forum

[Proposal] Timing #/8-signatures; Merge two timing-rules

posted
Total Posts
29
Topic Starter
Bonsai
Hi-ho, Kermit the Frog Bonsai the BN here, speaking on behalf of the timing-team of the UBKRC!

Every now and then, songs are being mapped that have a time signature that is not divided into quater-notes (e.g. 3/4, 4/4, 7/4, ..) but into eighth-notes (e.g. 6/8, 7/8, ..). Since the current editor does not support those #/8-signatures and we don't know whether this will change in the near future, we are proposing an addition to a rule of the ranking criteria (changes being underlined):
  1. Uninherited timing points must be used to accurately map the song's time signatures. If an incorrect time signature lasts for more than one bar, an uninherited timing point must be added on the next downbeat to reset the time signature. For #/4-signatures unsupported by the editor, metronome resets or editing of the .osu file are acceptable. For other unsupported time signatures, refer to this exemplary chart, and see this guide for further information.
Both the chart and the guide mentioned in the rule will be hosted on the wiki, separate from the ranking criteria. For now, I have included the guide in the box below, the chart can be found at the bottom of that guide.
box

Timing songs with #/8-signatures

Hi-ho, Kermit the Frog Bonsai the BN here!

Every now and then, songs are being mapped that have a time signature that is not divided into quater-notes (e.g. 3/4, 4/4, 7/4, ..) but into eighth-notes (e.g. 6/8, 7/8, ..). Since the current editor does not support those #/8-signatures and we don't know whether this will change in the near future, an addition has been made to a rule of the ranking criteria:
  1. Uninherited timing points must be used to accurately map the song's time signatures. If an incorrect time signature lasts for more than one bar, an uninherited timing point must be added on the next downbeat to reset the time signature. For #/4-signatures unsupported by the editor, metronome resets or editing of the .osu file are acceptable. For other unsupported time signatures, refer to this exemplary chart, and see this guide for further information.
To make that chart more understandable and state more precisely what is meant, I will now explain a bit of the relevant music theory behind this, and give examples of how it translates into osu!. I have prepared three mapsets for you to download, which I will reference during this guide.


Music Theory


Generally speaking, a measure contains a certain amount of beats. In musical sheet notation, one beat usually equals one quater-note ♩, and one quater-note ♩ is divided into two eighth-notes ♪♪. Hence, the term "Beats Per Minute" (BPM) describes the tempo of a song by stating how many quater-notes ♩ are being played in a minute. In osu!, beats are represented by the white ticks in the timeline, and eighth-notes ♪ are usually represented by red ticks. The start of each measure is indicated by a big white tick.

At least that's how it works for #/4-signatures. The fundamental difference to #/8 -signatures is that a beat doesn't always contain just two eighth-notes ♪, it can also contain three ♪s! A good example for this is the difference between 3/4- and 6/8-signature: Both of them contain six ♪s , but while 3/4 splits those into three groups of two (♪♪+♪♪+♪♪, [2+2+2]) by accentuating the first note of each pair, 6/8 splits them into two groups of three (♪♪♪+♪♪♪, [3+3])! These "groups" are also called "subdivisions". Time signatures only containing subdivisions of three eighth-notes ♪ (like in 6/8) are called "compound", time signatures containing subdivisions with variable amounts of eighth-notes ♪ are called "complex" or "irregular" (e.g. 7/8, more on that later). When a certain pattern made up of different time signatures repeats regularly, it is called "mixed meter".

comparison of subdivisions between a 3/4- and a 6/8-measure


Now to give you a first example of this in osu!, look at the "3/4"-difficulty in the provided mapset of Leonard Bernstein's "America" - In this difficulty, I timed everything in 3/4-signature, meaning that every measure is divided into three beats by white ticks. When you turn the BeatSnapDivisor to 1/2, every beat is split into two by the red ticks in the timeline - A red tick representing an eighth-note ♪. This would work perfectly fine for a 3/4-song, but if you listen closely you can hear something special about this song: While the overall tempo stays (approximately) the same, the rhythmical grouping of ♪s switches almost every measure - The first measure of the mapped part at 00:03:662 is split into [3+3], as it has only one more accentuation on the red tick at 00:04:232 -. The second measure, which starts at 00:04:801 is grouped into [2+2+2], as it has two more accentuations on the white ticks at 00:05:181 and 00:05:561 -. In other words, all measures are equally long (6♪s), but half of them only have two beats (-> 6/8-measure) while the other half has three beats (-> 3/4-measure), making this a mixed meter! There are different ways to handle this in osu! that all have their up- and downsides, which I will showcase in the next segment.


