Judging from this and the mostly similar opinions I'll have adjust the difficulty algorithm, so it can give more sensible values for 4K vs 7K maps. It should be noted, though, that the 7K version already _IS_ considered harder. Just not by enough to outweight the score differences you had relative to the maximum possible scores.Xenlon wrote:
Ok,I played this map again but with 4k and got 765.000 points.
I personally think the 7k version is harder then the 4k,but thats not the point.
The first time I got 815.000 points and thats more then 765.000,so the first try is my Top Score.
Does this mean that I have to play some maps with 4k and 99% ,when I could pass them with 7k and 95-96%,just for the Top Score?
I think the same 7k converted map is always easier in 4k then in 7k ,but I want to play with 7k,because 4k isn't challenging me enough.
(This is my opinion,because I never saw a converted map that is easier in 7k then in 4k)
I plan to make quite some changes to the mania difficulty algorithm. Yes, easier 4K diffs are overrated and higher ones underrated. And yes, the algorithm overrates certain patterns in general by a lot, which is bad.Luna wrote:
My thoughts as a noob mania player:
4k is overrated at the low level, but underrated for high tier maps.
After accumulating over 1.5k plays of 7k, I tried out 4k for maybe ~150 plays. Now I have 18 4k maps in my top 20 performances (and more beyond that, but I'm too lazy to count). I mean, I probably have an advantage with 4k-type modes due to my taiko experience, but I really don't think it should outscale my 7k performances that easily.
The point is, all of the 4k maps I played are extremely easy to read with simple patterns - yet they are rated as harder than much more complex 7k maps with a similar overall note density. I'm making assumptions about your calculation methods here, but maybe you could scale down the weight of per-lane density for low density values (keep it the same or even increase it for really high density)?
On the other hand, the hardest 4k maps are weirdly underrated. Imperishable Night 2006, Bangin' Burst and Kamui are all listed as "I"-level difficulties instead of "X". That can't be right, can it?
My top 7k performance also feels a bit out-of-place. Personally I feel like my FC on Utakata, Ai no Mahoroba [LV.14] should be my top performance, but the algorithm considers Stella [Pro Drum] better. I assume it's because of accuracy since Stella uses OD9, but even so, Stella has a slightly inflated star rating in my opinion. The map uses a lot of chords, but that's about it. They are not even hard chords at all, since most of them are symmetrical and extremely repetitive. And I didn't even FC or anything. So maybe reduce difficulty of highly symmetrical maps if others get the same impression from higher level maps?
Easy increases the HitWindow and decreases the accuracy pp you can get for your score by doing that. At your accuracy of ~88% you barely get any accuracy point anyway, so Easy won't really make a difference apart from allowing you to pass the map more easily. You would probably get a lot more pp by getting better scores and accuracy on slightly easier maps. Mind trying out and confirming, denying?Fudgyking wrote:
There seams to be something very broken about easy mod. I first played this map : https://osu.ppy.sh/b/279134 with the EZ mode and got a low 300k score with a 88,6% accuracy which then appeared on my top5 best performance on my profile. So, I decided to beat my score without the mode to get an even better performance. I got 2x my old score BUT a 87,8% accuracy and then I suddenly lost 22 global pp. When I went to my profile, my score went from being worth 180~pp to 116pp.
It doesn't look like the EZ mod gives any kind of penalty and thus can be used to pass maps way easily and still get a better performance then when you don't use it...
Ok here is my feeback. I played this map until I somehow manage to beat my EZ accuracy with no mod and I finally managed to get a 89,25% accuracy. I then went to my profile to see how much pp it was worth now and..... it was worth a mere 158pp; way less than the 181pp my 88,61% EZ mod score gave me. I do not understand how this can be possible if there is truly a decrease in the accuracy pp.Tom94 wrote:
Easy increases the HitWindow and decreases the accuracy pp you can get for your score by doing that. At your accuracy of ~88% you barely get any accuracy point anyway, so Easy won't really make a difference apart from allowing you to pass the map more easily. You would probably get a lot more pp by getting better scores and accuracy on slightly easier maps. Mind trying out and confirming, denying?
Can someone compare Easy with nomod while doing higher accuracy runs? Anything from 90% - 100% is fine, more feedback is always good.
In my easy run, I had a 149x max combo and 41 misses while in my no mod best score, I have a 177x max combo and 42 misses. So, yeah. Nothing that major hereLuna wrote:
Did you maybe have a significantly higher max combo or less misses in your Easy run?
