You might make a poll to know how much peoples are playing each mode so you can adjust multipliers properly...
I'm all for LNs being split into two parts for better feedback, and yeah I haven't tested scorev2 either because I can't but from what I've read in this thread the LNs were made tighter, when they should be the same or more lenient.Arras wrote:
I disagree. Scorev1 frequently has me beating my old scores where the old score has a higher accuracy and a similar combo. Hopefully this can mitigate that sort of thing, as it's really annoying when it happens. While I can't say anything about the release timing leniency as I haven't tried scorev2 (and I assume I can't, since cutting edge can't do multiplayer without supporter), the fact that LNs are split into two notes is a fantastic change imo. This gives the player much better feedback for how well they did on the start of an LN, and you can no longer cheese LN stuff by just holding everything and taking the 200s.
If anything, I'd say the game feeling good to play is far more important than "pp".
Yeah that's fine, sorry for derailing, I did however give my feedback on what I think about score v2 and what should be changed so take it or leave it.smoogipooo wrote:
Star rating and pp is irrelevant to this entire discussion. ScoreV2 will not even be ranked for a long time, and pp/sr calculations are a completely separate matter altogether.
We need more information, specifically:MegaAmoonguss wrote:
I tried out ScoreV2 in multiplayer yesterday and one person in the match literally SS'd the song but had one miss in the middle and got 896k for that. It probably would have been around 997-998k without the miss, and probably like 980-990k on regular Score. I'm not really a fan of how insanely the combo affects you score, and I'm personally hoping that it gets changed in some manner. One of the things I always liked about mania better than standard is the fact that in standard, it almost doesn't matter how well you actually do because if you hold a higher combo you get a higher score. I liked that in mania you can miss, but the emphasis is more on getting as many rainbow 300s and 300s as possible. I'm definitely all for some kind of combo system, but not one that punishes as hard as this lol
I guess it wasn't an updated version. I just used the ScoreV2 that is available in the latest stable release, which I guess must be an old version. The map was Sakura Mirage [ADVANCED], which is a low 3 star map and is 2 minutes long. I'll definitely check out the Cutting Edge version to see how it compares.smoogipooo wrote:
We need more information, specifically:
* How long was the map?
* Were they actually using ScoreV2? Or even the updated ScoreV2? Keep in mind they have to be on Cutting Edge to try it.
osu!mania ScoreV2 does not exist on anything but Cutting Edge at the moment, so it's possible that the player in question was not using it. Likewise you would've seen the old scoring for yourself.MegaAmoonguss wrote:
I guess it wasn't an updated version. I just used the ScoreV2 that is available in the latest stable release, which I guess must be an old version. The map was Sakura Mirage [ADVANCED], which is a low 3 star map and is 2 minutes long. I'll definitely check out the Cutting Edge version to see how it compares.smoogipooo wrote:
We need more information, specifically:
* How long was the map?
* Were they actually using ScoreV2? Or even the updated ScoreV2? Keep in mind they have to be on Cutting Edge to try it.
Mathisca wrote:
I think that we should regain life when we hold sliders, like in ScoreV1.
Actually just noticed this in a multi I played earlier too. Not sure if HP is fine with LNs atm and want more opinions.Kernaus wrote:
Mathisca wrote:
I think that we should regain life when we hold sliders, like in ScoreV1.
Iii dont agree with this, the HP rate in o!m is generally very lenient, and we dont have a constant hp drain like in standard, failing in mania is generally because the chart is way way above the player's abilities.
Yeap that's a bug, if they failed instantaneously when the map was completed :prohen04 wrote:
Actually I did notice that people have failed songs with V2 that they have never failed before. Maybe it's part of a bug, or because of the harder LNs. This definitely needs more testing.
Other than that, occasionally it showed me right at the end of the song that I have failed it, even though I had a FC. I haven't tested the newest version yet, maybe it's already fixed.
By the way, thanks for changing the Combo system. This should be more in line of what people expect from such a system.
I do have the suggestion to maybe not make the cap static, but dynamic (i.e. 10-20% of the total notes). This would help to make 4K and 7K equally hard in this regard, since 4K usually has less notes in a similar difficulty level. Also this might avoid problems with songs with less than 400 notes (not relevant for MWC).