two measures of "America", the first is in 6/8, the second in 3/4

Methods of timing songs with #/8-signatures


A) Emulating #/8-signatures

When talking about the 3/4-difficulty in Leonard Bernstein's "America", I mentioned that one of the beats lands on a red tick instead of a white tick at 00:04:232 because that measure is split into [3+3], which makes it a 6/8-measure. In order to "fix" where the beat lands, the BPM has to be multiplied with 0,666. Furthermore, the Time Signature shown in the TimingSetupPanel has to be set to 2/4 - This is currently not possible in the editor, but it is allowed to change it in the map's .osu-file. After doing that, the BeatSnapDivisor has to be changed to 1/3 so it splits one beat into three ♪-notes, which are then represented by the purple ticks. The result of this can be seen in the 6/8-difficulty.



the timelines of a song timed in 3/4 (upper) and in 6/8 (lower)

For a simpler example, take a look at the provided set of Queen's "We Are The Champions", which is a song purely in 6/8-signature. When you look at the timing-tab of the 6/8-difficulty, you can see that the small white ticks in-between the measures land precisely on the song's actual beats, and all the eighth-notes ♪ are mapped by using the 1/3-BeatSnapDivisor.

This method is rhythmically the most accurate in osu!. Since white ticks are always representing a beat, both the pulsating osu!logo in the SongSelectMenu and the Nightcore-soundeffects are always on point and properly convey the song's "feeling". It is very easy and practical to use this method for songs in signatures like 6/8 but also 9/8, 12/8, etc. - Many of those signatures are more common than one might think, for example Nekomata Master - Avalon no Oka is a song in 12/8-signature, which has intuitively been timed as 4/4-signature and mapped by using the 1/3-BeatSnapDivisor. This is why we are generally enforcing this method of timing for all songs in #/8-signatures where the beats are consistently divided into three eighth-notes ♪♪♪ (e.g. it would be enforced on Queen's We Are The Champions, whereas it would not be enforced on Leonard Bernstein's America because it is in mixed meter).

But this method can also cause some problems: First of all, the BeatSnapDivisor does not have as many layers of "triple-snaps" (3,6,12) as it has of "duple-snaps" (2,4,8,16). This can make it impossible to snap potential fast rhythms that might occur in some songs, which you might want to map in higher difficulties.

Secondly, there are more complex #/8-signatures than those that have multiples of 3 in the upper numeral of the signature. For example, while a 6/8-measure is always formed as [3+3], you can get measures with "irregular" beats by combining [2]s and [3]s. For example, a 7/8-measure can be formed as [3+2+2], [2+2+3] or [2+3+2]. If you still want to have the beats timed accurately with white ticks, you'd have to use multiple timing-sections within one measure to follow both the "fast" beats (the [2]s) and the "slow" beats (the [3]s). This can be seen in the 7/8-difficulty of "Le Mystere Des Voix Bulgares - Erghen Diado". As you can see, I had to set two timing-sections per measure (marked with bookmarks), and because of the different BPM-values I also had to change the SV-multiplier of half of the sections. Moving around in this map is rather tedious too, because while all of the mapped notes are equally long, half of them are snapped as 1/2-notes and half of them as 1/3-notes. This is why we are not enforcing this method for songs with complex time signatures (or when rhythms become unsnappable), and allow the alternative method used in the 3.5/4-difficulty:


B) Treating them as #/4-signatures

This method of timing is generally very simple: ♪-notes are always represented by the red (and white) ticks from the 1/2-divisor. This means that even when the song changes signature, the BPM remains the same and the BeatSnapDivisor doesn't need to be changed to 1/3 in order to map ♪-notes. Navigating in the timeline is easy, and the BPM that is shown in SongSelect and on the website is generally an accurate indicator of the map's usual tapping-speed ("I can singletap/stream 200 BPM" always refers to singletapping red ticks / streaming blue ticks in that tempo). However, for complex time signatures this might not always accurately follow the song's beat, and never accurately follows the beat of compound time signatures - For example, when looking back at the 3/4-difficulty of Queen's "We Are The Champions", you can see that the small white ticks in-between the measures do not land on the song's beats.

This is the go-to-method for timing songs with complex time signatures, as it significantly reduces the timing-effort and ensures easy handling of the timeline. You are always allowed to use the more accurate first method, but only forced to use it when the song's beat-value is consistently three eighth-notes ♪♪♪ (e.g. both of the provided timings for Erghen Diado are acceptable, as it has inconsistent beat-value [2+2+3]). For mixed meters, the start of every measure is required to land on a big white tick. Compound and complex #/2- or #/16-signatures are to be dealt with similarly.



tl;dr
Shoutouts to dsco for helping a lot with everything <3

Furthermore, we are proposing to merge these two rules as follows:
  1. An object which is wrongly snapped due to passing through or ending slightly before a new uninherited timing point must have its end snapped within the new timing section. For spinners and osu!mania long notes, this can be achieved through dragging an object’s tail in the timeline. For sliders, this can be achieved through slider velocity manipulation or editing of the .osu file.

