How did you get started with osu! after completing the tutorial? As in, before beatmapping contests resulted in the beatmap being packaged with the game and only the tutorial was present, how did you figure out how to find maps?

July 31, 2015

Not sure if this is a serious question. Just used the beatmaps section of the website. o_o

Liked by: DroppedBass

Hello Tom94. Do you considered dependence acc on pattern=pp from it? For example: if map has streams, and someone will get 100s on these streams, he will get WAY less pp than he would get with ssing these streams and getting the same amount 100s on easy part. It would stop people mashing.

July 28, 2015

Sadly pp can't do that, since information about where 100s occured is not stored at the servers. We also can't just start storing such information, because it would require huge amounts of space which in turn would bring many performance problems with it.

Liked by: DroppedBass

I don't know if you are even involved in this, but do you know if there are plans to have the editor support custom/more time signatures? Because right now mapping/timing something like this http://vgy.me/iXLuh5.jpg is extremely tedious as only x/4th is possible (with x not allowing for decimals).

July 28, 2015

I don't know of any plans to support that, but I agree that it would be a nice addition. When timing some maps for friends I also wished for such a feature.

Liked by: DroppedBass Shira

Hey, I was thinking about playing your rainbownuke build, and I think that in 2.0 they were supposed to change the discharge animation, so it only displays one at a time or something like that? I've heard about it before the patch, but I'm not sure if it's in place. Do you know? My computer is shit

July 28, 2015

It's not in place. :(

Is the pp system also like the star system where you say that 1000pp should be impossible for everyone and 800pp is like 8* and should only be accessible to a few? At least if std progresses to that level

July 28, 2015

No, I generally let pp rise higher and higher as long as it is because players are improving. I just try to keep the top more or less similar between modes for the sake of consistency.

Liked by: DroppedBass Shira

http://ask.fm/Tom94/answer/131076902606 http://puu.sh/jdsNk/aaf4dbf1ce.txt

July 27, 2015

The only cases where a sufficiently skilled player loses to a lower skilled player are luck-based cases. As soon as luck is not involved the better player will almost certainly beat a less lucky player. When going for topranks where you can retry as much as you want luck is pretty much a non-factor, so your call of duty example doesn't apply. I do agree, that when restricting to a single play, then in standard luck is a larger factor than in mania due to the scoring system, but this is irrelevant since star rating measures the cases where luck isn't a factor - the difficulty to get a good score given as many retries as you want. Thus is there is (almost) no variance in the results a player of a given skill can achieve, then up to a super small point it doesn't matter how squeezed together the theoretical skill ceiling is. The ordering of players will still be correct.

As a consequence it also doesn't matter how tight the star ratings are. Whether they go for 0-10 or from 0-1000 doesn't make a difference when judging how good a player is, since each player would have the same rank in the end anyways. If you want to obtain the stars from 0-1000 you just multiply the stars from 0-10 with 100. See how there is literally no difference apart from how they are displayed? That's the point. In osu! we decided for aesthetic reasons, that 0-10 stars would look good, so we simply scale the range of playable difficulties to that range. We further squeeze and stretch that range so that stars fit the Easy, Normal, Hard, Insane, etc. scheme. Still, the rank ordering of players stays the same, only the star numbers change.

Wanting larger star numbers is literally sacrificing the standard used throughout the game for a larger e-peen.

Now onto pp. There it indeed makes a difference whether a harder score gives 1200 or 600 pp, given all others stay the same. This however is not a necessary constraint. If a score truly warrants to be worth twice as much as another one, then literally all scores can be scaled down in their pp value to finally produce sensible player pp while still retaining the appropriate ratio of pp between different scores. Understood now?

Liked by: DroppedBass

http://ask.fm/Tom94/answer/131091602382 I don't think I quite understand that? I mean as opposed to 4.00 to 4.99, 5.00 to 5.99 etc.

July 27, 2015

Oh I never actually thought about that. That has been behavior ever since I can remember, but I do see your point.

Liked by: DroppedBass

in mania 10stars should be at the same difficulty level than a ~overjoy7 then, and 8stars ~overjoy5-6. Do you have any experienced player that could help you, or that you can check it with them?

July 26, 2015

Yes, I do, Aqo is one of the people I talk to the most about these kind of things.