This is a lot better, but combo still has some influence on score. I think a purely accuracy based scoring system is what osu!mania really needs:smoogipooo wrote:
- Combo
Combo is the harder one to talk about. We want to award holding combo, but at the same time not punish holding 4000 combo and missing once too much. To achieve this your individual hit scores are weighted by the combo you have after hitting the note. In the previous iteration this was a linear relationship, which resulted in punishing for missing after holding 4000 combo. In the new iteration it is logarithmic, with a cap at log_4(400) (meaning combo > 400 will be weighed as if your combo was 400), as shown in this graph (red = old, blue = new): https://u-gi.me/oJ6sa
So, its better than before but still, i dont think the combo cap has to be at log_4(400), we are osu!mania the lair of the TVSAIZUS DESU, but being serious, maybe 200~300 should be fine, need to test to see the differences and see what is best, but i think 400 is kinda high, meaning that taiko is ''2k'' the combo cap is 100, so why osu!mania should be 400? this is my opnion btw.smoogipooo wrote:
- Combo
Combo is the harder one to talk about. We want to award holding combo, but at the same time not punish holding 4000 combo and missing once too much. To achieve this your individual hit scores are weighted by the combo you have after hitting the note. In the previous iteration this was a linear relationship, which resulted in punishing for missing after holding 4000 combo. In the new iteration it is logarithmic, with a cap at log_4(400) (meaning combo > 400 will be weighed as if your combo was 400), as shown in this graph (red = old, blue = new): https://u-gi.me/oJ6sa
Here you go.FelipeLink wrote:
So, its better than before but still, i dont think the combo cap has to be at log_4(400), we are osu!mania the lair of the TVSAIZUS DESU, but being serious, maybe 200~300 should be fine, need to test to see the differences and see what is best, but i think 400 is kinda high, meaning that taiko is ''2k'' the combo cap is 100, so why osu!mania should be 400? this is my opnion btw.smoogipooo wrote:
- Combo
Combo is the harder one to talk about. We want to award holding combo, but at the same time not punish holding 4000 combo and missing once too much. To achieve this your individual hit scores are weighted by the combo you have after hitting the note. In the previous iteration this was a linear relationship, which resulted in punishing for missing after holding 4000 combo. In the new iteration it is logarithmic, with a cap at log_4(400) (meaning combo > 400 will be weighed as if your combo was 400), as shown in this graph (red = old, blue = new): https://u-gi.me/oJ6sa
P.S: Some broken mechanics on that combo cap is that we have 9 keymods(6 can be ranked), to be FAIR i think every Keymod need to has his own combo cap, because more keys= more notes, less keys= less notes, 400 could be fine for 7k, but for 4k maybe not, 200 could be good for 4k but for the other keymods? this is a BIG problem in my opnion.
Still, i dont think the score system needs to be changed, maybe this is the small problem in the osu!mania is the score system, all people say ''i had the same combo and accuracy'' but what about the ''300g'' ? and the ''100,50'' ? you can get a 99,5 with Nx100 or with Nx200, and probably the 99,5 with only 200 will be the higher score(meaning that the two scores has the same amount of 300g)
I want to see some scores like:
Example:
992k 1500x 300g, 3x200, 0x miss
Beating a:
991k 1520x 300g, 3x200, 0x miss) i dont think this will happen.
(considering that the 200s were in the same places on the runs)
Hope thats help, my english is kinda bad )':
i dont think you understand what i said, mania on score v1 has a combo cap already so ''higher max combo'' dont do anything.Arras wrote:
Here you go.
Note how I have almost a full 2% higher accuracy, higher max combo, much less HP loss, more MAX, more 300 and less of every other judgement, yet a lower score.
They are. I pretty much stopped heavily playing not too long after that anyway. The only reason I picked those was because I remembered that as a score where I had a lower score even though it should be higher looking at the stats.Endaris wrote:
Am I dumb or are both scores from 2015, Arras?
Yeah, and player with 10misses on 1000notes map can still get 990k points, no, thanksDrojoke wrote:
This is a lot better, but combo still has some influence on score. I think a purely accuracy based scoring system is what osu!mania really needs:smoogipooo wrote:
- Combo
Combo is the harder one to talk about. We want to award holding combo, but at the same time not punish holding 4000 combo and missing once too much. To achieve this your individual hit scores are weighted by the combo you have after hitting the note. In the previous iteration this was a linear relationship, which resulted in punishing for missing after holding 4000 combo. In the new iteration it is logarithmic, with a cap at log_4(400) (meaning combo > 400 will be weighed as if your combo was 400), as shown in this graph (red = old, blue = new): https://u-gi.me/oJ6saSo basically just Keep It Simple, Stupid. Any thoughts on this?
- Inconsistent gameplay is already punished in a 100% accuracy based score. Inconsistent gameplay = missing notes and not timing well, resulting in lower accuracy and score. No need to involve combo at all here.
- Gone are the days of beating your personal best score on a song with a lower accuracy, sometimes even resulting in a loss(!) of pp.
- A purely accuracy based scoring system is a lot simpler and more predictable. Hit notes = higher score. Good timing = higher score. Right now you need to be mathematically inclined in order to comprehend or do calculations with both ScoreV1 and ScoreV2.
