first find a way to calculate your sensitivty in cm/360°

lol how ironic.... sujoy is just about to publish an uber mouse article within the next few days :) so hang fire and im sure you will have all the answers you need

Don't count on it.

dont count on what ?

It.

ye but what "it"

I belive that McFly has some doubts about Sujoys all knowing knowledge ... meaning that we might not get answers to all of our questions.

I might have misunderstod it ofcourse.

oh right , thnxs *shrug* ... im sure he will disagree with the findings then :)

You know you've reached the epica of human development when we can discuss on facts ;D

This one is a really simple question to be honest.

The idea is to work out how many pixels in 360 degrees, then how much mouse movement in inches. By dividing the number of pixels by the number of inches you get the CPI needed.

D = distance for 360 degrees in inches
W = horizontal resolution of screen
F = field of view in game
C = CPI needed

Number of pixels in 360 degrees = W * 360 / F

Distance moved = D

C = (W * 360) / (F * D)

If your FOV is 90 then you can simplify the equation a bit to:

C = 4W / D

Or to work out the distance:

D = 4W / C

For 1024 width and 400 CPI you get:

D = 4 * 1024 / 400 = 10.24 inches = 26cm

and this is just a teaser for the whole article

This wasn't actually in the article I'm writing! It's just a bit of maths.

Yep, nothing too special there but still thoughtful that you wrote something obvious down which is not obvious for quite some people ;)

fov doesn't affect sens though...

No but it affects the screen resolution which affects how much DPI you need to not skip pixels.

so you're saying fov affects dpi...

No, I'm saying fov affects dots-per-pixel.

This is because each pixel represents a larger angle when you increase the fov.

This is because each pixel represents a larger angle when you increase the fov.

You mean a smaller angle?
Maybe I'm just not getting it. If I move my mouse a certain distance, my view changes by a certain angle, regardless of fov. And wouldn't that mouse movement consists of a fixed number of dots counted?

The angle is the same yes, but using a different fox means the number of pixels in the same angle is different!

If that's not clear just imagine a scenario.

In one case you are playing with fov90. Each dot of the mouse turns you a small amount.

Imagine you now change to fov10. The same dot turns you the same angle, but the number of pixels you travel on the screen is far more. This is because your view of the game is zoomed in by a huge amount.

If that doesn't make sense, then just trust me. The formula is right.

Imagine you now change to fov10. The same dot turns you the same angle, but the number of pixels you travel on the screen is far more. This is because your view of the game is zoomed in by a huge amount.

ok, you're right :) I didn't think of testing an extreme fov in-game.

yeah it looks wierd to me that fov and resolution change the sensitivity.
Imo windows gets the messages from the mouse and then transmits it to the game that can interpret it with the sensitivity var or acceleration (changing your mouse settings in the control pannel affects the sens ingame).

The formula should look like Distance = n / dpi
(with n numer of dots you need in game to make a 360° and dpi the dpis your mouse has)

That would be true, except we're trying to work out dots per pixel. The fov changes the size of the pixel relative to angle.

For example with fov 90 and res 1024x768.

Each 1 degree of angle = 1024/90 = 11 pixels.

But if you change to fov 10:
now each 1 degree of angle = 1024/10 = 102 pixels.

Therefore if you want your mouse dots to still correspond to 1 pixel, then your dpi has to be a lot higher.

Making correspond dots of the mouse with pixels on the screen doesn't really have sense since you move in angles not pixels.
First because at a determined fov moving by one pixel when you look at something really close is very different from one pixel when you look at something very very far, and also because games don't really care of the pixels your seeing when your shooting. When you shoot, the game takes your position and angles and traces a line to know if it touches another player.
The game doesn't care if you saw the enemy or not but it's true people would want to make correspond their dpi to a small unit, but the unit should be an angle and not pixels (the tangent of that angle is the distance it does when you move just a bit divided by the distance you're looking at, but since you were using pixels it's like you were only using one distance and not the other one).

Deciding how much DPI is useful is a tough question and tbh we're only making vague answers here. I was just answering the question that the original guy asked and I'm not going to say this is the best way to decide what DPI mouse you need.

sorry sujoy but that isnt the case, in 1024 / 90 you cant just multiply it up to 360 degrees to find out what the amount of pixels is needed. In 3D you cant link a pixel as the smallest change, because the entire screen changes with 1 mousecount. I can explain it with a drawing it you want.

you are making the mistake of thinking that in a 3d game (like in a 3d game) you have a grid with a theorized maximum amount of points as well. You dont and it's not tied to your resolution

if that isn't the case, what is?

Of course the screen changes with 3D games even when you turn less than a whole pixel. I didn't claim that wasn't the case, it was the guy who posted the original thread who wanted to know the point at which DPI count corresponded to a pixel.

