[Proposal] Spread requirements based on song length

posted
Total Posts
360
show more
Lasse

Monstrata wrote:

1. Take out mp3 editing as a rule. Optionally you can make it a guideline that artificially edited mp3 extensions must sound natural/unnoticeable in order to be acceptable.

2. Remove the crossfade rule, and put it in as a guideline so people have the option to call a map out for its low quality crossfading, but not the ability to completely prevent it from being ranked. It is far too subjective to be an objective rule anyways.
agree with both of these, mainly the bold part

agree with Halfslashed's point about combining two songs, since there are a lot of cases where songs work much better if you combine them, just look at https://osu.ppy.sh/s/100348 which would be affected by this change since it technically combines two songs, it's split like this https://i.imgur.com/3fhDLNU.jpg on the album. where the actual kanshou no matenrou starts at 00:43:271 -


also with how this change is currently, if you're mapping a song that is just slightly above 3:30 you wouldn't be allowed to put breaks on your hard diff because then it wouldn't reach the drain minimum to not need a normal.
For example https://osu.ppy.sh/b/1072897 higher diffs on this are ~3:40, so lowest diff could be a hard, but then diffs below another are <3:30.
Or https://osu.ppy.sh/b/1590239 where Insane+ diffs are all above 3:30, but hard is 3:29 due to breaks.
Breaks can be very important on <Insane diffs of this length and this would discourage using them, so I think this should somehow be accounted for, but I can't really come up with a nice solution right now.

Other drain time related spread changes look nice to me
TheKingHenry

Nozhomi wrote:

There's some songs who don't support difficulties above Hard or Insane (depending of their BPM / density / etc...) who will lead this to mappers having forced difficulties who won't fit the song at all to have a spread as rules expect (example : https://osu.ppy.sh/b/481272 / https://osu.ppy.sh/b/550579). So I don't think than you should force people do to a Hard / Insane in case of the songs can't support it, or allow them to fall under Marathon rule.
Maybe I misunderstood something in your post, but I'm pretty sure this ain't forcing anyone to map those diffs or over them, but rather those diffs and under (for atleast the lowest diff); and thus in case of calmer songs having all diffs under those shouldn't be contradicting the idea of the wording there.

Halfslashed wrote:

I think an exception should be made for using two songs that were composed to transition into each other.
Same but not restricted to two, considering there's compositions consisting of more parts than that still working as one whole. Basically restricting this with any number shouldn't really be necessary. In most cases songs ain't so short that you'd need like half dozen to get to the 5 min mark (so it ain't really abusable anyways), so putting the max amount for this will only end up forbidding mapping some pieces as whole, while their length would be over the 5 mins for single-diff-mapset even without all parts.
Nozhomi
[quote="TheKingHenry"]Maybe I misunderstood something in your post, but I'm pretty sure this ain't forcing anyone to map those diffs or over them, but rather those diffs and under (for atleast the lowest diff); and thus in case of calmer songs having all diffs under those shouldn't be contradicting the idea of the wording there.

I see what you mean, and yeah agree I'm retarded.
Tho my first point is still something we should considerate.
moonlightleaf
Happy ! but if do this , e E&N diff will decrease ? (just imo

so i suggest if do TV size(<2min?) , should be map E N at least ?
-Arche

Lasse wrote:

Monstrata wrote:

1. Take out mp3 editing as a rule. Optionally you can make it a guideline that artificially edited mp3 extensions must sound natural/unnoticeable in order to be acceptable.

2. Remove the crossfade rule, and put it in as a guideline so people have the option to call a map out for its low quality crossfading, but not the ability to completely prevent it from being ranked. It is far too subjective to be an objective rule anyways.
agree with both of these, mainly the bold part

agree with Halfslashed's point about combining two songs, since there are a lot of cases where songs work much better if you combine them, just look at https://osu.ppy.sh/s/100348 which would be affected by this change since it technically combines two songs, it's split like this https://i.imgur.com/3fhDLNU.jpg on the album. where the actual kanshou no matenrou starts at 00:43:271 -


also with how this change is currently, if you're mapping a song that is just slightly above 3:30 you wouldn't be allowed to put breaks on your hard diff because then it wouldn't reach the drain minimum to not need a normal.
For example https://osu.ppy.sh/b/1072897 higher diffs on this are ~3:40, so lowest diff could be a hard, but then diffs below another are <3:30.
Or https://osu.ppy.sh/b/1590239 where Insane+ diffs are all above 3:30, but hard is 3:29 due to breaks.
Breaks can be very important on <Insane diffs of this length and this would discourage using them, so I think this should somehow be accounted for, but I can't really come up with a nice solution right now.

