forum

[New Guideline] About Difficulty Spread

posted
Total Posts
65
show more
ColdTooth

[Nemesis] wrote:

I think there should be added one more diff - beginner, under 0.75/1.00 stars :oops:
Having too many difficulties on a long map is slightly annoying, however if used on a song below 2 minutes, this can be applicable.

The way that most maps should be is E/N/H/I, and I have been seeing some maps w/ E/N/H/I/I/X, putting to the risk of balancing, because of the amount of easy difficulties (E/N), and harder difficulties (H/I/I/X) isn't very balanced. Two easy difficulties and two harder difficulties is enough, but if you wanted to put in an Extra, then the spread isn't balanced, adding maybe an Advanced or a Beginner or both can fix that balance.
PyaKura
Don't try to make rules more complex than they already are. It's easier to allow a higher number of hard diffs than easy diffs within a set. Tbh as I see it, it's more like E/N/H are the "beginner" diffs, and "I/X(/X+)" the "expert diffs".
xxdeathx

ColdTooth wrote:

[Nemesis] wrote:

I think there should be added one more diff - beginner, under 0.75/1.00 stars :oops:
Having too many difficulties on a long map is slightly annoying, however if used on a song below 2 minutes, this can be applicable.

The way that most maps should be is E/N/H/I, and I have been seeing some maps w/ E/N/H/I/I/X, putting to the risk of balancing, because of the amount of easy difficulties (E/N), and harder difficulties (H/I/I/X) isn't very balanced. Two easy difficulties and two harder difficulties is enough, but if you wanted to put in an Extra, then the spread isn't balanced, adding maybe an Advanced or a Beginner or both can fix that balance.
Let's not. Spreads tend to have more insane difficulties to provide more variety to the players in the skill level range who play it the most. Sure, there are hundreds of thousands of people who can only play normals and easies, but they won't play them many times, while the few tens of thousands who can play insanes for fun will retry over and over trying to pass/FC/DT FC for the pp or fun. That's why the presence of more insane diffs caters to this group of people.
CXu
I haven't read the whole thread, so I'm sorry if it's been mentioned, but possibly a guideline for how far apart two diffs in a spread can be? For example, that between any two adjecent difficulties, keep the SR difference lower than say 1.5 (just pulling some number). This way you can't have a 2.26 Hard and 5.24 Insane, and should cover the case of a 3 diff mapset's spread being too "wide".
Kodora

CXu wrote:

I haven't read the whole thread, so I'm sorry if it's been mentioned, but possibly a guideline for how far apart two diffs in a spread can be? For example, that between any two adjecent difficulties, keep the SR difference lower than say 1.5 (just pulling some number). This way you can't have a 2.26 Hard and 5.24 Insane, and should cover the case of a 3 diff mapset's spread being too "wide".
We already have direct rule that difficulties should go in consecutive order; due to star rating being not 100% perfect i don't think that adding such limitations with numbers as rule would be really good idea. If jump between two diffs diifs feels to harsh it definitely will be noticed while modding - imo, would be better to leave it just as common sense thing.
CXu
I was talking about it as a guideline. I haven't looked much into how the new Star Rating rates difficulty, but from what people seem to think, it should be enough to make some rough guidelines as to how big of a difficulty gap between diffs should be. Obviously, as a guideline, it can be broken in the case where SR fails to rate maps appropriately, but it will give mappers and modders something to follow or consider. I mean, in most (or all) cases the 2.26 hard and 5.24 insane gap would be too big.
p3n
Before yesterday it never even once crossed my mind that people are treating the "consecutive diff order" as not skipping web icons for difficulties. This is beyond stupid. Who started treating it like this?

Let me show you something:




This is the range that decides which WEB ICONS are used depending on the star rating of any given map. This does, in most cases, NOT indicate a good diff spread.

See the small window to achieve a "Normal" rating? Some songs make it almost impossible to achieve the E N H spread while properly utilizing the respective gameplay elements in each difficulty bracket. You can end up with either E E H or E H H depending on the BPM of the song but the difficulties themselves are actually a perfect spread.

