forum

Adopting a "Star Perceptual Map"

posted
Total Posts
108
This is a feature request. Feature requests can be voted up by supporters.
Current Priority: +367
Topic Starter
RBRat3
I brought this up in another request and thought it deserved its own request as to not side track the other thats completely different.

This is an alternative idea to display difficulty with more information and accuracy versus using the star system or the maps difficulty title.

(Measurement groups and image are pure concept)

Star Map is infinitely expandable to accommodate more attributes such as but not limited to Hitcircle counts, Slider Counts, Jumps, Etc...
Example: 4, 8, 16 respectively and is not constrained to even leg numbers also I wouldn't exceed 8 legs for simplicity sake.

Is possible to stylize the maps shape to an extent.
deadbeat
biggest problem would be location imo, where would you put it. cause i'm pretty sure putting it over the scoreboard won't go well x.x
also would this chart change when hard rock is used?
Topic Starter
RBRat3

deadbeat wrote:

biggest problem would be location imo, where would you put it. cause i'm pretty sure putting it over the scoreboard won't go well x.x
also would this chart change when hard rock is used?
You didn't pay attention to that image a bit, It was a button toggle at the top to show that page ( next to the public/local score button )... I would have erased the scoreboard out but im not gonna put that much effort into a mock up than I already have. Technically the black BG for that panel will be transparent to match the scoreboard page, it's solid to half-ass hide the scoreboard to get the concept across.

In a nutshell the whole thing is pretty flexible so yea its skies the limit for things like HR to be shown, its more about how much effort you wanna throw at it.
Rei Hakurei
hmmm if i could switch from difficulty settings to map statistics like DDR then i would support this :)
(also circle size is ranged from 0 to 10 , but the allowed ones is only from 3 till 7)
Archangel Tirael
Support. Looks pretty functional and beautiful, but honestly, it would be nicer if Star Perceptual Map would be made more similar to the standard skin osu!, Would skinable and would be the same size as the scoreboard.
Topic Starter
RBRat3

Archangel Tirael wrote:

Support. Looks pretty functional and beautiful, but honestly, it would be nicer if Star Perceptual Map would be made more similar to the standard skin osu!, Would skinable and would be the same size as the scoreboard.
The mock up is made purely to get the point across if I had it my way I would have made it plain black and white, Ideally the whole thing would be made to match while using the scoreboards BG and width... Its just over sized to cover it mostly.
deadbeat

RBRat3 wrote:

deadbeat wrote:

biggest problem would be location imo, where would you put it. cause i'm pretty sure putting it over the scoreboard won't go well x.x
also would this chart change when hard rock is used?
You didn't pay attention to that image a bit, It was a button toggle at the top to show that page ( next to the public/local score button )... I would have erased the scoreboard out but im not gonna put that much effort into a mock up than I already have. Technically the black BG for that panel will be transparent to match the scoreboard page, it's solid to half-ass hide the scoreboard to get the concept across.
ah yup, sorry. at second glance i see the toggle button. my bad. also if hard rock will play a factor in the star map, can i suggest having the cap at 11 instead of 10?
peppy
You sure love pink blue combinations, don't you :P. I see this is a feasible addition, and for a lot of lower players, would likely be more useful than the top players display.
Topic Starter
RBRat3

peppy wrote:

You sure love pink blue combinations, don't you :P. I see this is a feasible addition, and for a lot of lower players, would likely be more useful than the top players display.
Huh well it practically kinda is osu!'s primary colors ^_^
Ohh and just to note the map is actually the song selects Red & Blue... It just happens that glossy overlay I slapped on it made it slightly pink so thats my defense :P

Sakura
I like this idea, although like my post in the other thread: t/98070 i would prefer if it had more map statistics, this is a good start tho, so Support!
Topic Starter
RBRat3

Sakura wrote:

I like this idea, although like my post in the other thread: t/98070 i would prefer if it had more map statistics, this is a good start tho, so Support!
I included the possible expandability in the top post, There is one issue with adding said things (jumps?) is that you need a "max" in order to graph into a percentage which means someone is going to have to go figure out what ranked map has the the most of said things unless this is already recorded in peppy's secret laboratory :P
bwross

RBRat3 wrote:

I brought this up in another request and thought it deserved its own request as to not side track the other thats completely different.
I'm sure I've seen star maps come up even before that... but again that's probably lost deep in some other thread, so it's best that it has it's own now.

The main problem with star maps is that they're not actually very good at anything.

There are a number of problems with using them here:

1) Star maps work best when the arms are weighted to be the same. This makes the area represent something consistent (although the human ability to judge that isn't very good, which is part of the reason why star maps aren't very good). This is pretty much impossible to do here objectively. The shape itself does carry some meaning even if the area is inconsistent... however, I'm not so sure how useful those shapes would be with these particular values.

