Hello there.
I am making an interesting system that will potentially/hopefully work in parallel with pp (or become a modification of it).
The idea is to give a map some values that will tell "skill requirements" to succesfully play the map. And when you beat the map you gain certain amount of skill points depending on the number of points you already had.
So far I've finished basic skill requirement calculations for streams, stamina and general aim skill.
My next milestone is making calculations for the most subjective one - reading skill.

I made this thread here hoping that you can tell me your views on what's "hard to read".
Please, be as precise as possible. This will help me a lot.

1) Object density. This is the number of hitobjects on screen at once. IMO it's the most important factor, and most of what makes EZ hard.
It's probably more accurate if you don't count 1/4 notes in this calculation, since fast streams can drive up this number without actually affecting the reading difficulty. Also scaling slightly by note distance would be good too, since more eye movement is required to take everything in and react in time.

2) Overlapping notes. If there are two notes on roughly the same spot that are visible at the same time, and you have to move to another spot and back in between hitting them, that's really confusing to read. This isn't such a big deal unless the object density is high enough though.

2.5) Suddenly changing patterns. If you're doing star motions and the star suddenly turns into a pentagon or even reverses direction, it's really easy to trip up if you're not paying close attention. Even so, it's hardly an issue unless #2 is present, in which case it becomes a total nightmare!!

3) Suddenly changing time intervals. If the spacing remains constant (e.g. 100 osupixels between each note) but the pattern suddenly changes from 1/2 to 1/4, 1/4 to 1/2, etc. then that's really tricky to read. (Anything changing from normal beatsnaps to 1/3 is especially hard.) Mappers usually put new combos to fight this.


Just my thoughts for now. Those are the things I find most difficult when playing EZ, so I assume it extrapolates even to higher AR as well.

Add slider clusters and slider speed changes. Sometimes there's no way to tell and you have to either react fast or memorise them.

Ask Tom94, I think he'd know more. Some stuff I'd like to point out as well are:

- Stream spacing (Shuffle Heaven and Okeanos are infamous for this, IMO)
- Reaction reading vs. Focus reading (AR-based, if you get my drift)
- The weird boundary between singletaps and streams (Kyouki Chinden and Worldwide Choppers both have this)

You can break reading difficulty down into a visual component and a non-visual component.

The non-visual component can be summarized as a measure of variation. Between every pair of objects (and every slider start and slider end) there is a minimum speed you must move your cursor to hit both objects. This speed can vary between adjacent pairs of notes, and as the difference increases it becomes more difficult to read. The same thing is true for the angles between notes; if the angle between a pair of objects is substantially different than the angle between the next pair of objects the whole pattern becomes more complex to read.

It's easy to understand if you think of the entire map as a sort of differentiable function; the function itself is just position, but you can differentiate it to get velocity, then again for acceleration, etc.. These derivatives can be aggregated in order to find the map's non-visual complexity.

Visual complexity is a measure of visual obfuscation caused by overlapping stuff; overlapping graphics, obviously, but also things like overlapping cursor paths, overlapping patterns, etc. This is what makes EZ seem difficult; overlaps become more likely as the number of objects on-screen increases. However, the real difficulty isn't in the visual obfuscation itself, but rather its effect on how much time you have to process each note, so visual complexity can come not only from the "noise" of low AR, but also the sheer speed of high AR.

Thanks for the posts!
So far I've managed to detect object overlaps at any point of time and ignoring streams in this calculation
And also object density value.
The idea of considering angles came to my mind before but I never realized it can affect reading and not just difficulty of certain jumps (squares e.t.c) because of "difficult" cursor moves you must do
I'll definetly try this

They say 90 degree angles are the hardest to play, so the amounts of right angles or the proximity of being close to 90 degrees may also play a role in map difficulty. Check out DJPop's maps for that matter.

circles that are arranged in certain ways

skystar with extra pizza

ar >= 9.3

Pattern complexity and AR (note density) are probably the biggest ones, since basic patterns like squares and triangles etc that follow a clockwise rotation are much more easier than squares and triangles that have their order change clockwise and anticlockwise. Note density which is dependent on the AR and bpm translate to readability as the illusion of "more beats" or "less beats" affect your gameplay by making you either go too fast and missing beats or hitting them too fast

buny wrote:

Pattern complexity and AR (note density) are probably the biggest ones, since basic patterns like squares and triangles etc that follow a clockwise rotation are much more easier than squares and triangles that have their order change clockwise and anticlockwise.