These proposals will be up to discussion for one week and close on 14th of February for revision.
pishifat
I approve this message!
Arzenvald
new timing when
Xenok
I approve this post. It would also be very cool if osu supports more complex timing signature, hope to see this comming with lazer :)

I would also add something, musically, a 6/8 should always be composed by two dotted quarter note, making [3-3], you say here it can be [2-2-2] (you also said [3-3-3] which looks like a typo) https://i.imgur.com/uJRHaj7.png
GoldenWolf
Yes Yes Yes! Sounds very good to me, totally approve of this.
Topic Starter
Bonsai

Xenok wrote:

I would also add something, musically, a 6/8 should always be composed by two dotted quarter note, making [3-3], you say here it can be [2-2-2] (you also said [3-3-3] which looks like a typo) https://i.imgur.com/uJRHaj7.png
Oh thanks, both of those were a mistake, fixed!
-Mo-
Definitely needed, and nicely written.
Xenok
I would also add it's "musically wrong" to say that America is mixed between 6/8 and 3/4. Actually, the song is only composed in 6/8 but composed with notes that sometime could tend to look like 3/4. We can see this in this music sheet for example: https://s3.amazonaws.com/static.musesco ... core_0.png

But well I think it's not a problem in the post to show the difference between 6/8 and 3/4 since it sound like a 3/4, but I think it's a good think to make things more precise!

Sorry for being annoying... :cry:
Topic Starter
Bonsai
Well, wouldn't that kinda imply that 6/8 can be [2+2+2] since every second measure is subdivided like that? :P "Mixed meter" is actually a proper term, and wikipedia describes it like that too! Those sheets you've got are just by someone who arranged it and didn't care about indicating it like that ¯\_(ツ)_/¯

No worries about being annoying, discussion is what this thread is for! ;)
Xenok
Hmm yeah it looks like this song is kind of an exception, I never saw that time signature before LUL

Yeah, so on osu, we should also alternate between 2/4 100 bpm and 3/4 150 bpm? Don't look kinda annyoing sv wise?
Topic Starter
Bonsai

Xenok wrote:

Yeah, so on osu, we should also alternate between 2/4 100 bpm and 3/4 150 bpm? Don't look kinda annyoing sv wise?
Ah right, we didn't actually consider mixed meters very much and I assumed it would fall into the complex-category and doesn't need to be timed like that, but we (read: dsco) found a ton of loopholes with that right now. The two options to avoid all of those are either 1) requiring them to be timed this way, which I find annoying for cases like this too, or 2) adding the following sentences at adequate places in the guide:
When a certain pattern made up of different time signatures repeats regularly, it is called "mixed meter".
For mixed meters, the start of every measure is required to land on a big white tick.
which we will do for now, along with mentioning that "America" is an example of that. If anyone knows some examples of mixed meters where this could be problematic, please bring them to us!
Kagetsu
so uhh i have some concerns

the proposal assumes that all irregular time signatures divide beats in the same way?

the proposal doesn't take care of the main thing it should be addressing, the metronome hits. a lot of those would be left omitted with this proposal (the whole purpose of the upper number is to denote how many times the metronome hits within a certain measure, as far as my understanding goes, please correct me if i'm wrong)
the metronome should not follow the beats only, because that's a thing that depends on different factors, and it isn't important for osu really, if anything it helps with note placements on the timeline and stuff like how to divide patterns and such.

you guys are implying that the metronome should emit irregular pulses which is wrong as far as i understand. the whole purpose of a metronome is to have a regular pulse aligned with the music.

wondering why isn't doubling bpm a valid solution?
chainpullz

Xenok wrote:

I would also add it's "musically wrong" to say that America is mixed between 6/8 and 3/4. Actually, the song is only composed in 6/8 but composed with notes that sometime could tend to look like 3/4. We can see this in this music sheet for example: https://s3.amazonaws.com/static.musesco ... core_0.png

But well I think it's not a problem in the post to show the difference between 6/8 and 3/4 since it sound like a 3/4, but I think it's a good think to make things more precise!

Sorry for being annoying... :cry:
Xenok is correct that the time signature isn't determined by how the beats break down in a certain measure. The score he linked to is very indicative of many 6/8 songs so it shouldn't be looked at as an exception. It is much less common for music to switch time signatures every measure actually. It's *usually* pretty obvious whether something is 6/8 or 3/4 based on the emphasis patterns/style of the music if you have the ear for it.

Kagetsu wrote:

the proposal doesn't take care of the main thing it should be addressing, the metronome hits. a lot of those would be left omitted with this proposal (the whole purpose of the upper number is to denote how many times the metronome hits within a certain measure, as far as my understanding goes, please correct me if i'm wrong)
You're about half right. Time signature as a whole along with tempo marking (i.e., BPM) typically is what determines how many metronome hits are in a measure. 6/8 is almost exclusively counted in 2 and then further subdivided as triplets (i.e., tri-ple-let tri-ple-let) even in measures consisting of 3 quarter notes. 3/4 on the other hand is either counted in 3 or in 1 depending on the tempo marking (i.e. waltz is in 1 typically). There are a variety of ways that the tempo marking could be formatted to include this information (i.e., the tempo marking might be "fast waltz").