- No weird edge cases like '1 miss 10 seconds in' being punished more severely than '1 miss on the last note'.
A pure accuracy system can be based in something else besides the sum of the judgment values (which is not really a very good way to measure accuracy, since the value of each judgment is pretty arbitrary).Tidek wrote:
Yeah, and player with 10misses on 1000notes map can still get 990k points, no, thanks
I agree with it being percentage based instead of 400, and I agree with the 5-6% max combo values, but smoogipooo and other combo lovers would probably want a bigger percentage. So I suggest 10% of max combo as a compromise (please note that combo bonus still scales up after the cap but much slower).Remyria wrote:
I suggest that the combo cap could be scaled on song at something like 5-6% of max combo as the cap
There's always the option of making misses/50s/100s reduce accuracy. Using Stepmania's scoring system (which punishes for misses by reducing score) scaled up to a maximum of 1,000,000 points, you would get a score of 960k when missing 10 notes and hitting 990 r300's.Tidek wrote:
Yeah, and player with 10misses on 1000notes map can still get 990k points, no, thanks
You are forcing more anxiety with the combo based system, actually. Frustration as well. Currently the source of those emotions in mania is primarily when trying to PF a map with nothing but MAX or at least SS for less skillful players, but with this, you are going to extend that further. Having your score screwed for the rest of the play because of these mishaps on a non skillful level doesn't sound like a good gameplay mechanic. Also, a map can be as long as you want to be, but unless a player is being skillfully challenged, the player will get tired only from boredom. The way you made it sound is that you were referring to physical tiredness, which is not always the case.smoogipooo wrote:
As maps progress anxiety builds up and you become tired, both of those are indications of how good of a player you are, or otherwise, how consistent of a player you are.
Gladly! I have my own proposal for a scoring system which should be better than accuracy based scoring and combo based scoring. The concept of this scoring system is influenced by the difficulty of the parts the player is playing.smoogipooo wrote:
Explain what the "better ways" are? As I mentioned in the OP we are taking feedback, and we have lots of time to make changes.
Score += (diffScore[t] * accScore[t] * log(t)) / missesPunisher
I agree. This will work only if star rating is fixed, though it I think it could be better than scoreV1 even now. Also due to the nature of this thread I tried my best to avoid mentioning star rating. When I say difficulty, substitute that for what you percieve the difficulty as instead of star rating.FrenzyLi wrote:
diffScore take the difficulty of the map at point t
this is not a post about star rating. Any score system that takes into account the difficulty at time / point t should be deferred until a better, non-jumpstream-inflated, workable improvement or revamp to the SR system is made.
Whatever works. The formula I put up is psuedocode and not in proper mathematical notation.FrenzyLi wrote:
AccScore take the player's score point (50, 100, 200, 300, MAX) at point t
scalar sum of score values? I would rather not take a vector.
See aboveFrenzyLi wrote:
log(t) will a pause in a map affect this? You can't just take a single formula for everything. Divide map into discrete parts (by combo, timestamp, whatever) because a beatmap is not a continuous function. Smoothing will help but how do you smoothen this discrete chunk of data of user replays / beatmap hitobjects?
Not sure what you mean by the first sentence. To address your second point, this is the formula at the core without any constants. Ofc you would shift the function to the left to avoid the t=0 situation.FrenzyLi wrote:
Also, the objects at the end of a map is exaggerated in their importance compared to the first few notes. What if beatmaps has hit objects at point t=0?
I will experiment with this bit and see what works best.FrenzyLi wrote:
missesPunisher: v1 thanks.
dynamic last X milliseconds: seems interesting, please propose a working formula.
smoogi wrote:
Please note that HD/FI/FL mod multipliers have not yet been removed. These are slated to be removed in the next iteration of changes.
Sounds like the same thing to me. The main difference is that your system works backwards from mine. You are putting the resultant score and getting out the corresponding difficulty of obtaining that score. I am putting in difficulty and result and getting out the score. Your system work of potential worth while my system works off immediate worth.Full Tablet wrote:
I think it is better to not make the scoring system difficulty-dependent.
Instead, make the star rating calculation output a function f(score) that maps a certain amount of score with the difficulty of achieving that score in the specific map.
I had a feeling I missed something. 15 pages was a lot to catch up to.Endaris wrote:
@abraker:smoogi wrote:
Please note that HD/FI/FL mod multipliers have not yet been removed. These are slated to be removed in the next iteration of changes.
Visual mods won't have a multiplier, relax.LinkTaylord wrote:
How disgusting! ._.
You'll see ''best global players'' will be using HD/FI/FL only.
Goodbye Jackads, Inteliser, Yuko.