Other drain time related spread changes look nice to me
-Arche
When there is a loop in the song, or the album have an intro, and your song is 4:55, it's really bad to not been able anymore to extend it. R3 music box are really autistic way to extend because it have basically nothing to see with the original song, and an extend of 30 secondes aka 1/10 of the map, is not an extend anymore. Adding a loop in the song to reach 5 minutes is maybe lazy, but when its inaudible, this rules is just bad and wont help beginners mappers to rank there stuff. Making it more like a guideslines as said Monstrata or Lasse would be much better.
-Arche

peppy wrote:

Cutting shorter is done to make it more playable/suited to a rhythm game. Making longer is done to avoid mapping certain difficulties with basically no exception.
No, cutting a full version of a song to make it "tv size" is basically just been lazy to map a spread of 1:30 minutes for each diff despite a spread of 4:30 for each diff lol
pimp

moonlightleaf wrote:

Happy ! but if do this , e E&N diff will decrease ? (just imo

so i suggest if do TV size(<2min?) , should be map E N at least ?
shorter songs will not be affected, anything shorter than 3:30 minutes will still have to follow the current spread rules
Mir

timemon wrote:

I've only read the spread proposal. I quite like it, though 3:30 drain time is very hard to reach.

1) full ver maps have lots of break (because they don't have to care about drain time)
2) some full ver songs can be quite short (3:30 to 4:00) so they might even fail to hit the rules designed to help them.

Maybe adjust the drain time a bit? I think 3:00 is good for the lowest point.
I actually agree with this, full version songs are more likely to have breaks in them so having slight leeway for breaks in longer songs would be nice. Instead of 3:30 -> 4:30 -> 5:00, I would prefer 3:00 -> 4:00 -> 5:00. Not only is it more linear (there's a lot of room between 3:30-4:30 to require a hard but half that from 4:30-5:00 to require an insane) but it could also allow more breaks in 3:00+ songs.

This pretty much solves Lasse's concern I think?

Lasse wrote:

Breaks can be very important on <Insane diffs of this length and this would discourage using them, so I think this should somehow be accounted for, but I can't really come up with a nice solution right now.
The rest of the proposal I pretty much agree with.

The matter of Songs Compilation is kinda iffy. I think I agree most with what Monstrata wrote:

Monstrata wrote:

1. Take out mp3 editing as a rule. Optionally you can make it a guideline that artificially edited mp3 extensions must sound natural/unnoticeable in order to be acceptable.

2. Remove the crossfade rule, and put it in as a guideline so people have the option to call a map out for its low quality crossfading, but not the ability to completely prevent it from being ranked. It is far too subjective to be an objective rule anyways.
I wouldn't say mp3 extensions are always bad, because sometimes songs are literally milliseconds away from being 5 minutes long, and a small extension would be fine in that case if it's natural or unnoticeable. I also think crossfades can vary in quality so I agree having it as a guideline works much better than as a rule.

As for "EN diffs will decrease" nobody is forcing people not to make difficulties lower than a hard/insane for songs that don't require them, if mappers want to continue mapping low diffs for those songs they are free to do so. Most newer players don't play long songs and don't stay around Easy/Normal level for too long anyways so for mappers that like to do less work I view this more as a QOL option.

Regarding Nozhomi's concern about participation, I think having the spread rules more lenient would encourage mappers to make more diffs themselves since they don't have to map as much. At least, that's how it would affect me most likely. Getting GDs wouldn't be so high priority if you only had to map two difficulties as opposed to 4 perhaps.

