to be clicked somewhat on time. Therefore, I propose having a separate OD for sliders in which the timing window for sliders is 2x longer than the timing window for circles. For example, a map has OD10. There would be +- 19.5 ms leniency for circles and +- 39 ms leniency for sliders (effectively OD 6.66~). The multiplier may be adjusted based on testing of the score system in the future.

For other components of the scoreV2 system:
I agree and second Halfslashed's idea for combo scaling. Implementation of Halfslashed's idea would further differentiate and reward high combos. This change still would make team carry (which scoreV2 tries to prevent) difficult to do as the difference between full combo score and half combo score (one miss in the middle and assume same accuracy) would be 8.75% of the score cap.
hehe

MiruHong wrote:

with this in mind maybe a change in the percentages of combo and accuracy, while adding a final multiplier for # of misses. this avoids unnecessary complication and all data can be pulled from the results screen. something like 50/50 and 0.98 or 0.99 score per miss. 0.97 would feel a little too punishing for misaiming a stream etc.
Endie-

Spaghetti wrote:

i love it, seems like its going in the right direction
TakuMii

handsome wrote:

with this in mind maybe a change in the percentages of combo and accuracy, while adding a final multiplier for # of misses. this avoids unnecessary complication and all data can be pulled from the results screen. something like 50/50 and 0.98 or 0.99 score per miss. 0.97 would feel a little too punishing for misaiming a stream etc.
The problem with adding a final multiplier for misses is that it won't affect the score during gameplay, and if it did, it would cause sudden score decreases. I don't think this makes sense, especially in a competitive environment, as it'd only make the teams' performance during gameplay more ambiguous to the viewers. Misses have always counted as zero, and I believe they should be kept that way.
If anything, misses could be handled as Halfslashed has suggested, with a decreasing combo multiplier each time a miss occurs. I somewhat agree with the concept, but the numbers and scaling would have to be tweaked significantly as to avoid punishing early misses more than late misses.
show more
Please sign in to reply.

New reply