However if you asked me how much a sensible amount of DPI would be, I think an amount corresponding to the pixel size is enough. Sure you could have more but it would be overkill. Infact you could have quite a lot less without a noticable problem. I remember at one point I used to play QW with extreme high sensitivity on a 400dpi mouse and it wasn't stopping me being able to aim well.

ok good to hear

I think for a any sensitivty that goes above 3 inches / 360 degrees is fine, most people are around 10 inch /360 and at 400 dpi that leaves 4000 (!) positions horizontally to aim at, that's pretty much enough for anything.

thinking the dpi that way is the same that thinking ----------> 125 usb mouse rate fits just perfect with your 125hz of the quake3.

i know that you need to have (sujoy case) AT LEAST 400dpi (90fov, 26cm/360 1024resolution).
even in that case i would add some comments on that.

for the ones that followed me ok.

for the ones that didn´t follow me at this point:
it is almost impossible to measure EXACTLY the sens in cm/360, because with the sujoy´s graph you can see that there is a little bit of error and i am almost sure that the kind of measurement equipment that we (gamers) use to measure is very unaccurate (ruler and pencil). so suppose that you calculate it aproximately... an example like this could happen:

the mouse move 1 dot, the screen moves 2 pixels = choppy

the more unaccurate the cm/360 measurement, and the more error the mouse give. the more 1x2 pixels you have

so i THINK measuring "the best that you can", with double the DPI on the mouse, would correct that.

beeing said that...
shame that the mouses still skips and have so much error, or data loss. i think that STMicroelectronics (microsoft sensors) and Avago/Agilent should be working on a better sensor, maybe with a 60x60pixel sensor instead of 30x30pixel sensor.

an interesting page for you to see:
http://www.mstarmetro.net/~rlowens/OpticalMouse/

sujoy, how did you measure de count/s lost on the mouses??
i remember that one time i measure the counts lost on a regular ball mouse, and it was horrible, even moving the axis wheels very slow.

I agree, mapping 1 pixel to 1 DPI might not always work. You probably get small rounding errors etc so that sometimes you will move 2 pixels etc.

More DPI might be a better thing, it's like when you use 4 * Antialiasing to make the screen smoother. The graphics card is working out a screen size of 4096x3072 then scaling it down to make a much smoother picture. You could make the same argument for CPI.

The real question is whether you need that much accuracy in a game that is constantly moving. It doesn't seem to bother players like Toxic.

The graphics card is working out a screen size of 4096x3072 then scaling it down to make a much smoother picture.

Modern AA definitely doesn't work like this. You're speaking of a 16-fold increase in the pixels rendered, which would slow most games to a crawl.

Initial implementations of full-scene anti-aliasing (FSAA) worked by simply rendering a scene at a higher resolution, and then downconverting to a lower-resolution output. With "multisampling", support for sampling and computing multiple samples within a pixel is added directly in the graphics pipeline hardware, providing much better performance.

(Wikipedia)

I'm not really into graphics cards, so I was just guessing what it might do. But the theory is similar? You take more information and then sample it down?

Yes, basically.

The AA you are thinking about is SuperSampling (SS for short). Only a few cards are powerfull enough to utilize it with good framerate. Basicly the high end GeForce 7 series (I'm not sure if ATI offer supersampling FSAA). The most commonly used way to do FSAA is multisampling which works in a different way, as SS is way to GPU and Memory intensive on all but the very high end cards.

why would you use antialiasing if you can simply bump up the resolution, i never understood..
most kinds of bluring are relatively taxing on cpu/gpu anyway

Sometimes you can't bump the res (say you're already at your monitors max res) or running FSAA is less taxing than increasing the res. Also in some engines bumping the res doesn't help as much with jaggies as running FSAA. Say you're a lucky bastard and running a 8800GTX, then you'll usually max out your monitor before you max out the gfx without FSAA.

might be, thanks for clarifying :))

i guess its more an issue with lcds (or maybe im thinking dumb stuff again)

anti alias is not intensive at all on high end cards. I run oblivion in 1024*768 with 6x to remove pixel borders :O

the funny thing is that here in the forum, WE (gamers) make tests (sujoy´s) and analyse things more than the gamer mouse manufacturers (razer, microsoft, logitech).
shame :(

again, sujoy how did you make those tests? could you tell us a little bit more about that?

I'm writing a HUGE article right now. It will be finished and published soon. You'll see everything and it'll all make sense.

You, if it's so huge, you should have published as paperback. With a pdf edition.

Pretty much finished now, 22 reasonably long pages.