The quality of a diff spread can only be measured by checking gameplay elements used in each diff.
A nice even SR curve without huge jumps is also good to have. There have been guidelines for years and they have been consistent for the lower difficulties. I gathered most of the stuff that has been the "norm" but it is by no means complete. This only covers the lower spread as changes and new trends in mapping will steadily evolve the Insane and Extra tiers of beatmapping.

[Easy]
Generally following the baseline and the most distinctive features of a song. Leave some recovery times and use sliders to avoid monotonous 1/1 spam. Following the vocals or melody is a possibility. Consistent spacing is required to teach beginners the correlation between timing and distance.

Difficulty setup should include a low slider velocity, generous drain rate, low to medium hit accuracy, and a big circle size. Approach rate should be set to match the BPM and slider velocity of the map and allow enough time to plan the next move without cluttering the screen with too many hit objects.

Avoid stacked notes
Avoid overlapping notes
Avoid kick sliders



[Normal]
Generally following the most distinctive features of a song. Recovery times should be plenty but less than in the related Easy map. Use of 1/2 notes is encouraged as well as the developement of more interesting beat placements if the song allows it. Spacing should be mostly consistent to teach beginners the correlation between timing and distance.

Difficulty setup should use a low to medium slider velocity, medium drain rate, medium hit accuracy, and a big circle size that is the same or smaller than in the related Easy map. Approach rate should match the higher slider velocity and pacing of the map while leaving enough reaction and planning time without cluttering the screen.

Avoid triplets/streams (1/4 notes)
Avoid spacing inconsistencies



[Hard]
Following mostly "what you can hear". Try to capture the essence of the song and don't leave out too much. Interesting patterns and beat placements are encouraged along with jumps and short streams/triplets. If your mapset features a Normal and an Insane make sure your Hard map is a reasonable stepping stone between them.

Difficulty setup should include a BPM appropriate slider velocity that captures the pacing of the song. Use multiple slider velocities if needed. Drain rate and hit accuracy should balance the difficulty of your patterns and the frequency of streams and other 1/4 notes. Approach rate should match the BPM and slider velocity.

Avoid extremely long streams
Avoid full screen jump patterns






_________________________________________________________________________________________



Now that this is covered:

The "rule" will stay. The mis-interpretation will probably result in a rewording and the addition of a set of guidelines that will have to be worked on before they will be included.

We want to avoid spreads like N I I I. We also never intended to rely on web icons to evaluate the qualitiy of a diff spread. Depending on the BPM and nature of a song the SR completely fails to capture the actual difficulty of beginner or medium maps. Things like beat placement, readability, and usage of established gameplay elements will always be judged by the modding community and the BAT (or equivalent).
IamKwaN
I don't think any new guidelines are needed. That's the responsibility or, say, modding routine for modders or BATs to check the difficulty spread. Trust yourself and judge it confidently.
Moreover, star rating is somehow incomplete and therefore not comprehensive enough to judge the real difficulty. Personally, it's totally unnecessary to make things complicated. If it's only a guideline, why do mappers need to follow it? They can still be lazy and things won't change in my opinion.
xxdeathx
Damn p3n I was waiting for you to write up insane and extra too
Kodora

p3n wrote:

a big circle size
As huge CS hater i can't really agree with it, cuz CS4 easy and CS5 normals can be mapped pretty nicely and easy enough (CS5 was pretty good default CS size days back)

Overral really great guidelines, p3n! I believe all these can be added to wiki after discussion here - that would be really awesome guides for a new mappers! :)
Oracle

p3n wrote:

We want to avoid spreads like N I I I. We also never intended to rely on web icons to evaluate the qualitiy of a diff spread. Depending on the BPM and nature of a song the SR completely fails to capture the actual difficulty of beginner or medium maps. Things like beat placement, readability, and usage of established gameplay elements will always be judged by the modding community and the BAT (or equivalent).
This.

EDIT: I try to make my point a bit more detailed.

Providing another example of spread problem i faced : https://osu.ppy.sh/s/153776
The spread is perfect, but the star rating is inaccurate due to the high bpm. I think we're having problems with both the web icon and the star rating.