That's about all that's on my mind right now... :?
-Mo-
Had a chance to play around with map data now thanks to x86's TSV. I personally think that there should be some adjustments to the proposal.

I just finished gathering the research and writing the script for my next video on song length (relevant polls here and here), and the conclusion is that most players when it comes to enjoyment prefer full length songs over shortened versions. A lot of the time full length songs do go above 3:30 in length.


Figure 1: Scatter plots of plays against song length for each difficulty. Red line is the current proposal's cut-off point for the respective difficulty.


Figure 2: Single scatter plot of plays against song length for Normal difficulties between 0 to 6 minutes and up to 6 million plays.


Figure 3: Same, but for Hard diff.


Looking at figure 1, it is true that the longer the map length, the less likely that the map becomes super popular, but keep in mind that this could be due to there being less maps in the 4 minute range compared to maps in the 1:30-2 minute range (see: these). It's unreasonable to say that lower level difficulties in the 4 minute range don't get played at all, because they clearly do. In figure 2 you can see that there are still a good handful of normal diffs with at least 250k plays in to 3:30 to 4, with a few being above 1 million plays.

The main concern for me is locking out those 3:30-4:00 songs from the newbie players, since they still do play them, and the jump from normal to hard can be difficult for some players because of how large that gap can be compared to other difficulty gaps.

You could say that we already have a large number of ranked normal difficulties in the game already, but this is a long-term change for the game, and mapping standards and quality change over time. Most of us would recommend a newbie player to play something ranked within the last few years rather than something from say 2009, for example.

Suggestions:

- Move the cut-off point for Normal difficulties up to 4:00. I think this is a more reasonable place to put it given the demand for full length songs.

- Add a new guideline that relaxes what a reasonable spread is for maps above 3:00. This is my compromise for raising the normal diff cut-off. One of the problems mappers face is having a low level normal diff and a high level hard diff, and requiring either remaps of the current diffs or the addition of an advanced diff to fill the gap. This problem is amplified for longer maps because more effort etc etc. This new guideline would allow spreads with wider gaps between each difficulty, lowering the workload on the mapper by not having to worry about spread too much and having less drain time to map whilst keeping maps accessible for lower level players.

- An (unwritten) guideline should be added/agreed for bundled maps or maps of featured artists to require normal spreads. It's probably no surprise that the most played maps in the game are the bundled default maps, and more specifically the lower level difficulties of those maps, since those are usually the first maps one plays when they first download the game. To me it wouldn't make sense to add official/bundled content into the game that isn't accessible to newbies.

I should say that I'm all for the general direction of this proposal. I just want to be weary about this sort of change since it's going to have a long term effect on the content we get in this game.
LwL

-Mo- wrote:

Suggestions:

- Move the cut-off point for Normal difficulties up to 4:00. I think this is a more reasonable place to put it given the demand for full length songs.

- Add a new guideline that relaxes what a reasonable spread is for maps above 3:00. This is my compromise for raising the normal diff cut-off. One of the problems mappers face is having a low level normal diff and a high level hard diff, and requiring either remaps of the current diffs or the addition of an advanced diff to fill the gap. This problem is amplified for longer maps because more effort etc etc. This new guideline would allow spreads with wider gaps between each difficulty, lowering the workload on the mapper by not having to worry about spread too much and having less drain time to map whilst keeping maps accessible for lower level players.
I like this solution a lot. The main problem would be how to relax something that is already somewhat subjective. My idea would be to perhaps for each star level of top difficulty define a minimum number of difficulties, which need to make a somewhat even spread down to the required minimum difficulty. Final judgement would still be done by BN/QAT as it is now, and that way we could also a bit more objectively define what constitutes a reasonable spread in the current RC. Everyone knows a hard limit on SR would be garbage because of how flawed SR is, but I think just saying "spread requirements are laxer" would be rather meaningless.
deetz
dunno if this already got resolved but I just wanna say that I support two song compilations if they were intended to be listened to together
Nevo

Monstrata wrote:

1. Take out mp3 editing as a rule. Optionally you can make it a guideline that artificially edited mp3 extensions must sound natural/unnoticeable in order to be acceptable.