Combo / misses don't really matter in osu!mania. The actual score relative to the maximum possible score is looked at (non-linear before someone complainsLuna wrote:
Did you maybe have a significantly higher max combo or less misses in your Easy run?
) and accuracy is factored in, too, by a smaller portion. (See the wiki.)Tidek wrote:
Does SV/BPM changes affect pp?
Its impossible to change scrolling speed while playing now, but as I remember It was possible to do while pausing a game (so, we have pretty high scores on "bpm jumpy" maps, now its pretty hard to get high acc on maps like this http://osu.ppy.sh/s/107565 because of scrolling speed).
Was already implemented, but then removed. And tbh I don't like the fact of playing w/ mods to get more points.Tidek wrote:
Tidek wrote:
Does SV/BPM changes affect pp?
Its impossible to change scrolling speed while playing now, but as I remember It was possible to do while pausing a game (so, we have pretty high scores on "bpm jumpy" maps, now its pretty hard to get high acc on maps like this http://osu.ppy.sh/s/107565 because of scrolling speed).
Bumping this and 1 more question, does DT in future will affect PP? Because atm its pretty useless mod to use because you are not getting any bonus for map difficulty...
There haven't been updates to the pp system for any of the modes in a while, so don't worry - Tom is not specifically ignoring this thread or anything. I'm assuming he's busy with personal stuff; just give it some time and eventually updates will happen.-Kamikaze- wrote:
also we didn't really get any response about EZ mod and I don't see changes in mod key weighting, is this thread already abandoned? D:
It's already been determined that score alone is not an accurate measure of performance and that your best score may not mean your best performance.mathl33t wrote:
I just lost 1 pp by beating my best score on a map. My accuracy went down, so maybe it's that. If it is, then scoring needs to be looked at. One or the other of pp and scoring needs to be fixed so that this situation doesn't happen.
Funny thing that I've beaten my Utakata, Ai no Mahoroba [LV.28] S (95,3%, 800k pts) with an A (94,92%, 821k) and it went from 230pp worth to 237pp. I believe acc has nothing to do with it, it's just weighted by score.Meseki wrote:
It's already been determined that score alone is not an accurate measure of performance and that your best score may not mean your best performance.mathl33t wrote:
I just lost 1 pp by beating my best score on a map. My accuracy went down, so maybe it's that. If it is, then scoring needs to be looked at. One or the other of pp and scoring needs to be fixed so that this situation doesn't happen.
However, I believe that fixing that issue would require either a scoreboard wipe (of over 6 years of scores, since I'm pretty sure the issue applies to all modes) or storing two scores on the server, which would drastically increase the space used.
It's a problem that we may have no choice but to deal with.
That's true.PyaKura wrote:
IIRC Tom said acc DOES have to do with it, but it's far more negligible than actual score.
You still got a crap ton of pp for a halftime mod o_ozKskita wrote:
I just beat my highest EZ score in Zirkfied with HT by quite a wide margin and went from A to S. The result was losing 16 pp...
This is why I thought we should have a MAT (Mania appreciation team) or something like that. A computer can't compute the difficulty of a mania map like it can for any other game mode difficulty since mania hasn't been around long on osu!. The minds of 20 players more or less if needed can come up with a better solution of difficulty rating. The only problem with starting a team like that is finding a group of well balanced people. I have a few on my mind who could possibly do something like this, but only a few and not knowing if they'd be willing at first.Drace wrote:
Well that's to be expected with an algorithm without any pattern recongnition
The reason I said "I" was because Tom brought up the topic again to me after a suggestion to peppy I gave weeks before the chat feud. (That was when the whole DT PP giving too much feud was happening.)Drace wrote:
"You" were not the one who thought of it, watch your wording. It was first brought up by Envi and me on page 7.
Considering how we can't even find BATs, a diff grading team might actually not be the best idea. Well for now atleast.
HAHAHAHA XD17VA wrote:
http://osu.ppy.sh/b/370840
native faith 1K master? - why 1K
ACs' SR may very well be broken. You're also comparing an extreme 7K mania spec with a (extreme ?) 4K AC here..._Gezo_ wrote:
https://osu.ppy.sh/s/92190 <= 7.11, passed by a lot of people, some being very close to 99%
https://osu.ppy.sh/s/179779 + p/3160605 <= 7.13, having played it I am sure Staiain will have a very hard time passing it.
Why ?