So additionally, I would like to add a bit more on the icon-spread/star-rating-spread thing. Since quite a lot of unnecessary unqualifications happened these days,

Ranking Criteria wrote:

The difficulties in the mapset must be in a consecutive order.
a consecutive order does not always mean a consecutive web-icon-spread, please see the 4-diff spread example below,

it's a perfect spread from the SR side. But it's a bad spread according to the web icon(E H I X).
That's totally weird, isn't it? Actually we will never know if the spread is good or bad until we actually check through the patterns in the diffs.

Mentioning this with p3n's point above and fycho's point below as well. Most Insane diffs are over 4.5 stars nowadays(especially on high bpm songs, 170~). As far as i know, the new star rating system values a diff depends on the physical difficulty for players to achieve it. So the SR varies with different bpm. Two Hard diffs containing exactly the same objects in 150 bpm and 220 bpm will have totally different star rating. While it does not mean the pattern usage is anywhere different. I don't think star rating is perfect atm.

Then, is it really necessary to have everything with 4 diffs or even more according to star rating? I will never know.

But yeah, if we're going to word a guideline. We must have a specific description about the case(4.5 in this case). With the star rating being inaccurate and not reliable. We can't really word anything depends on a not reliable source. Plus if it's a guideline instead of a rule, then it can't force anyone to do anything about the spread. Guideline does not lead to unrankable stuff so sadly they're quite powerless.

tl;dr:
I believe spread can be manually checked and judged by modding community and BATs from case to case. Do not be too relying on star rating. They're not perfect. Personally, I do not think we need such guideline.
Fycho

p3n wrote:

We want to avoid spreads like N I I I. We also never intended to rely on web icons to evaluate the qualitiy of a diff spread. Depending on the BPM and nature of a song the SR completely fails to capture the actual difficulty of beginner or medium maps. Things like beat placement, readability, and usage of established gameplay elements will always be judged by the modding community and the BAT (or equivalent).
I almostly agree p3n's post, and I don't think there needs any new guideline, also the "icon-spread" shouldn't be taken considerated when ranking/unranking a map.

Star rating is not completed atm which can't be used as reference imo, for example this map, https://osu.ppy.sh/s/97925, spread was totally fine at the beginning, also charles is quite sure of it, constant jumps of usage in insane diff makes an "EX" icon, and triplets/jumps in hard made spread fine too.
However, recently I was told to change some 1/2 sliders to constant 1/2 circles to make hard a "Insane icon" to avoid being unranked. This sounds even silly.

BATs/QATs should judge spread case by case, not relying star rating.
Natsu
so basically, unranks like this one: https://osu.ppy.sh/forum/p/3381966 were pointless?
Mao
Looks like that, Natsu.
I personally also think we shouldn't judge according to the Icons but as remember we were told (like this unqualify shows) not to qualify Maps that don't have a consecutive Spread (according to thethe amount of Stars in the RC).
Also about two weeks ago I've asked a QAT about it and he also said that a Gap within the Icons is unrankable.
But yeah, completely agree on what Oracle said. Spread is a case by case thing that also depends on BPM and overall playability that the SR probably doesn't get accurately.
ErunamoJAZZ
About the differences in difficulty due to BPM, I think it is clear that a 220 song itself is harder than a 150, and what you can change this is simply the SV.
At 220, the density of notes is higher. I do believe that this should be judged by a human, but do not remove the credit staring system to show this.
PyaKura
I'm lost. Are we supposed to be following a consecutive icons order or not ? And what if my EZ and NM are tagged [NM] and [HD] respectively on the website ? Does it get considered an NM/HD/[Insert 2 higher diffs here] spread basically making it unrankable ?
haha5957
As long as you have reasonable Easy and appropriate difficulty spread it should be fine I guess?

let's say a map has a difficulty spread of 1.4-2.7-4.0-5.8

Would you really require hard difficulty (around 3.4ish) on the mapset only because the difficulty gap between normal and insane is too big, when the average difficulty difference is actually 1.3~1.4SR?

Why not require another insane or extra between 4.0 and 5.8 cuz their difficulty difference is also way too big for average insane player and they cant prepare for appropriate extra?