If osu! properly took timing into account then 6/8 would look like 2/4 but when you hold down the control key instead of hearing 1/2 snaps your would hear 1/3 snaps. 3/4 would need the option to specify whether it should be counted in 1 or in 3 with the former subdividing by 3 and the latter subdividing by 2.

Edit: I misinterpreted the mixed measure as not being something taken directly from a score (it does seem to show up on most scores for America actually). It must not be that common in concert band or symphony orchestra as I was unfamiliar with it but I was always too lazy to do pit orchestra when I had the chance in high school so maybe it shows up more often there.
Topic Starter
Bonsai
Discussed most of Kagetsu's post with them privately as there was a bit of a misunderstanding of how proper metronomes work, here's a little summary, additionally evolving into the reasoning of why we don't deem double-BPM valid:

A "traditional" metronome only makes one type of tick-sound which represents one beat, and doesn't really have anything to do with the time signature that you want to play in that tempo - When a metronome is set to 60 BPM it makes one tick per second, if you're playing 3/4 in that tempo you take three ticks and mentally sub-divide each tick into two, if you're playing 6/8 in that tempo you take two ticks and sub-divide each into three ticks - Both signatures have the same amount of eighth- or quater-notes, but those notes don't have the same length, only the beats, and that's all what a metronome cares about. That's what the metronome in osu! does too, if you set the BPM to 60 you have one white tick per second and will hear one tick-sound per second in the timing-tab, only when you hold ctrl it will further divide that into two or three ticks depending on what the BeatSnapDivisor is set to. A single metronome can't properly indicate irregular time signatures, because irregular time signatures aren't limited to one tempo.

So this should make clear why using multiple BPM-sections is theoretically the musically most accurate method of timing irregular signatures in osu!. But since that's a relatively high effort, we only enforce the "simplified" version of sticking to one of those tempi, which is then consistent even if the song changes signature.
Now one of the reasons for not deeming double-BPM valid is obviously that it completely incorrect musically, since it uses neither of the tempi that a song actually has, and messes up the meaning of both note-value and beat-value. But some more osu!-related reasons are that 1) it also loses any indication of gameplay, as it doesn't even say how fast the actual "tapping-speed" of the song would be, and 2) it would make songs that switch between signatures be timed even more inconsistently, as you'd have 100 BPM for one part, then 200 BPM for another and then again 100 BPM, which messes up the SV along the way too. On average, the effort for using doubled BPM would be lower (though it could be higher in some cases too), but at the cost of not only losing any musical significance, but also losing an< significance for osu!. It wouldn't represent tempo in a consistent way, and wouldn't give the player any indication of what they can expect from the song. Basically, BPM would be completely meaningless because it would have a different meaning for different songs (or even for the same song when a signature isn't clearly identifyable), so at that point it wouldn't make much sense to have any timing-regulations in the first place.
Kagetsu
@chainpullz i think the following lines cover the thing i was referring to

Bonsai wrote:

A "traditional" metronome only makes one type of tick-sound which represents one beat, and doesn't really have anything to do with the time signature that you want to play in that tempo -
i gotta disagree with this, first of all why do "traditional" metronomes matter at this point? i mean we're in 2018 and i'm pretty sure electronic metronomes exist and are capable of representing compound or even complex time signatures.

Bonsai wrote:

When a metronome is set to 60 BPM it makes one tick per second, if you're playing 3/4 in that tempo you take three ticks and mentally sub-divide each tick into two, if you're playing 6/8 in that tempo you take two ticks and sub-divide each into three ticks - Both signatures have the same amount of eighth- or quater-notes, but those notes don't have the same length, only the beats, and that's all what a metronome cares about.
yes, but you gotta realize the reason why representing the subdivisions with the metronome on simple time signatures (such as 3/4 or 4/4) isn't necessary: the beat on those is always a quarter note, so the notes always get grouped the same way, whereas for x/8 signatures, the way of grouping may be completely different from one song to another, you can have beats that last a dotted quarter note, for other cases it may be 2 eighth notes, there even exist 3/8 which uses eighth notes as the main beat. hence why electronic metronomes do mark the subdivisions for those time signatures.
also regarding the thing you're saying about the metronome should be taking care about the beats only , i think it's incorrect as the metronome is supposed to help you maintain a constant pace aligned with the music so i'd say the metronome should mark the subdivisions too (although quieter than actual beats)

Bonsai wrote:

That's what the metronome in osu! does too, if you set the BPM to 60 you have one white tick per second and will hear one tick-sound per second in the timing-tab, only when you hold ctrl it will further divide that into two or three ticks depending on what the BeatSnapDivisor is set to.
this point was already explained, it just happen to be like that because osu uses simple time signatures only, which take quarter notes as the main beat, so i believe subdivisions doesn't have to be there to the metronome be aligned with the music