The complete formula for calculating the highest sensitivity at which you will get "pixel perfect" accuracy in Quake 3 is as follows, where:

Definitions
cg_fov: horizontal field of view, Q3 variable
m_yaw: horizontal mouse sensitivity factor, Q3 variable (if m_pitch is higher then use this instead - see below for explanation)
R: horizontal resolution (e.g. 1024 if the resolution is 1024x768)
sensitivity: mouse sensitivity, Q3 variable

Formula
sensitivity = ( arctan(2/R) * tan(cg_fov/2) ) / m_yaw

Is there any proof of this? It seems a bit far fetched just to say "this is the formula" :-)

Could you provide some source to this or a mathematical proof?

Regards

it doesn't seem to work out anyways.

Use "degree" mode on your calculator maybe. It produces reasonable numbers.

I recall working something like this out, but I don't have the will to confirm this formula ATM.

This formula has no connection to the mouse. It seems to be a formula to work out how high you can set mouse sensitivity in game before skipping pixels. In this formula the sensitivity is a game value, and not necessarily related to how far you move the mouse to turn 360 degrees (for example).

That's why this formula doesn't use the DPI of the mouse and doesn't tell you how far you have to push the mouse to turn 360 degrees.

It's a nice formula, but nothing to do with the original thread.

Well i wanted to post this formula, because its completly accurate.
"That's why this formula doesn't use the DPI of the mouse and doesn't tell you how far you have to push the mouse to turn 360 degrees."
He asked for formula, to count highest possible distance before skipping pixels, this does excatly that.
DPI doesn't matter, the number you get is always the highest possible before skipping pixels; with 1000000000dpi or 400dpi.
Just count the distance with correct sens, and you'll get the maximum distance in cm.
I figured this would be easy way to do it.

Ok that's true, but it assumes you have the mouse already. If you are considering buying a mouse but don't have it to test, then you need a formula that can work out the distance just based on the specs of the mouse.

True, but then again mouse using 800dpi moves twice as much as 400dpi mouse with same sensitivity, and 1200 three times as much as 400 and so on.
So its easy to do the math.

To be honest, I don't think it works....
Mouse: MX518
Windowsens: 10(in the middle)
Game: ET and Quake 3
Fov: 96
Resolution: 1024-769
DPI: 400


1024 * (360/96) / 400 = 9.6
9.6 is way to high for me as a sens!

EDIT:
My bad, 9.6 is in INCHES, didn't see that.
So in CM it would be 24.384CM.

24 CM?! To get pixelperfect sens? Man! That means that I should become a highsens player instead of a lowsens to get pixelperfect!

640 * (360/103) / 400 = 5.59

5.59 inches = 14.2 cm

Going to see what it feels like to have pixel perfect sensitivity!

*shock* no difference!!!!!!!!!!!

The formula gives the MAXIMUM value for the "sensitivity" cvar in Q3. It doesn't give you your IDEAL sensitivity.

Most people already use a sensitivity value lower than the results this formula will give them, so it's not really an issue.

If you DO get results that are lower than your current sensitivity, you can double the DPI on your mouse and halve your sensitivity value to get below the threshold.

How was this equation derived? I'm getting reasonable results, but am puzzled by where it came from.

It was posted @ RazerBlueprints, IIRC.

I did some math. Not sure if that equation is right.

Take your FOV and divide by resolution (Horizontal) to get the amount of degrees each pixel represents. Unless Q3 is different from the HL engine, the pixel to degrees conversion between mouse input and movement on screen is:
Degrees = Pixel * Windows Sensitivity * m_yaw * sensitivity.

For degrees, plug in the value each pixel on the monitor represents, and 1 for pixel. Plug everything else in and solve for sensitivity. This ended up returning results different than what was posted.

i am anxiously waiting for your article sujoy : )
i suppose that with the formula you can get the sens in your mouse perfectly if you can calculate EXACT sensitivity with lettu´s formula:

inches = 1024*(360/fov) / mouseDPI

with lettu´s formula you can get a pixel perfect sens, and with this common formula you can get how much the mouse has to travel to make a 360 turn.

cool !

:D

please lettu be kind and try it.

BTW, if it works, your formula is cool :D

great

cool

acceleration is fuck up them pixels fo sho nigga!

oops, accidetly double posted... delete pls^^

sensi: 5cm per 3600° (1,97 inches)
resi: 1280*1024
fov: 100

(1280*360)/(1,97*100)
=2339,0862944162436548223350253807 dpi

my copperhead is 2 bad :<

i did´t catch that fel1x...
maybe your are measuring your sens/360° in a wrong way. try with lettu´s formula.

^_^

Dont give me any credit of that formula, its just copyed from razer blueprints forum found by Hopkins.

5 cm --> 360 degrees ingame... :D

my q3 sensi is 6 and the calculations says me, 7.99 is possible ;>

but dno, i just my driver 2 get a high sensi, this calculation ignores the driver-settings completly^^

I agree man

nevermind

http://www.legras.com/mouse/

is it accurate?