And most of you guys honestly don't even play normal or below but it seems like that you are pretending to be "the guy who thinks about newbie".
The truth is newbies take no longer than a week to be a average normal+~hard player, and there are more than enough of those diffs.


In addition to that please realize that osu already has ridiculously big amount of very easy maps. sort it by difficulty, and you will somehow realize that osu! has much smaller group of 3.75 or higher diffs than you imagine.

Some other rhythm games (specificly BEMANI) some musics have 6 or even 7 (of 10) as their easiest difficulty and nobody complains about it. - mainly because there are many other easy musics and they admit that some higher BPM songs deserve higher diff (even to the easiest ones). However osu has the most music in all of rhythm games (I have like 33000 difficulties in my hand, most of them are easy and normals) and calling osu is lacking "difficulty spread" seems like nonsense to me. if A mapset is lacking difficulty spread, other mapset that is ALSO lacking difficulty will somehow make it fit.



Thing should be simple as :

if you have a extra difficulty map (and is obvious) you should have at least three other difficulties with appropriate difficulty gap, and you should have at least two other difficulty with appropriate difficulty gap if you have insane difficulty.

Bullshit+bullshit = legit
1.0 2.2 4.3 5.7(lacking hard) + 1.5 3.2 4.8(lacking normal) = 1.0 1.5 2.2 3.2 4.3 4.8 5.7 (lacking nothing)

and maybe force that 2.0SR rule as the easiest difficulty if you really wanna force it
PyaKura
can't make that a general hard rule for every mode
blissfulyoshi
I was thinking about the diff spread issue, and something cam up in my mind, dt versions of the map should be considered as part of the spreads. What I mean by this is that the dt version of a every diff in a mapset should be considered when rating a map's spread.

Before you think I'm crazy, first let me explain my reasoning. As pointed out in the op, a lot of hard diffs use a lot of 1/2s and a lot of X diffs use 1/4s, or in other terms a 2x increase in note density, but with no real middle ground. While this might be okay with lower difficulty maps, as you go to higher BPMs, the gap between hard and X becomes increasingly large. Luckily, dt is only a 1.5x multiplier, allowing the creation of that middle ground, which could solve that problem.

As a personal example, I am currently mapping a 130BPM song (https://www.youtube.com/watch?v=gsqNTQoO1mw#t=5631) that has drums every 1/8th tick, so lets just call it a 260BPM stream map. As you might expect, the Hard diff plays like a normal 130BPM stream diff, and extra plays like a 260BPM stream diff. The question is, how do you bridge them? The only reply I got about this was use triplets and/or short sliders t/256562. While this solution works, 260BPM triplets isn't exactly something I would consider insane, regardless of spacing. However, a 195 BPM stream 130 * 1.5 is definitely a lot closer to an insane diff, which leads to the necessity that dt diffs should be considered in the spread.
Kinomi

blissfulyoshi wrote:

I was thinking about the diff spread issue, and something cam up in my mind, dt versions of the map should be considered as part of the spreads. What I mean by this is that the dt version of a every diff in a mapset should be considered when rating a map's spread.
most humorous joking 2014.

blissfulyoshi wrote:

Before you think I'm crazy, first let me explain my reasoning. As pointed out in the op, a lot of hard diffs use a lot of 1/2s and a lot of X diffs use 1/4s, or in other terms a 2x increase in note density, but with no real middle ground. While this might be okay with lower difficulty maps, as you go to higher BPMs, the gap between hard and X becomes increasingly large. Luckily, dt is only a 1.5x multiplier, allowing the creation of that middle ground, which could solve that problem.
how about normal to hard or easy to normal? there are all 1/1 in easier diffs but 1/2 in harder diffs

blissfulyoshi wrote:

As a personal example, I am currently mapping a 130BPM song ( https://www.youtube.com/watch?v=gsqNTQoO1mw#t=5631 ) that has drums every 1/8th tick, so lets just call it a 260BPM stream map. As you might expect, the Hard diff plays like a normal 130BPM stream diff, and extra plays like a 260BPM stream diff. The question is, how do you bridge them? The only reply I got about this was use triplets and/or short sliders t/256562 . While this solution works, 260BPM triplets isn't exactly something I would consider insane, regardless of spacing. However, a 195 BPM stream 130 * 1.5 is definitely a lot closer to an insane diff, which leads to the necessity that dt diffs should be considered in the spread.
BPM is a sign of difficulty too, if you ignore the other tracks just because of difficuly. sorry, that is not mapping.
blissfulyoshi