Are you sure that just doesn't depend on the person

Kheldragar wrote:

buny wrote:

Pattern complexity and AR (note density) are probably the biggest ones, since basic patterns like squares and triangles etc that follow a clockwise rotation are much more easier than squares and triangles that have their order change clockwise and anticlockwise.

Are you sure that just doesn't depend on the person

readability is a very subjective topic, since it doesn't have an actual metric such as speed, aim (and arguably accuracy)

Readability absolutely does have an actual metric, we just haven't figured out what it is. People tend to get caught up in the whole "not everyone finds the same patterns easy or difficult to read" argument without realizing that readability and reading skill are two different things, just like accuracy as a statistic and accuracy as a skill. Differences in pattern recognition stem from differences in the relative development of all of reading's sub-skills, not from any inherent differences in reading ability, and while it might be absurdly complicated, there absolutely does exist a metric by which readability can be measured.

Yeah, I think the same.
I am not good at making hypotheses, so that's why I am just planning to gather angle+distance data from various difficult map sections and then try to find something common in them.
If only I had a list of them..

Narrill wrote:

Readability absolutely does have an actual metric, we just haven't figured out what it is. People tend to get caught up in the whole "not everyone finds the same patterns easy or difficult to read" argument without realizing that readability and reading skill are two different things, just like accuracy as a statistic and accuracy as a skill. Differences in pattern recognition stem from differences in the relative development of all of reading's sub-skills, not from any inherent differences in reading ability, and while it might be absurdly complicated, there absolutely does exist a metric by which readability can be measured.

when you compare metrics of speed and aim together and then you compare them to readability or reading skill, then readability or reading skill isn't something you simply "measure" and is something you observe and give arbitrary values

like the pp system, where it's insanely complex/impossible to create a system that rates the performances on maps due to different difficulty aspects that are subjective to the player in question, e.g. they may be better playing on the left side than the right side, corners than the centers etc.

basically, I don't think readability can truly be measure, other than artificial values that probably won't even model it correctly (pp anyone? How long did it take for people to figure out what gives the most pp?)

We need Tom94 for this...

And yes, readability is often a very subjective thing IMO. But I'd say variations in note placement beat-wise is what makes a map hard to read as well. Like mixing up pattern's of 1/4s and 1/8s with 1/2s for melodic accentuation, not like techno-style maps with mostly repetitive 1/2s and 1/4 notes.

dagambler999 wrote:

We need Tom94 for this...

And yes, readability is often a very subjective thing IMO. But I'd say variations in note placement beat-wise is what makes a map hard to read as well. Like mixing up pattern's of 1/4s and 1/8s with 1/2s for melodic accentuation, not like techno-style maps with mostly repetitive 1/2s and 1/4 notes.

Yeah, but stack/density make it even harder. Reading rhythm is all about just listening to the beat, so it's rather intuitive. However, if you've ever played one of those shitmaps where they stack all the notes on top of each other and there are like a million approach circles closing in on a small space at different timings, that's impossible yo.

pandaBee wrote:

Yeah, but stack/density make it even harder. Reading rhythm is all about just listening to the beat, so it's rather intuitive. However, if you've ever played one of those shitmaps where they stack all the notes on top of each other and there are like a million approach circles closing in on a small space at different timings, that's impossible yo.

Yep, AR8 Renard is an impossible read. How did those maps even get ranked? (I'm talking Banned Forever w/ lesjuh and co.)

Sentenced to a lifetime in gaol forever*

buny wrote:

like the pp system

I don't have an intricate understanding of the pp system, but I doubt it's as complex as you seem to think. And you're missing my point, players playing better in the corners or on one side of the screen is irrelevant to the inherent readability of the map in the same way that some players being better at streams or jumps is irrelevant to Tom's difficulty calculations. The final result might be an aggregation of a large number of variables, but we can absolutely measure things like intramap variation and object overlap, and we could even forgo the concept of general "readability" altogether in favor of separate metrics for all the various components like rhythmic complexity, geometric complexity, visual clutter, etc.

The point isn't to figure out how difficult it will be for the player to read the map, it's to better understand the composition of the map.