Liked by: DroppedBass Shirotsu

http://ask.fm/Tom94/answer/131075860174 take each mod seperetely, you shouldn't make it that a sertain diff in standart should represent the same shit in mania, there's absolutely no point in that

July 26, 2015

That's exactly what I am doing. I'm making the difficulty system depend on what the players of the mode are capable of playing such that we have a reasonable scale from 0 to 10 stars, because that's the range the game displays.

Liked by: DroppedBass

http://ask.fm/Tom94/answer/131075860174 But isn't that ever changing? I mean, the skill always improves. If you would've installed such a system for Beatmania in 2005 (8 years uptime, just like osu! now) it would be COMPLETELY thrashed by now.

July 26, 2015

Yes, and stars and pp are also ever changing. I don't see any issues with the difficulty system adapting to what players are doing.

Liked by: DroppedBass

Mania doesn't have any hard maps at all. If MANIERA(o!mania version) was a BMS chart It would be around a ★17-★19 and there would still be around 175-190 people who could get a S rank. (estimating off of global ★17-★19 stats, but even then o!mania has easier timing. solution is to rank harder maps

July 26, 2015

That's pretty much what I've been saying. Turn stars down in preparation for actually hard stuff, don't overweight what many people can do.

Liked by: DroppedBass Rafacar

Will you implement a graph that tracks raw pp with time? This way we can see the actual growth and not just the rank fluctuations.

July 26, 2015

Will have to see. The rank graph got introduced back with ppv1 where pp itself also fluctuated with what other people did, so it made more sense at that time.

Liked by: DroppedBass

Is there any reason that the osu client rounds the difficulty of each map to assign it to "4 stars" , "5 stars" etc.? I've always wondered that.

July 26, 2015

The reason is, that when grouping you don't want a shitton of different groups. If you just sort by difficulty and keep some other grouping mode active, you won't have any quantization.

What do you think of tuning the accuracy part of PP calculation to depend on something like UR in order to not punish high accuracy players on low OD maps? Obviously it's not a perfect solution as UR is harder on maps with trickier rhythm patterns but I think the current system is outdated.

July 26, 2015

For that to work UR would have to be stored in the database. For this to be done either the client just uploads his UR and the server needs to trust the client (not gonna happen, because that'd allow for pretty much undetectable pp cheats) or the client sends his whole replay, the server analyzes the replay, computes the UR and then saves the UR. This would require a serversided replay playback system which is not trivial to add. It'd also add extra load to the server should it already be programmed. Lastly for old scores it'd be impossible to recover UR, so pp would be unfair towards them.

Liked by: DroppedBass

If you don't mind my asking, why is a difficulty per point in time graph for every song and a table with where each miss was made in time for every score (AKA per-Hitobject data) so hard to implement? Am I oversimplifying the issue? I am not good with computers

July 26, 2015

Because it would mean storing multiple kilobytes per score rather than just multiple bytes. That's quite a few orders of magnitude of more storage. As is we already have maaaany gigabytes of raw score data, and multiplying that by a thousand doesn't seem like a good idea.

Assuming we would do that however, computing pp would also require additional time. Difficulty data would have to be streamed into memory when required instead of readily having it all in RAM like right now and thus pp calculations, especially re-calcs when the algorithm changes, would take a long time.

So much for the technical difficulties. Of course it is also difficult to find an algorithm rating map difficulty properly in the first place.

Liked by: DroppedBass

i can already tell you as an experienced mania player that you wont be able to make 10* in mania impossible and 7* playable for the best players like in osu!std. you could only scale the pp you get from maps a bit down. pls ask experienced mania player for help.

July 26, 2015

And... why would that not be possible? I could make 7 star maps the easiest shit in the world and 7.1 star maps completely and utterly impossible to play, by simply squishing the current 0-20 star range into 7-7.1 stars. Not like that would be a good idea.

Liked by: BJ DroppedBass

Why are you trying to make the game modes have a star rating that scales the same when the modes themselves are vastly different? In mania you have multiple inputs, rather than just one (aiming) so it's obvious the skill ceiling will be way higher, even though it's actually harder for beginners.

July 26, 2015

You don't make any sense. How do you even define skill ceiling? No matter how simple the gamemode is you throw at people, they'll push it as far as they are capable, period. If a gamemode is inherently simpler, then the boundaries are pushed further than in a more complicated gamemode. It does categorically not make sense to compare things like osu! standard and osu!mania, so nothing is lost by making star ratings not depend on each other.