2) OD, AR, DR, and CS are discrete parameters. They have definite set values that people want to know. If they were calculated stats on a continuous scale where people might not care so much about the exact value, then star graphs would work better... because reading values from star graphs is a relative pain. Bar graphs/histograms (ie the stars on the web page listing) work better (most notably because they're aligned to the same direction, which is easier to compare that the star graph's, each one in a different direction). However, in this case, just giving the numbers works best for these four... there's no need to make people count stars in a histogram or ticks on a star, because the specific values of OD and AR for a map are of more interest than a visual representation of the difference between OD and AR for a map (this ties back to why the shapes produced won't be interesting... the correlations between these values is less interesting than the values alone).

3) The star graph is so large it needs to be hidden away and brought up on demand (or alternatively, for those that default to it, it blocks other information that needs to be brought up on demand... for one, it hides that precious "max combo" information that some people love). A more compact format that's small enough to be always up is far better. For example, numbers in icons (eg a number inside a solid circle for CS, a number in a double circle for AR, a number under a left pointing arrow for DR, and a number in a square for OD) are nice and small, give the exact values, and could be squeezed in just about anywhere (ie they can be arranged as a 2x2 square and should be able to squeeze in beside the current displayed Length/BPM/Object/rating info with room to spare)... even on the song tabs (allowing for comparison across maps at a glance).

That isn't to say that a star graph couldn't be pretty... it's just that I wouldn't use it this way. A better way would be to use in not for the set map parameters at all, but only for calculated continuous ones. By calculating a bunch of factors from a map, you could then organize the arms so that the shape carries meaning... ie putting factors that are tied largely to object density on one side and putting the factors related more to movement on the other will create blobs that lean one way or the other depending on how streamy or jumpy a map is.
Topic Starter
RBRat3
My only reason for wanting it is the ability to be expandable without occupying any more space than the UI can offer and the ability judge at a glance without consolidating numbers to figure out what that means for my play experience. Its not intended to just show those 4 values for the rest of osu's life and yes an elegant arrangement of value sets needs to be in order to visually show known difficulty shapes. As for making it severely smaller would negate the point of having rule markings essentially losing its precision visually but yes number values can be substituted for it.

More or less this is a start point and a idiot thing if experienced players need the absolute value its already readily available or the number can be included with it. If you know of better value sets to be used along with a calculation method then by all means suggest away ^_^
Winshley

Rei_Fan49 wrote:

(also circle size is ranged from 0 to 10 , but the allowed ones is only from 3 till 7)
You can still force-select Circle Size as 0~2 and 8~10 via Notepad though. :P

I think I mentioned about the "radar graph" about this too, but I forgot where I posted it. :?
Sakura
Reposting:

I think it would be better if we start basing it off DDR's star graph:



And then change some values to osu! meanings, like Air = Jumps, Voltage = spacing changes?, Chaos = speed changes?, things like that.
SkyDevil
Really great idea :3 SUPPORT.
Topic Starter
RBRat3

Sakura wrote:

Reposting:

I think it would be better if we start basing it off DDR's star graph:



And then change some values to osu! meanings, like Air = Jumps, Voltage = spacing changes?, Chaos = speed changes?, things like that.
Still need a max to percent off of and a method of calculating the value like how far and time between does a note have to be in order to be called a "jump"?. Should probably poll that :P

Also to note that title says adopt, This really isn't meant to replace a solid number (can be shown with it) nor replace any current requests that involve showing map statistics in any way... I found a spot for it and I don't see any reason not to bench test the thing while allowing for other requests to be fulfilled, It can always be removed if it doesn't live up to any usefulness just like anything else :)
Luna
The Jump/Speed/whatever ratings could be determined similarly to Tom's difficulty calculator, that does a damn good job at estimating all kinds of stuff.
Topic Starter
RBRat3

Luna wrote:

The Jump/Speed/whatever ratings could be determined similarly to Tom's difficulty calculator, that does a damn good job at estimating all kinds of stuff.
Got a link for that?.... Also no one likes estimations XD
Luna
t/92485

It's really damn accurate for how basic it is, with a couple of improvements this would work great I'm sure
Zare
Luna is right, combine this with Tom's Calculator and this would be a relly nice addition.
Topic Starter
RBRat3
Well that took it's sweet ass time for 23k maps :P This can be added with no issues with the map display when arranged right. I didn't look if a max was being factored or not but when turned into a percentage and graphed it should work great ^_^

If it isn't then all ranked maps need to be calculated to find a max value to at least be remotely accurate by having a boundary to go by otherwise its just a number to any newb.
peppy
Any calculations can be easily worked into osu! to calculate when a map is displayed, so that is no issue. I think if this was to be implemented, it'd be best to keep the number of stats to a bare minimum. Four looks really nice, five is alright, but any more just seems crowded and reduces the usefulness of the display.
Yuzeyun