Bonsai wrote:

A single metronome can't properly indicate irregular time signatures, because irregular time signatures aren't limited to one tempo.
so uh i don't understand why you say that, i mean if it was like that why would we even bother to ask developers to add complex time signatures into the editor?
as far i know there already exist electronic metronomes that can actually represent these type of time signatures. they use loud sounds to represent beats and quieter ones for subdivisions which helps a lot when it comes to play music in my opinion.
i can also tell that i've already searched from a ton of different sources finding always the same results from youtube videos, google search others forums, even music making related software such as cubase or guitar pro deal with those time signatures in a similar way, so why would osu use something different?

Bonsai wrote:

So this should make clear why using multiple BPM-sections is theoretically the musically most accurate method of timing irregular signatures in osu!. But since that's a relatively high effort, we only enforce the "simplified" version of sticking to one of those tempi, which is then consistent even if the song changes signature.
this wouldn't be a problem if we were with the doubling equivalency method as there would be always a subdivision hitting the beats no matter what

Bonsai wrote:

Now one of the reasons for not deeming double-BPM valid is obviously that it completely incorrect musically, since it uses neither of the tempi that a song actually has, and messes up the meaning of both note-value and beat-value
actually doubling the bpm and making all the notes half of the duration is just an equivalency, it would be still readable if you were to read a music sheet (although it would be more complicated because there wouldn't be an indication as to where the notes should be emphasized or how they are grouped), but it's none of osu's business honestly, if it was for this, the equivalence for almost all 1/3 based songs would be also labeled as incorrect as you would be converting a compound time signature into a simple one mapped with tuplets.

Bonsai wrote:

1) it also loses any indication of gameplay, as it doesn't even say how fast the actual "tapping-speed" of the song would be
i'd say it's the complete opposite of what you're stating, as for example this map's listing says it's only 72 to 108 bpm but the tapping speed required to play it is actually a lot higher than that

Bonsai wrote:

2) it would make songs that switch between signatures be timed even more inconsistently, as you'd have 100 BPM for one part, then 200 BPM for another and then again 100 BPM, which messes up the SV along the way too. On average, the effort for using doubled BPM would be lower (though it could be higher in some cases too), but at the cost of not only losing any musical significance, but also losing an< significance for osu!. It wouldn't represent tempo in a consistent way, and wouldn't give the player any indication of what they can expect from the song. Basically, BPM would be completely meaningless because it would have a different meaning for different songs (or even for the same song when a signature isn't clearly identifyable), so at that point it wouldn't make much sense to have any timing-regulations in the first place.
well that's something that can be addressed with some rulings too, i don't think it would that difficult to so, also the method you're mentioning of halving/doubling because of time signature changes had happened a lot of times, refer to maps like this or this where you could theoretically change the time signature instead of changing tempo.

to end with i'm just gonna say that after our talk yesterday, i'm not against the current proposal for regular time signatures (anything that can divided into beats of the same duration) as it makes a lot of sense and align with the music anyways. for irregular ones though, (5/8 7/8... etc) it's unpractical and not really intuitive to follow when playing/listening to the metronome
the fact that adding multiple red lines shifts the note placement on the timeline should be considered too, at least in my opinion.
chainpullz
I think the bigger concern with double bpm in my mind is that it screws up the metronome with NC. At least in the genres I am most familiar with, a tempo marking of prestissimo isn't typically used as the tempo marking for the entire piece. Most songs that you would want to use double bpm for are actually cut time (e.g., 2/2 time signature). The tempo marking then usually indicates the bpm in terms of half notes instead of quarter notes since half notes get the beat in cut time. Hence with NC you end up with the added effects twice as fast as they should be.
dsco
timing ♩•=60bpm 6/8 (as an example) as 180bpm 6/4 poses many problems and is extremely musically inaccurate. timing it as 60bpm 2/4 is not correct either, as its quite obviously not 6/8, but it bears a much stronger musical relation to the way the music is played and written. a 6/8 or a 12/8 drum beat will have a kick on the first eighth note and then a snare on the 4th, (or kick on 1st and 7th, snare on 4th and 10th for 12/8), just as would 2/4. in fact, nearly every song in 12/8 would be instinctually mapped as 4/4 in 1/3 snap divisor by any mapper. to imply it should be timed as 1.5x or 3x the bpm with a 1/2 divisor would not only sound awful in the editor but be extremely musically inaccurate. further, the tempo for #/8 time signatures is very rarely given in terms of the eighth note, it is given in terms of the pulse of the music, for 6/8, for example, it would be given in terms of a dotted quarter-note (3 eighth notes, hence the reason the tempo would be chosen).

