forum

Maximum Speed a Player Can Handle

posted
Total Posts
31
Topic Starter
abraker
The following is an attempt to describe the maximum speed at which a player within the osu! community can execute a micro pattern. This description assumes a linear micro pattern which involves back and forth movements or a V shape. This description ignores tapping and human reaction. This is the first of the several upcoming descriptions in an effort to describe the difficulty of maps in the standard gameplay mode.

Let's assume the micro pattern is as long as possible. The game's distance is measured in osu!pixels, which is 1 pixel in 600x480 resolution. Therefore, the max distance of any 2 hit circles can be is
sqrt(600^2 + 480^2) = 768.37 osu!px

Since the real world is not measured in osu!px, and is based on the unknown distance the stylus or mouse goes from one end to the other, we need a basis that describes how fast a player's hand or fingers can move.

Staiain has an impressive finger speed, capable of spamming up to 32 keys/second with 4 keys with remarkable accuracy. To figure out how fast one of his fingers moves on average, we divide the rate by 4. So Staiain can tap 8 times/second. Let's keep it in milliseconds since we will be measuring the time between the notes in milliseconds in the upcoming descriptions. So Staiain can tap 0.008 times/ms, or a tap rate of 0.008 kHz.

Putting it in terms of frequency gives us the ability to calculate the speed at which a player's finger is moving the fastest. I know that mouse players don't use fingers to move the mouse and use their wrists instead, but at this point I am assuming the human muscle within fingers and wrist are capable of achieving close to the same speed. This is where the max distance in osu!px comes in as that will be the amplitude. It doesn't matter how far to the side the stylus or mouse is moving in reality since a player can change the mouse sensitivity or tablet area, but it does matter onscreen.

We start by determining how long one period of the oscillation lasts.
period = 1/frequency = 1/0.008 kHz = 125 ms

Therefore, it will take the player a minimum of 125 ms to go from point A to point B and then back to point A at max distance. To find out how long it would to go just from point A to point B, we divide by 2 and get 62.5 ms. We will need this number later in future descriptions, but for now we need the time at which the cursor is moving the fastest.

At 62.5 ms the the cursor is slowing down to 0 osu!px/ms and is about to turn back to point A. The point at which the cusor is moving the fastest is right in the middle of point A and point B. So we divide by 2, giving us 31.25 ms. Therefore, it takes 31.25 ms to accelerate to the fastest speed. To find how many osu!px that is, we need to half the distance from point A to point B. So 768.37 osu!px/2 is 384.19 osu!px. To find the average speed, we divide the distance by the time:
384.19 osu!px / 31.25 ms = 12.29 osu!px/ms

So we found out that the fastest speed a player can do a micro pattern without screwing up due to inertia and strain is 9.6 osu!px/ms. But let's take a step back and take a look at the minimum time the player can preform such micro pattern: 62.5 ms. Reacting to that is impossible due to the human reaction time being 180 ms minimum (220 ms average). Therefore, this speed is the max speed for which a player could do if movement is memorized.

Now for your feedback! Did you think I did something wrong? Are my calculations correct? Is my assumption to use Staiain's tapping speed reasonable? Is my assumption the max wrist speed is similar to finger speed correct? Can you, the players, even achieve 12.29 osu!px/ms? Try setting up a map with a V shaped micro pattern and try doing it with the relax mod.
Rilene
Oh well, this is so complex, I might have to read this twice.
Endaris
It doesn't really make sense to assume a maximum movement speed by how fast you can tap and then tell people how fast they can do it with relax-mod on.
Also some of your assumptions don't make a lot of sense, for example that the movespeedfunction v(x) is uniformly continous or even a harmonic oscillator etc. Maybe try it with 2 acceleration phases and a constant v(x)=v_max inbetween accelerations.
I don't know precisely which data from Staiain you used but if we're looking into the speed at which a top tier player can perform a triple it might very well be even faster and should also give a more precise unit than "keystrokes/s" which is kind of worthless when we're looking at durations of less than 200ms all the time. According to common rounding rules your calculations have an incredibly big margin of error.
Also there is no need to calc a "reaction based micro pattern" as lowest approachtime is longer than human reaction(300ms). And if you did it you'd have to add up reactiontime and movement time anyway as 180ms or the respective 220ms only apply to an expected event you only have to click a button for(no aim component).
Topic Starter
abraker

Endaris wrote:

It doesn't really make sense to assume a maximum movement speed by how fast you can tap and then tell people how fast they can do it with relax-mod on. I don't know precisely which data from Staiain you used but if we're looking into the speed at which a top tier player can perform a triple it might very well be even faster and should also give a more precise unit than "keystrokes/s" which is kind of worthless when we're looking at durations of less than 200ms all the time.
What do you suggest to base it on? And I'm asking players for data. I want to know what the real value, non theoretical value is myself. Regarding tapping speed, Staiain's livestreams are the only reliable source of that info I came by. If you know where other players record their info, please share. At this point, I have no real clue what the max is, therefore it's only logical to base it on existing data.

