http://www.esreality.com/index.php?a=post&...pid2087765

#edit: I actually visualized that using your values and the formula in that link:

#edit2: I updated the picture to show the same dimensions in each plot.

so we learn:
Power = 1 -> No acceleration at all.
Power = 2 -> Linear Mouse Acceleration
Power = 3 -> Parabolic mouse acceleration

#edit3: As I learned, this post does not address the question raised by the thread starter. My bad :o But for those, who don't scroll through everything: Here's a tool to generate your own plots like these:

bit.ly/14ssx49

Thanks for the concise answer. I know what injx said and it puzzles me because he is way smarter than me when it comes to this stuff but in-game it just feels different. Try it out for yourself if you don't believe me. I'll probably post a video up later demonstrating this. I know this isn't a very scientific way to test stuff but why does 9/10 times i do a 180 with cl_mouseaccel 0 using accelpower 1 and 9/10 times i do a 270 with cl_mouseaccel 3? i'll also try and get rid of the placebo factor using 100 different configs so i don't know which one has cl_mouseaccel 0 and cl_mouseaccel 3.

Barring a strange bug it's simply not possible.

I've played around with power 2.1-3.0. It's definitely interesting, but it's hard to imagine the benefits.

I now installed that into an interactive tool, if you want to play around with it:

bit.ly/14ssx49

if you want to repost it, please contact me beforehand.

#edit: I'm not quite sure about the labels on the y- and x-axis. If somebody knows the correct labeling and unit, please tell me and I can change it.

nice work! But I'm not so sure your implementation is correct.
I'm pretty sleep deprived atm, so forgive me if I'm wrong here...
Regardless of input speed, the resulting output speed will be a flat line while under the offset threshold.
The base sens should be the origin of the y-axis, since no matter what ( positive ) accel value is used, the output speed will never drop below the base sens value. Wouldn't the acceloffset value just offset the origin of the x-axis?

Like I said, I think the labels aren't correct. After thinking about it myself, I came to the conclusion that the Y-Axis shows the factor, the raw input mouse speed is multiplied with.

So let's turn off acceleration completely (power = 1). Then factor is just a constant (the sensitivity+1 for some reason). In the graphic this is represented by a horizontal line. So if I turned off acceleration and I move my mouse with 1cm/s it results in an effective movement on screen of 2cm/s when my sensitivity is 1. So what the constant line tells us, is that regardless of how great the speed is I input, it is always multiplied by the same constant.

If I turn on acceleration however, the number my input speed is multiplied by changes with regard to how great the speed is I input. (This would be represented by anything else than a horizontal line.)

So if power = 2, I get a sloped line. This means, that the multiplication factor of my raw input speed rises in a linear fashion with the magnitude of the raw input speed. This is what we would expect for mouse acceleration. The faster I move the mouse, the higher the factor my mouse movement is multiplied with.

I'll now change it in the tool.

p.s. What we learn from this, is also that:

* The Multiplication factor is never smaller than the base sensitivity.
* The offset is a lower cap for the acceleration to kick in. If I set the offset to 2 for example, my multiplication factor will be just the base sensitivity until my input speed exceeds the value 2. (For 3, acceleration only kicks in after input speed is greater than 3 etc.)

Cool tool!

http://www.youtube.com/watch?v=C5v3SCaML8o

Edit: acceloffset is expressed rather confusingly in this graph though.

Ideally say with power 3, accel ~1 offset 5 you'd see a straight line until 5 counts/msec and then it would curve upward.

nah, that's because I didn't implement the lower bound for (v-c) in the formula. v being mouse input speed and c being the offset. it is supposed to be (v-c) > 0. So everything left of the lowest point is supposed to be just flat.

You did not do a test with power = 1 and accel = 0, which is what the OP had in his post.

He states, in a response to your post, "I know this isn't a very scientific way to test stuff but why does 9/10 times i do a 180 with cl_mouseaccel 0 using accelpower 1 and 9/10 times i do a 270 with cl_mouseaccel 3?".

"Obviously" it should be the same as power = 1 and accel = 0, but the test was not done and so he doesn't have a proper direct comparison.

I still gave you a + for putting in the work that you have. It's just an amusing little aside that it didn't actually address his question :)

Edit: To clarify, bugs can happen in the strangest ways. There is a non-zero possibility that the interaction between a zero accel and a non-zero power could cause strange issues.

A while back I ran a bunch of tests and came to the conclusion that when mouseaccel is not zero the mouse handling enters a different sub routine entirely. (dy = M (B) dx)

Like when cl_mouseaccel is 0, accelpower, offset, accel, and senscap just plain aren't considered, as if they don't exist.

So there's really no room for "anomalies" there.

When accel is non-zero, it allows the user to do some silly things:

1. Set senscap equal to or lower than sensitivity (effectively making accel 0, something you find in strenx configs)

2. Use a negative offset when accelpower is 2, which just raises the base sens by the value velocity+offset*accel. For example sens 3, offset -2, accel 0.2 will start at 3.4. Velocity 1 it will be 3.6. Looks like this: 3 + 0.2(x - -2) You'd get the same exact effect from using offset 0 and setting sens to 3.4.

3. Set accelpower to 1 with a non-zero accel, which raises the base sens by 1 and removing any mouse accel. Would be like if you use sens 3 accel 0, and then set sens to 4 accel 0.

yes, you're right, I kind of misread there. But you can use the tool to generate as many plots as you want with different values. The first post was just a 'sketch'. But too bad I didn't read the question correctly ;)

I guess if you have mouseacceloffset your sens will rise only if you swing your mouse over a certain speed

i set cl_mouseacceloffset to 0 and same thing still happens. :/. All i'm asking for is for you guys to do your own tests with your mousepads and just post here. Yes, i know it's not very scientific but if a lot of people to do it with the same results then somethin must be up.

I just tried it out what you said and it does indeed feel like there is some accel with accelpower 1. But it's more like a 30 degrees difference between cl_mouseaccel 0 and 3

Setting accelpower to 1 results in raising the base sensitivity by 1

Remember it's like this:

B + (A(v-c))^(P-1)

Sens + (Accel(Velocity-offset))^(accelpower -1)

So with accelpower 1 you get:

Sens + (Accel(Velocity-offset))^(1 -1)
Sens + (Accel(Velocity-offset))^(0)

Anything to the 0 power is 1.

So you have Sens + 1

This could get significant depending on the dpi.

Increasing the sensitivity by 1 with 800 dpi increases the sens twice as much as compared to 400 dpi.

So if you were using a really high value, like 3200 dpi, and say you had sensitivity 0.5 (~25 cm/360) and your sens increased by 1 due to accelpower 1 and some non-zero mouseaccel value, your sens would be 8 cm/360. That's a huge difference!!

However you'd have no acceleration at all, just higher sens.

TL;DR

Accelpower 1 raises sensitivity by 1 if mouseaccel is not 0 but makes it unaccelerated.

does that mean with enough force it raises your sens as a step rather than gradually?

No, it means that your sens will increase by 1 and stay there at any velocity.

nice touch, bringing in the dpi here. that of course is also something one ought to look at!