you mention that DAWs are capable of producing these complex/compound/irregular time signatures. this is completely true. in fact, if you open any DAW, select 120bpm and select 7/8, the 120bpm will be referential to a quarter note still. if we were to time it as double bpm, 120bpm 7/8 in a DAW is now 240bpm in osu? there is no musical clarity in timing songs this way.

here are the problems that doubling bpm presents:
for music that switches between 4/4 and 7/8 or 5/8 or any other #/8 signature, you will have two bpms for the exact same song and tempo, even if the music is quantized. this makes 0 sense whatsoever and there is no way you can argue that part of said song is 180bpm and the other part is 360bpm. this is extremely common to math rock, and very common in genres like jazz as well.

doubling the bpm results in incredibly inaccurate bpm readings. this is not only the case for #/8 time signatures, but even more-so #/16 time signatures, which are common as pickup measures in jazz and some classical music. quadrupling the bpm for these cases is absolutely mad. for the final movement of Stravinsky's 'The Rite of Spring,' an extremely important classical piece, you would have to quadruple the bpm for long sections of music, resulting in a completely ludicrous bpm that has a very deteriorated relation to the actual pace of the music. for a jazz example, take 'Big Eater' by The Bad Plus. this song has pickup measures of 3/16 after every 'chorus' section, with mostly 7/8 but also 3/4 and 4/4 sections in the music. it would be absurd to time parts of this song as 120bpm, some as 240, and some as 480. or take this song by tigran hamasyan, where hypothetically you would time the entire song as 560bpm (it's around 140bpm). surely you can see the complete lack of accuracy in doing this.

as for adding red ticks shifting the bpm, i agree it could be added to the guide (with such songs doing it back to beginning prevents this entirely), though its already covered in songs requiring being timed perfectly, if you mean in the ranking criteria.

hopefully this addresses your concerns and explains the reasoning that this timing is necessary, and most accurate musically.
chainpullz
Does anyone actually know the precision used for timing points by the way? I know the editor only displays out to the nearest milisecond but the .osu file format seems to allow higher precision. I know this is definitely respected for SV where you can specify decimal points out to whatever bit limit osu uses (either 32 or 64). It will only display 2 significant digits for SV in the editor but it uses the higher precision when computing slider length.

I managed to hit the space bar and set a red line down that seems to have had a decimal time value the other day from the look of things. If it does indeed allow floating point precision then you could use text editing to eliminate the drift due to rounding when putting down additional red lines.
Kagetsu
some clarifications and ideas that i feel would explain the way i see this whole thing and i think need discussion:
  1. the metronome is meant to mark regular pulses. so that it helps you mantain a certain pace. using metronome resets (irregular pulses) defeats the purpose of having a metronome in the first place, since the pulses wouldn't be regular, it would make following music a difficult task to achieve
  2. the way a metronome works doesn't have much to do with music theory. no matter what the rhythms of the song are, the metronome should always follow the same regular ticking
  3. you won't find metronomes marking irregular pulses only. even if some metronomes do mark beats with stronger sounds, they still mark the subdivisions on /8 time signatures. this happens on /16 or whatever 2^x the bottom number is
  4. the metronome doesn't depend on the song, so any song using a specific time signature should allign with the metronome. this means that if i take two songs using the same time signature, they both should use the same type of metronome, accentuations within the measure are meant to be executed by the player, not the metronome.
  5. in any music making software, if you double the bottom number of a time signature, the metronome would sound as twice as fast. this is because it does mark the subdivisions. note that this doesn't necessarily imply that the song pace is faster.
    now i already mentioned that the current proposal covers regular time signatures just fine so that shouldn't be a thing to debate about. but just to make it clear, we're here talking about metronomes only. the purpose of doubling method is not to denote a faster pace, but rather to emulate how a metronome (not music) would sound on those time signatures, especially on irregular ones.
it would be appreciated if you use the numbers i put to reply, so that it's easier to discuss
Topic Starter
Bonsai
1 2 3 4 5 german counting:
Who says that the metronome-concept you're describing is relevant in the first place? The osu!metronome™ never really helps anyone with maintaining a certain pace at all, because you can't hear it in the editor (except the timing-tab where you usually don't spend much time in) and you can't hear it during gameplay (except for Nightcore-mod, which sounds terrible with doubleBPM in almost every case). The field where you put in the value that you are proposing to double is literally titled BPM for "Beats per Minute", if you double that value it's not the correct BPM anymore. I don't see how these characteristics that you are demanding of a metronome are relevant for osu! at all.