Endaris wrote:

Also some of your assumptions don't make a lot of sense, for example that the movespeedfunction v(x) is uniformly continous or even a harmonic oscillator etc. Maybe try it with 2 acceleration phases and a constant v(x)=v_max inbetween accelerations.
The point of this was to find out how fast a player can go in a V shape or back and forth. For this, I assume the player would need to slow down when approaching to the 2nd note. In reality, according to the replays I watched, top tier players tend to jump between notes, not move smoothly between them unless the notes are close. Regarding your suggestion, can you explain it a bit more?

Endaris wrote:

According to common rounding rules your calculations have an incredibly big margin of error.
Figures. It's something I tend to ignore, and it kills me. I'll have to fix that when and if I fix the discrepancies within the calculations.

Endaris wrote:

Also there is no need to calc a "reaction based micro pattern" as lowest approachtime is longer than human reaction(300ms). And if you did it you'd have to add up reactiontime and movement time anyway as 180ms or the respective 220ms only apply to an expected event you only have to click a button for(no aim component).
The reaction based case is not ready then and flawed. I will include it in my next description which would be about using minimum reaction time to derive some values.
Synpoo
spoken like a true mad scientist
ZenithPhantasm
@OP You're a fucking idiot
Topic Starter
abraker

ZenithPhantasm wrote:

@OP You're a fucking idiot
Support your claim
pandaBee
Lol wakarimasen
Bungie2117
That avatar is most fitting.
Bauxe

ZenithPhantasm wrote:

@OP You're a fucking idiot
Don't steal my line bruh.

play more
deletemyaccount
this is fookin brilliant m8, *clap clap*
Risa
Cool.

At least OP's post is still better than a bunch of shitposters shitposting everywhere
StephOsu
i don't understand anything so...
Play more
E m i

StephOsu wrote:

i don't understand anything so...
Play more
Yes
-sev
mind=gelified
Full Tablet
A test I did some years ago to see how fast I could aim.

The position of the first circle is (0,0), while the second circle is (512, 384), so the distance between them is 640 osu!pixels.

The circles are a 1/16 snap from each other, at 200bpm, so the time distance between them is 18.75ms.

http://puu.sh/k7t2y.mp4

I hit the first circle 57ms early, and hit the second circle 62ms late, so the total time the jump took was 137.75ms

So, the average speed (only considering the movement in the direction of the jump) was 4.65 osu!pixels/ms.

There are ways to make achieve a better result, for example, by reducing the tablet area from Full Area (reducing the movement needed to perform the jump, at the cost of making the jump harder to aim), or making the jump be from bottom left to the upper right instead (allowing for more arm movement, instead of mainly using fingers+wrist).

A good player would most likely achieve a better result on the same test. I don't think I would do much better now than when I did the test, since I have barely improved at the game since late 2013 (even, I am probably worse now than then).

Doing the test with a back-and-forth or a V-shape jump sequence would result in a lower speed, since it requires direction changes.
E m i
I wouldn't even be able to reach the second circle
Zhuriel

abraker wrote:

So Staiain can tap 8 times/second.
(...)
So Staiain can tap 0.008 times/ms, or a tap rate of 8mHZ.
That's 8Hz, not 8mHz. You fix it later in the calculation though and arrive at the correct period.
Topic Starter
abraker

Zhuriel wrote:

abraker wrote:

So Staiain can tap 8 times/second.
(...)
So Staiain can tap 0.008 times/ms, or a tap rate of 8mHZ.
That's 8Hz, not 8mHz. You fix it later in the calculation though and arrive at the correct period.
Yea I made a mistake on the units there. It was supposed to be in ms^(-1) units, which is kHz not mHz. I just put it in the wrong units, but values seem to be correct. The frequency is 0.008 kHz with the period still remaining at 125 ms (knew the answer without doing the actual math I guess).

Full Tablet wrote:

So, the average speed (only considering the movement in the direction of the jump) was 4.65 osu!pixels/ms.
That's pretty impressive, though you were reacting to it. My extremely flawed prediction for reaction based movement (which I removed from the description) is 4.27 osu!px/ms, so that tells me I'm kinda close.
ZenithPhantasm

Bauxe wrote:

ZenithPhantasm wrote:

@OP You're a fucking idiot
Don't steal my line bruh.

play more
:( but its so generic

Reset- wrote:

Cool.

At least OP's post is still better than a bunch of shitposters shitposting everywhere
Ftfy
show more
Please sign in to reply.

New reply