Sakura wrote:

Reposting:

I think it would be better if we start basing it off DDR's star graph:



And then change some values to osu! meanings, like Air = Jumps, Voltage = spacing changes?, Chaos = speed changes?, things like that.
I'm pretty sure Chaos and Stream would fall under the same category, Chaos in DDR is the non 4th/8th notes ratio in the map (that's why a very streamy chart gets easily 1.00 Chaos) :P

Anyway I'm for this idea, but actually it should be adjusted to every mode (In Taiko the graph would be very different, I'm pretty sure that we would need as equivalent : Stream, Chaos, Freeze (Spinners and Sliders or Dendens and Drumrolls will be under this). Voltage I still don't know wtf it means, though it's been 7 years I have been playing it, wtf.
Topic Starter
RBRat3

peppy wrote:

Any calculations can be easily worked into osu! to calculate when a map is displayed, so that is no issue. I think if this was to be implemented, it'd be best to keep the number of stats to a bare minimum. Four looks really nice, five is alright, but any more just seems crowded and reduces the usefulness of the display.
Well I came up with a min of 8 for simplicity, But the only way it could be useful is getting rid of the ruler and putting actual numbers at the rim like so :S

Somewhat similar to this

Yea I know Pink all up in this *^&$@!
peppy
I meant the different type of stats displayed (in the above case, 4) rather than the range of each stat. So yeah, this is fine.
Mithos
I wouldn't mind this, but we should add Speed/BPM to the graph as it is a huge part of map difficulty.
Topic Starter
RBRat3
You can add anything to it but the name of the game is getting an arrangement of value sets where it can form a shape you can understand. Putting things in a random order will just spit out a goofy shape that will leave you sitting there going wtf is this.

For the mock up its pretty straight forward, Bigger the diamond the harder it is.
deadbeat
personally, i don't think circle size would be as needed in the diamond as the other 3. maybe that could be replaced with bpm or something like that?
Topic Starter
RBRat3
Gimmie a max bpm estimate to work with and a map and we'll see how it translates :)
Zare
Let BPM cap at 220. Every map above 220 gets the max and maybe an additional small "extra star"
Mithos
I was going to say 250 because anything faster than that is usually in the approved.
deadbeat
i think 220 bpm is fine. i can only think of one ranked map higher than that. so 220 should do
MMzz
But if you're on doubletime won't the map need to adjust to the new BPM. Mods like hardrock should also effect it.
Topic Starter
RBRat3

MMzz wrote:

But if you're on doubletime won't the map need to adjust to the new BPM. Mods like hardrock should also effect it.
With mods on there can be a dead zone in the star where it would intersect a red ring of sorts...

Id show you what I mean but I'm busy trying to pull something out of my ass and it really isn't working so well XD
deadbeat
what about a second star map that is toggled by the use of mods?
Mithos
The problem with having a high max number for BPM is that while other elements like approach rate and overall difficulty often reach the higher parts of it's spectrum, BPM will only reach 2/3rds of the way for some of the hardest ranked songs in the game (without mods). BPM should go to 220 or 250, and any BPM that goes over that through mods or whatever get a star or something saying it goes over the max.

Next thing for bpm... What do we do about BPM changes?
Topic Starter
RBRat3

Mithost wrote:

The problem with having a high max number for BPM is that while other elements like approach rate and overall difficulty often reach the higher parts of it's spectrum, BPM will only reach 2/3rds of the way for some of the hardest ranked songs in the game (without mods). BPM should go to 220 or 250, and any BPM that goes over that through mods or whatever get stars.

Next thing for bpm... What do we do about BPM changes?
Take all the changes and average it?
Mithos

RBRat3 wrote:

Mithost wrote:

The problem with having a high max number for BPM is that while other elements like approach rate and overall difficulty often reach the higher parts of it's spectrum, BPM will only reach 2/3rds of the way for some of the hardest ranked songs in the game (without mods). BPM should go to 220 or 250, and any BPM that goes over that through mods or whatever get stars.

Next thing for bpm... What do we do about BPM changes?
Take all the changes and average it?
That could work.

Should other elements that hit their max (AR 10 for example) get an above max notification? I think BPM being the only one to have it would be kinda weird.
Topic Starter
RBRat3
Well the thing about averaging it is that it doesn't account for length...

like 300bpm changes to 80bpm but the 80bpm section only lasts 30 seconds which makes the average (190) useless and inaccurate without factoring its length of use.