(3: You can mark subdivisions in osu! without doubling BPM by pressing ctrl)

(4: A metronome that marks beats with stronger sounds and subdivisions with weaker sounds does depend on the song, and the osu!metronome™ is one of those. Also, you can just write irregular time signatures as "additive signatures", in which case they are not written as 7/8 but as (2+2+3)/8, which is just a stylistic difference. Thus, 2+2+3 and 3+2+2 would be different signatures, and hence require different metronomes)
dsco
@1: you seem to have some misconception that not doubling the bpm results in a less-accurate tempo to what the music would be played as. for starters, this is an incredibly subjective thing, as some people will alternate 1/2 divisor at bpms even as low as 180 and above while single tapping 180 below, and some will single tap 1/2 divisor all the way to 280.

secondly, doubling the bpm is incredibly inaccurate in many cases too. take these three songs: [1] [2] [3] [4]. are these songs 384bpm, 580bpm, 320bpm, and 345bpm? or are they 192, 140, 160, and 115bpm?

@2/3/4/5: i don't get the obsession with metronomes? i don't know where to begin in addressing this. you admit firstly that metronomes don't have much to do with music theory, and then in the same breath imply that the way music should be timed in osu should be based entirely off of metronomes? i don't see the basis for this conclusion? beats per minute is not respective to metronomes only. music is very commonly marked in beats per minute now that metronomes are widely available for practice and performance, and in the case of #/8 time signatures it's very rarely given in values of the eighth note. when you're timing music in osu, just as if you were to transcribe a piece of music, you wouldn't just go by the smallest subdivision that is commonly played (it would be ludicrous to do this for any song primarily in sixteenth notes, i.e.: stream maps) so why would you do it for music in #/8 time signatures?

beats per minute is just that. beats per minute. there are two beats in a measure of 6/8; it is a compound duple meter, meaning two beats of 3 eighth notes. all compound meters in #/8 will have the beat written in terms of a dotted quarter note. perhaps you are getting confused by the fact that a beat has multiple meanings in music, but this is all basic music theory. this is why 3/4 is different from 6/8 in feel; they have the same number of eighth notes both, but a measure of 6/8 will have two beats and a measure of 3/4 will have three. this is why music in 6/8 has the same drum beat as a 2/4 drum beat, and also why the tempo markings for 6/8 will be played in terms of a dotted quarter note. for example, if you have a 4/4 piece in adagio tempo, that is roughly 70bpm. for a 6/8 piece in adagio tempo, you would have 3 eighth notes roughly equal to 70bpm. bpm is an objective measure of tempo that did not previously exist in music due to clocks not being objective enough. it would be as if you were to mark a 6/8 piece as prestissimo tempo because each eighth note is being played much faster than the quarter notes at 4/4. similarly, if you were asked to play 6/8 at 100bpm, you would play such that 3 eighth notes are equal to one beat. or 7/8 at 100bpm, two eighth notes equal to one beat and so on. it honestly seems you are passing judgment on something you are unfamiliar with or do not have background/trained experience in.

why would you change this when marking bpm in osu?

metronomes do not always play to the lowest subdivision of the time signature anyways. take 12/8 or 6/8 metronomes for example, in compound meter. it is extremely common to practice these to 4/4 or 2/4 metronomes respectively, especially at higher bpms. in fact, when taught, they are essentially always taught as triplets in 4/4 and 2/4, because the eighth notes are more easily and comfortably played as triplets than individual beats.

just because there is a metronome that plays when you insert a bpm in osu doesn't make the bpm revolve entirely around the rate at which the metronome ticks (which can be changed simply by holding ctrl in the timing panel anyways).
Kagetsu

Bonsai wrote:

1 2 3 4 5 german counting:
Who says that the metronome-concept you're describing is relevant in the first place? The osu!metronome™ never really helps anyone with maintaining a certain pace at all, because you can't hear it in the editor (except the timing-tab where you usually don't spend much time in) and you can't hear it during gameplay (except for Nightcore-mod, which sounds terrible with doubleBPM in almost every case). The field where you put in the value that you are proposing to double is literally titled BPM for "Beats per Minute", if you double that value it's not the correct BPM anymore. I don't see how these characteristics that you are demanding of a metronome are relevant for osu! at all.
i don't even know what to say, if you think the metronome isn't important at all then i don't know why we are discussing this proposal in the first place. why not set any time signature if the metronome isn't important at all?

Bonsai wrote:

(4: A metronome that marks beats with stronger sounds and subdivisions with weaker sounds does depend on the song, and the osu!metronome™ is one of those. Also, you can just write irregular time signatures as "additive signatures", in which case they are not written as 7/8 but as (2+2+3)/8, which is just a stylistic difference. Thus, 2+2+3 and 3+2+2 would be different signatures, and hence require different metronomes)
doesn't change the fact that they would still mark the subdivisions which is the point i'm trying to convince you about though

dsco wrote:

@1: you seem to have some misconception that not doubling the bpm results in a less-accurate tempo to what the music would be played as. for starters, this is an incredibly subjective thing, as some people will alternate 1/2 divisor at bpms even as low as 180 and above while single tapping 180 below, and some will single tap 1/2 divisor all the way to 280.