Tsukuyomi wrote:

BPM is a sign of difficulty too, if you ignore the other tracks just because of difficuly. sorry, that is not mapping.
I think something seriously got lost in translation because I really don't understand you, but first off, difficulty should never be purely judged off BPM. BPM is an indicator of how fast the downbeats are in a music, nothing else, assuming it is timed correctly.

Now, if you mean by tracks, you mean by other things to follow, these things have to follow the BPM, which usually means 1/1, 1/2 and 1/4, nothing inbetween, so not sure how you are planning to bridge by following other things when they still all fall under 1/(2^x) where x is an integer. Usually, a lot of maps these days bridge maps by difficulty of jumps, but more difficult jumps does not prepare you for faster streams and vice-versa.

In short, I am not joking, but I don't understand your argument at all.
Topic Starter
Myxo
Uhh, this is getting in a completely different direction than I wanted it to go.

At the end, it might be best to just leave the Ranking Criteria as it is (since it doesn't seem to up-to-date anyways and there seem to be many flaws) and let the BATs judge it case-by-case.
I have the feeling nobody reads the RC anyways
p3n

Mapset

  1. The difficulties in the mapset must be in a consecutive order. Easy or Normal can be skipped if the gap in the star rating spread allows it. The order can be seen in the chart below. If your mapset has two difficulties, one of them cannot be an Insane or Expert. The lowest difficulty must be below 2.0 stars. The difficulty level of Taiko-specific and osu!mania-specific difficulties must also follow a well-designed spread and might contain an Hard/Insane only, if there are standard difficulties present. In CtB, the spread evaluation is upon the BATs discretion. The difficulty spread is determined by the map's star rating. A map falls under a certain difficulty when having a specific star rating:
    1. Below 1.5: Easy
    2. Below 2.25: Normal
    3. Below 3.75: Hard
    4. Below 5.25: Insane
    5. Above 5.25: Expert


Guidelines

  1. The difficulty spread should be linear and reasonable. Linear difficulty spread means your difficulties have a comparable gap in star rating between each other. A reasonable spread means the usage of difficulty appropriate gameplay elements. If your mapset does not have an Easy difficulty your Normal difficulty should follow the general guidelines for Easy difficulties. If your Insane difficulty is at or above a 5 star rating it is recommended to include another Insane level difficulty between Hard and the 5+ Insane.
Topic Starter
Myxo
Nice! Here is my question:

p3n wrote:

Easy or Normal can be skipped if the gap in the star rating spread allows it
Does that mean you can have Easy Hard Insane sets now (telling from the icons), when the gaps between the difficulties are even? Or did i misunderstand something.. Also, is it still allowed to name the difficulty Normal in this case instead of Easy, although the icon is an Easy one?
p3n
1.4 -> 2.6 -> 3.8 can work depending on the song and the mapping style. Overall if the BATs think it would be beneficial to the spread to include another diff at 2.0 in this imaginary set then that is what should be done. But theoretically there could be a mapset that is mapped with an E-H-I spread that is properly balanced. That will not free the BATs from actually checking the lower diffs and the usage of gameplay elements and the general setup of the spread like SV, beat patterns, spacings, overlaps, difficulty settings and so on.
Loctav
Amended for now. We might link a new article for the "difficulty appropriate gameplay elements" to clarify this.
DakeDekaane
I'm assuming "skipping Easy or Normal" relates to the web icons and not in the real spread, right?
Loctav
It always does, yes.
Kibbleru
but what if this scenario happens:

mapset has a proper spread say like 1 -> 3 -> 5

but then the mapper decides to add another insane diff between the 3 -> 5

would the mapper then need to add a difficulty too between 1 -> 3 to make the spread linear again?
Please sign in to reply.

New reply