2. Remove the crossfade rule, and put it in as a guideline so people have the option to call a map out for its low quality crossfading, but not the ability to completely prevent it from being ranked. It is far too subjective to be an objective rule anyways.
I completely agree with these since I dont even see what can make a crossfade low quality and mp3 extensions that are done well you cant tell are extended.
Nevo

pimpG wrote:

not really looking forward to see the beginner difficulties decreasing but the boss has spoken and he doesn't really mind that...

"only allow compilations to be mapped as the same difficulty level each song..."
^ I think this should be considered
Problem with this is not all songs will be the same difficulty example being https://osu.ppy.sh/s/744238 which has a 206bpm song and a 86bpm which are completely related as its part of the same album project so its a completely logical compilation
Stack
also agree on what Halfslashed said a few pages back as 2 songs are sometimes even advertised as bundled together and this would make them unrankable.
For example https://osu.ppy.sh/s/710305 and http://akatsuki-records.com/drcd0013_2.html, these 2 songs clearly belong together and even the album title hints to it.
Halfslashed's version already covers this and seems quite good so I would honestly just use that one
jas
hi im too lazy to read thru the entire thread and im not sure if this is mentioned but ima mention it anyways~~

i think it would be better if the times were changed from

- 3:30 to 3:00
- 4:30 to 4:00

just for consistency reasons, idk ab you guys but 3:30>4:30>5:00 doesnt make a lot of sense if theres a 1 min gap and then a 30 sec gap.
timemon
I'm thinking about the drain time rule things.
How about we allow some leniency on lower difficulties.
For example, The insane diff drain time is 4:10 it is qualified for the 4:00 rule, however due to the gameplay elements of Normal and Hard the maps drain time are 3:40 and 3:50 instead which I would like to extend the rule back to help those lower diffs.

The difference in drain time has to be reasonable and relative to the top diff and the overall length of the song. And the extension will only help Hard difficulties or lower.

The issue with this solution is that it is very vague and needs to be handled subjectively case by case by the Beatmap Nominators to work.

Edit: I might add some people are really concerned about less content on the game which is justified, but I think it might even add more content with this change. Think about all the full ver maps that never got ranked because of the mappers giving up/spread issues. And remember mappers can still opt to ignore this rules and map full spread should they wish to.
Kibbleru
the only exception i could make with the 3 song rule is red like roses 1 and 2 from rwby. but iirc that was an official track released by the artist?
Okoratu
very good list kibb



on a different note we'll be collecting feedback until the weekend (so probs around this sunday) and then change the draft accordingly to the consensus of this thread
squirrelpascals

Mir wrote:

timemon wrote:

I've only read the spread proposal. I quite like it, though 3:30 drain time is very hard to reach.

1) full ver maps have lots of break (because they don't have to care about drain time)
2) some full ver songs can be quite short (3:30 to 4:00) so they might even fail to hit the rules designed to help them.

Maybe adjust the drain time a bit? I think 3:00 is good for the lowest point.
I actually agree with this, full version songs are more likely to have breaks in them so having slight leeway for breaks in longer songs would be nice. Instead of 3:30 -> 4:30 -> 5:00, I would prefer 3:00 -> 4:00 -> 5:00. Not only is it more linear (there's a lot of room between 3:30-4:30 to require a hard but half that from 4:30-5:00 to require an insane) but it could also allow more breaks in 3:00+ songs.

This pretty much solves Lasse's concern I think?

Lasse wrote:

Breaks can be very important on <Insane diffs of this length and this would discourage using them, so I think this should somehow be accounted for, but I can't really come up with a nice solution right now.
This is what I also suggested here https://osu.ppy.sh/forum/p/6655193

UndeadCapulet wrote:

@squirrel

please read the first few pages for why we can't do 3/4/5 for time limits. it was almost exclusively seen as too lenient.
That was the very beginning of the thread though. Since we have a lot more people in on this discussion now and we're approaching spreading from different perspectives, I think we should reconsider a linear 3:00 -> 4:00 -> 5:00 draintime spread.
show more
Please sign in to reply.

New reply