So to sum it up:
- Note clustering
- General Pattern Complexity
- Slider patterns & speeds (Hollow Wings, Broccoly, 0108)
- Note Pattern Complexity
- Stream spacing
- Reaction reading vs Focus reading

Did I miss anything in this topic?

dagambler999 wrote:

So to sum it up:
- Note clustering
- General Pattern Complexity
- Slider patterns & speeds (Hollow Wings, Broccoly, 0108)
- Note Pattern Complexity
- Stream spacing
- Reaction reading vs Focus reading

Did I miss anything in this topic?

If you think about it, most of this can be described by some kind of metric of "chaos", as Narrill mentioned on the previous page
Except the visual part

I like where this is going. Maybe we'll actually come up with some exact definitions

dagambler999 wrote:

So to sum it up:
- Note clustering
- General Pattern Complexity
- Slider patterns & speeds (Hollow Wings, Broccoly, 0108)
- Note Pattern Complexity
- Stream spacing
- Reaction reading vs Focus reading

Did I miss anything in this topic?

Spaced streams aren't hard to read, they are however very hard to do...

pandaBee wrote:

Spaced streams aren't hard to read, they are however very hard to do...

I meant varying stream spacing/accel streams (Shuffle Heaven is a prime example).

stack leniency

dagambler999 wrote:

So to sum it up:
- Note clustering
- General Pattern Complexity
- Slider patterns & speeds (Hollow Wings, Broccoly, 0108)
- Note Pattern Complexity
- Stream spacing
- Reaction reading vs Focus reading

Did I miss anything in this topic?

-Too much DT?

I can relate!!

Narrill wrote:

Visual complexity is a measure of visual obfuscation caused by overlapping stuff; overlapping graphics, obviously, but also things like overlapping cursor paths, overlapping patterns, etc. This is what makes EZ seem difficult; overlaps become more likely as the number of objects on-screen increases. However, the real difficulty isn't in the visual obfuscation itself, but rather its effect on how much time you have to process each note, so visual complexity can come not only from the "noise" of low AR, but also the sheer speed of high AR.

Any ideas on how to measure the processing time for each note taking overlaps and density into consideration?
I can only think of using objects' fade-in times as a base and altering these somehow depending on the overlap time/percentage

As long noone can explicitly determine the metric and definition of a map's readibility I will remain convinced it is a totally subjective matter.

I agree with the points which were given earlier on and would like to add distance snap aka time/distance relation. More inconsistent DS can constantly mess with your reading. Sudden DS changes can catch a player off guard. Jumps and Anti-Jumps are not merely covered with good aim and acc/timing.

Kert wrote:

Any ideas on how to measure the processing time for each note taking overlaps and density into consideration?
I can only think of using objects' fade-in times as a base and altering these somehow depending on the overlap time/percentage

Yeah, that sounds like a decent approach. I would work on developing an algorithm that compares all the visual objects at a particular time position and spits out a composite "noise" rating. Once you have that you can just run that algorithm at each object's time position and use the resulting noise rating to come up with a per-object weighted processing time. Obviously the algorithm itself is the real challenge, and you'll likely have to heavily tweak the factor by which the base time is altered to get the weighted time, but that's how I'd start.

Evaluating the non-visual component is the tougher problem I think.

Edit: Just to spitball some ideas for the algorithm...

The most obvious component is the proximity of all the visual objects. Next to that is probably overlapping cursor paths (e.g., the path between the current note and the next note overlaps the path between two notes later in the pattern). Notes fade in over time, so you'll need some way of determining how much each object has faded in at any given time position, and once you can determine that you could probably use that fade-in percentage as a multiplier for the note and its outgoing cursor path. This would make notes closer to the current time position more heavily weighted, which I think is accurate.

However, you won't just need to evaluate the objects that are actually on-screen, you'll need to look at hot spots as well, meaning areas of the screen that tend to have more notes than other areas (the easiest way to understand why this matters is to think of taking all of a map's patterns and putting them in one corner of the screen - the object density in that corner would skyrocket, making everything happening in that corner tougher to read). I can't think of an efficient way to implement this, but I think an ideal implementation would work very similarly to the noise algorithm, but would be localized to a specific point; that is to say, you pick a point and a time position and the algorithm looks through all the objects that occurred before that time and, for each object, adds some value corresponding to the distance, both in a time and screen position, between that object and the given point/time.

Now that I think about it, those two algorithms could actually be combined. Instead of running one to check for hot spots and one to check noise, just give visual objects a higher weighting than non-visual objects, and have that additional weight correspond to how visual the object is.