Your argument itself about aim being just a single input is flawed, too. Yes, you can interpret 2d aiming as a single input, but it's completely and utterly different to a binary input such as a key (pressed or not pressed). You can't just go and say they are the same. They are not.

All these comments about "mania being harder because more keys" are just elitist bullshit. No matter what game you make, as long as you have an equal amount of people trying equally hard it will be of the same difficulty to reach the top.

Liked by: Remyria Tatch dcv Psychic Scarecrow DroppedBass pyon VnSoul Rafacar

http://ask.fm/Tom94/answer/131019614158 what does " MANIERA being 8 stars while the top 50 is filled with S scores isn't okay." have to do with anything... ofcourse it would have alot of S ranks first because it will be aimed alot, second because the players can actually play 14 star maps : )

July 25, 2015

The star system is meant to scale such that 10 stars is impossible to play for literally everyone. Players being able to play over 10 stars - excluding gimmick maps such as tag4 with absolutely no drain in standard - clearly shows that star scaling is messed up. Similarly 8 stars are meant to indicate something that the vast majority of players can absolutely not play properly.

Liked by: Try Harder DroppedBass Shimox Supreme-

If you scaled the pp to the top mania players, which are capable of S-ing 11-12 star maps in the current star rating (getting 3000-4000 pp from each), would mean that new/intermediate players get extremely little pp and play 0.3-1.5 star maps. Mania is already the hardest to get pp on for most peopl

July 24, 2015

The star ratings (and pp) can get squished together more at the higher ranks / difficulties while retaining the scaling for weaker players and easier difficulties. Just look at the pp top50. All modes have a reasonable difference from #50 to #1 whereas in mania is just flatout starts at 9400pp and barely rises by 1300. To be fair, that can also be attributed to almost no actually "hard" mania maps being ranked.

Liked by: DroppedBass

But relative to osu standard players, mania players are more skilled, as most of them have been playing it for a much longer time.

July 24, 2015

Not twice as skilled as a 1200 pp score would suggest. That's not how it works. Even if they are more skilled, there is no way to even subjectively gauge this as they are completely different gamemodes. Thus I'll treat each game mode in isolation and balance it around the best players existing there.

Liked by: Psychic Scarecrow DroppedBass Jin

But what if the top mania players are just that good in respect to the star rating system?

July 24, 2015

That statement doesn't make any sense. The star system is not absolute. If all the stars and hence pp are being scaled down, the ranking wouldn't change at all, the pp and star numbers just would be smaller.

Liked by: DroppedBass

http://ask.fm/Tom94/answer/130923524814 But the top mania player said it was correctly balanced? I mean, he surely knows more about mania than you. And I doubt he wants it just for the pp, he could take #1 if he wanted anytime.

July 24, 2015

You sure you linked the right answer?

And yes, I am aware, that MANIERA is actually the hardest o!m map with respect to the o!m scoring system at the moment. Other parts of the mania pp algorithm also need adjustment, for instance some insane low-bpm jack weights. Furthermore mania gives way higher star ratings to maps than other modes due to a previous lack of properly hard maps, so overall pp and star ratings need to be pushed down.

MANIERA being the hardest map is okay. MANIERA being 8 stars while the top 50 is filled with S scores isn't okay. Same applies to imperishable night and other maps.

Liked by: lukass pyon _bloo DroppedBass

http://ask.fm/Tom94/answer/130956687054 are you sure? because I can still feel an awful delay input on my Tablet even while playing on fullscreen (been using Tablets for drawing for 14 years so I can "feel" any delay, on Windows/Border mode it's unplayable) maybe it's something on the drivers side?

July 23, 2015

Maybe you still have VSync or a low frame limiter enabled. Those also cause a noticeable delay.

Liked by: DroppedBass

http://ask.fm/Tom94/answer/130941828814 By average difficulty, you mean the average of the difficulty at each note, rather than the average of every point in time, right? Wouldn't want to penalize a hard map for having a long slow section with few notes (and thus little influence on final accuracy).

July 22, 2015

Yup. But even using the flat average is not yet decided yet. Making, progress, though.

Liked by: DroppedBass

Tom, under what circumstances I would still have delay input while fullscreen on 8.1?

July 22, 2015

Under none. If you have the fullscreen button toggled you don't have input delay anymore. Borderless windowed doesn't count as fullscreen.

Liked by: DroppedBass

Tom Long Dong

Nothing here yet