I suck at math but there is a way to factor it
Cyclohexane
that wouldn't really work if you look at songs like ICARUS which get up to 251bpm for less than one second and are otherwise around 140bpm. Same thing for songs like New Castle Legions by Dirty Androids, whose bpm goes up throughout the whole song (starts at 120, finishes at 180bpm with 150 and 170 sections in the middle)
Plus imo bpm isn't always a good indicator of a map's difficulty. If you take Skrillex - Bangarang, LC's diff, it's 128bpm and yet not for the faint of heart. And that's an Insane. It makes even less sense on easier difficulties.
Mithos
I tried to make an equation to do it but honestly for now if you make a minimum length requirement (in bars) for timing sections with BPM changes, you can weed out the small BPM jumps and get an average from that.

Also, BPM isn't the only factor going into the diamond/pentagon.
MMzz
Honestly we should just stick to the 4 facotrs from difficulty settings. BPM is already displayed anyways, and BPM difficulty is pretty opinionated.
For instance 220 BPM in taiko is nothing for me.
Wishy
Agreed, some people find some BPMs harder than anothers. For example, I find playing 220~240 BPM streamy maps easier than 165 BPM@accuracy, while it is usually the opposite since lots of players can't stream that fast, but are better at lower BPMs.
Mindwaves
really good idea,support.
Zare
I still think we should add "Jumps" and "Streams" as additional factors by using Tom's calculator.
Topic Starter
RBRat3
This is what I mean by a dead zone, The red ring would represents overages when a mod is applied
bwross
BPM shouldn't be used... it's a set parameter of the map like OD, DR, CS, and AR. It has a specific value and it's already displayed in a clear way for people that want to know it. What you want is a measure of EBPM (Effective BPM)... because things like significance of BPM changes (ie how important is the range of BPM displayed) and what baseline the map is mapped to (ie 1/2 beats vs 1/4 beats) are the things that are currently hidden, but can be calculated.

As for how to measure EBPM, well, it's essentially the baseline object density of a map. Objects/second is one measure of that. Also interesting are the peak burst rates (ie the streamy bits) and the lengths of bursts (because the occasional triple isn't a stream). Movement rate is the same... a baseline average velocity (the "air speed" of the map... a measure of the size of the beat spacing), the peak burst rates (how extreme jumps are) and the lengths of those bursts (how long the jump sequences are). These are the things I would be focusing on, and recall that Tom was working on, and had a pretty good grasp of what was needed to extract meaningful numbers (which is why I've never felt the need to play with things personally further). They're also the exact things that you want on a star graph... things like set parameters should just be listed. Derived stats from the mapping are the only things that need be on a star map, and Tom seemed to have a good handle on those, so I'd go with his stuff (although I haven't seriously looked at his stuff, just read the post a while back, I remember it to be fairly solid, and people seem to like the numbers he's producing). If peak burst rates and lengths are combined that gives four stats: an EBPM measure, streaminess, average velocity, and jumpiness.

This leaves room for say some sort of measure of chaos... ie the jerkiness of the map, jerk technically being changes in acceleration, but in this case it could probably be derived somehow from the amount and size of changes in adjacent intervals between objects.

Average queue length would be another possible derived stat that could be used. However, that depends on if AR remains not a preference, and it's also going to be correlated to both AR and the measures of object density (both the average and the burstiness), in a way that might be easily enough judged (assuming that the object density is available on the star graph and the AR is listed elsewhere). So it could probably be left out to keep the map simple (it also has the problem of interpretation... both ends are hard in their own way).

That leaves five stats: two for streams, two for jumps, and one to represent rhythm/flow chaos. Which seems a good mix to me for giving a feeling about what the map itself might be like (as opposed to just the parameters it will be played under).
Mithos
I thought the polygons were made to measure the intensity (thus the difficulty) of the map. I agree with bwross, but BPM should not be ruled out as it still affects difficulty greatly.
bwross
The thing is that BPM doesn't directly affect difficulty at all. What matters is how it's used... a 320BPM map can easily be mapped like it was only 160BPM (or vice-versa). So Just knowing 320 or 160 can be deceptive. That's why the focus should be on the objects in the map, not the music. Which is why you want stats based on the density of objects... which is very much like the BPM (same units), but is actually things the player has to do.

Now, the BPM itself is still useful to know. People like to judge themselves against the BPM level of stream they can do well. But for that you want to know the exact value of BPM, not just an impression of how large it is... and that's something that's currently displayed (and should remain). However, the listing of a song at a BPM you have difficulty streaming at doesn't mean that it has any real streams that you have to go up against (or whether the streams in the map are divided into chunks of a size you can manage, or are put together into a single long death stream where you're going to eventually slide off). To know that, it comes back to the objects and how they're packed in the map. Right now that can be done to an extent by using the number of objects and the length (which can give objects/second)... but deeper analysis of the map and it's bursts could be far more accurate.
show more
Please sign in to reply.

New reply