If I had any familiarity with osu's file formatting I'd put together a prototype for you, but alas...

What do you guys define as "reading"? In my mind, reading an osu! Chart is seeing the circles and processing what motions you have to do to hit them all correctly. Of course you need fundamentals and knowledge to adapt and read new songs but the concept is the same for all songs.

The problem with this thread is that most top players have their reading skill capped. The key elements to reading seems to be circle density and AR, however both of these things are SUPPOSED to be tailored to the song so it flows correctly and can be ranked. The only thing really left are the patterns themselves. If we are talking about ranked songs, when you play thousands and thousands of songs, you can get to a point where any pattern is readable even if it's too fast/hard for you. There are maybe like 1% of songs that will have a gimmick pattern that's unlike any normal pattern and may trip you up but 2-3 replays later you should still be able to understand and adapt.

So when we are talking about making a value for readability, why does it matter when most people can already read and have trouble playing? I agree that rotating squares are haf and maybe specific patterns should have a say but reading in general is the most basic skill. Also as far as ar and density goes, it'd be better for old maps to get unranked then for an arbitrary value be put on something that's supposed to be correct when the map gets ranked in the first place.

-Rinku- wrote:

The problem with this thread is that most top players have their reading skill capped.

Nope, not even close.

Because the map pool is overwhelmingly populated by simple maps most people don't even consider reading a trainable skill, but the reality is that your ability to decipher the visuals of the beatmap into usable knowledge in a timely fashion is paramount to your ability to actually hit the map's notes, and every extra millisecond you take figuring out what you're looking at is a millisecond you don't get to spend moving your cursor. Good reading skill makes literally everything easier, and I can tell you for a fact that all it takes for a map to go from a full scoreboard of 99%+ HDHR plays to just a handful of HDHR plays is a bit of complexity. Reading is important.

As for why it matters, it's a skill like any other, and people should be rewarded for proficiency. There are tons of skystar/fanzhen/hollow wings/etc. maps out there whose star rating grossly underrates the actual skill required to play the map, and this discrepancy comes entirely from improper handling of complexity and readability. People who manage to FC those maps deserve pp that corresponds to the map's true difficulty.

-Rinku- wrote:

Also as far as ar and density goes, it'd be better for old maps to get unranked then for an arbitrary value be put on something that's supposed to be correct when the map gets ranked in the first place.

I don't know what purpose this sentence is supposed to serve. When did anyone mention old maps? Why are AR and density relevant to the shoddy quality of old maps? What particular shoddy quality are you even referring to?

Narrill wrote:

However, you won't just need to evaluate the objects that are actually on-screen, you'll need to look at hot spots as well, meaning areas of the screen that tend to have more notes than other areas (the easiest way to understand why this matters is to think of taking all of a map's patterns and putting them in one corner of the screen - the object density in that corner would skyrocket, making everything happening in that corner tougher to read).

Are you sure about hot spots? I think object density will skyrocket because new overlaps will appear (overlaps with objects that appear ontop of previous objects / that haven't faded out yet e.t.c). If they are no new overlaps I don't think the same patterns will become any harder (atleast for reading)
Considering proximity is a must, but I am not sure how to measure that.
I am thinking about taking all object positions at a certain time and applying some bonus for those which have a lot of other objects close to them. Basically like a heatmap. Then again I'd have to figure out which are streams, because this won't really apply to them

Your heatmap is basically what I was describing. The idea is that the weighting of each object would be tied to its temporal proximity to the current time, so only objects within the same region and around the same time would have much effect.

The amount of circles on the screen.
The order of which you need to hit the circles.
The layout/pattern of the circles.
The time of which you need to hit the circles.
The AR (Approach Rate) of the map.

Reading doesn't really have a definite definition in osu!, some people think of the term differently than another.However, most say that it is, as -Rinku- said, is seeing the circles and processing what motions you have to do to hit them all correctly.

Stacks and super high AR are a pain in the ass.

What is harder to read: full overlap or let's say.. half-overlap?

Kert wrote:

What is harder to read: full overlap or let's say.. half-overlap?

What does this mean? Example?

Kert wrote:

What is harder to read: full overlap or let's say.. half-overlap?

Depends how long the pattern goes for,
patterns like tornado squares become hard because they require constant "adjusting" to the changing angles

Full


Half