secondly, doubling the bpm is incredibly inaccurate in many cases too. take these three songs: [1] [2] [3] [4]. are these songs 384bpm, 580bpm, 320bpm, and 345bpm? or are they 192, 140, 160, and 115bpm?
so uh i think i already addressed most of the points you're stating now.
as for the songs you posted you could virtually use any bpm i want, the song pace doesn't only rely on a bpm value, it needs a time signature for context so you can't just say "this is x bpm" because it depends on that factor too. theoretically i could arrange those songs in any bpm i want and it would sound exactly the same as long as i do convert everything accordingly (time signature and note values).

second paragraph is just the same, it doesn't really explain how a metronome would actually handle those complex time signatures, in fact, under your reasoning, you couldn't even use a metronome to play a song on irregular time signatures (because it would get off after the first measure and as we already know, metronomes only mark regular pulses)

now you mention that i'm "obsessed" with metronomes and i'd have to say that it's because it's the only thing osu should be caring about, you gotta realize that the metronome ticking is totally independent from music and it's merely used for performance/practice and as such, doesn't have any impact on music. (hence why i don't really care if the bpm gets doubled/halved, as long as it's synchronized with music it's okay for me)

if i were to take these both pieces of sheet music:

it's obvious that they're equal, meaning that musicians would produce the same rhythms no matter if reading the first or the second one. the accentuations are the same, rhythms are the same. it literally doesn't change at all.
now the question is: would the metronome sound the same for both cases? the answer is that it probably wouldn't. it would sound way slower for the 2/4 measure on most of metronomes, but that doesn't change the fact that they're completely equal (actually, this is the way this proposal is addressing those cases). well now you might ask: if metronomes do sound different for those two cases, why don't you disagree with this conversion? the asnwer is quite simple: both align with the music and makes sense musically speaking so i don't see it as problematic.
it works completely different on irregular signatures though. the metronome would sound desynchronized on things like 5/8 7/8 etc, not to mention the pointless effort it represents, the fact that the metronome sometimes wouldn't even land on beats if you don't hard enforce the second part of the rule and lastly the point that it shifts the offsets

dsco wrote:

metronomes do not always play to the lowest subdivision of the time signature anyways. take 12/8 or 6/8 metronomes for example, in compound meter. it is extremely common to practice these to 4/4 or 2/4 metronomes respectively, especially at higher bpms. in fact, when taught, they are essentially always taught as triplets in 4/4 and 2/4, because the eighth notes are more easily and comfortably played as triplets than individual beats.
i'm fully aware of this and it's why i agree with the proposal for regular time signatures, actually this is the fourth time i mention it. i have yet to see an example of what a metronome would sound like on irregular times signatures though

dsco wrote:

just because there is a metronome that plays when you insert a bpm in osu doesn't make the bpm revolve entirely around the rate at which the metronome ticks (which can be changed simply by holding ctrl in the timing panel anyways).
yes, the whole discussion is kinda pointless anyway because the metronome would only show on the main menu screen and when using nightcore mod, right?
but why not ensure it sounds as any other electronic metronome would sound? you would always expect metronomes marking subdivisions on irregular measures and i've already explained why it isn't needed for simple time.
even if that means we would be changing a number, i think it's worth the change, though i'm realizing it's just a matter of personal opinion. some people would prefer to keep the original bpm value while others (like me) would rather synchronize the metronome to the music.

i think that at this point we're just beating around the bush, everyone has provided their own point of view already and we would probably be spinning around the same matter over and over so i don't think i'll be returning to discuss this thing further. in any case i appreciate how smooth the discussion ran, it was certainly a good talk.
Topic Starter
Bonsai
Oh well, since you're feeling that it's just a matter of personal opinion, I'll use this point to conclude that topic. As everybody else on this thread seemingly agrees with the proposal, and prior talks to people dealing with these signatures gave me the same impression, we'll stick with that.
just a sidenote: the fact that you can hear subdivisions by pressing ctrl resolves most of the issues you still seem to have, that's how a metronome would sound on irregular signatures and how any other good electrical metronome would sound too

On a different note, we've noticed that the chart is actually nonsense without indicating what note-value the BPM on the left side refers to, so I've added "in ♩" to that lol

pinging pishifat to lock this thread (one day late bc we've all been busy on Valentine's Day)
pishifat
seeing as bonsai's been updating the proposal as people have shown concerns, there's nothing to really close and revise lol

so unless anyone has any last minute problems to discuss, we'll just go ahead and port this to the wiki
Okoratu
Xinnoh
is 1/5 rhythm addressed or is that another proposal
Topic Starter
Bonsai

Sinnoh wrote:

is 1/5 rhythm addressed or is that another proposal
1) no
2) no

(that topic would prolly need to involve the devs on discussion which nobody of us has been eager to do yet xd)
abraker
I think making an issue for 1/5 beat support in osu!lazer is a start
pishifat
thread moved because finalized amendment
Please sign in to reply.

New reply