Best USA coefficient for Battlefield 1?

[DPI Wizard]

7 hours ago, Skwuruhl said:

Nobody has addressed this potential issue with using chord as "resolution" or whatever else

If I understand you correctly, the green chord is a 4:3 monitor and the black chord is a 16:9 monitor, so they must have a different chord length to have the same radius.

I'll look more at all the discussion here later, working on a video at the moment

[Skwuruhl]

4 hours ago, DPI Wizard said:

If I understand you correctly, the green chord is a 4:3 monitor and the black chord is a 16:9 monitor, so they must have a different chord length to have the same radius.

I just realized I made one error in that image, here's the fixed version:

http://i.imgur.com/2AWlClN.png (still "fixed" but irrelevant)

Better explanation of the image too: The image represents a 1920x1440 4:3 monitor and a horizontal FOV of 90. The black chord is horizontal obviously, green is (assumed to be) vertical. Since they are the same "zoom level" (that is to say they are two different ways to represent the exact same FOV) they should share the same radius. Based on that, if you find the chord of 73.74° using that radius you'd expect that the calculated chord would be equal to the vertical resolution. Instead you calculate a resolution much higher than that. Since the resolutions do not match I don't believe it's accurate to associate the monitor with the chord.

Looking back over the recent posts it appears if you extend the radius of the black cone (or shorten that of the green one) so that the chords lie on top of eachother then the ratios between them match.

6 hours ago, DNAMTE said:

 

Heres something to consider. Take this example:

 

Putting alternate theory aside for a second and lets look at zoom factor. Essentially we are going to multiply the 45' 'viewspace' by 4.64, thereby filling up the 125' viewspace'. Now we have two arcs with equal chord length, however the expanded 45' viewspace does not share the same geometry as the 125' viewspace. The 125' arc is 1.1966 times longer.

Point being even when we have a FOV over two and a half times larger, the actual arc length differential can be minimal.

Ultimately even before we figure out exactly how far the apple drops from the tree, ALL FOV are scaled in some correlated way to fill the same size screen.

 

EDIT. WIP... thought I'd add this diagram here if anyone would find it useful. The numbers between dimensions are ratios.

 

 

I also realized this is literally just 0% match distance but 100 times more complicated.

tan(90°/2)/tan(45°/2)=2.41

tan(125°/2)/tan(45°/2)=4.64

Just like this image way earlier in the thread

180° does not work using this, but it also doesn't work you how you applied it in your image. The chords have to be lying on top of eachother. If you tried to do this then the chord would tend towards infinity as implied by lim x->180[tan(x/2)]

Edited June 27, 2017 by Skwuruhl

[DPI Wizard]

Ok, I see your point now. However the calculator is using formulas so the vertical chord much closer to 1440 (1519), but still off.

I'll look into the formulas

[fenriquez]

I have been testing out drimzi's formula and I agree with him that it does feel better than the current viewspeed implementation. Not sure if it is a placebo but transitioning between zoom levels in-game feel that much more fluid.

[TheNoobPolice]

I don't really see how what you guys are trying to achieve can exist. I say this because as we all know, If any distance on the monitor IS synchronised, then all other positions are, unavoidably, NOT synchronised as one FOV transitions to another. The only way around this would be to have some kind of magic mouse which knew where you were aiming to for a particular enemy and adjusted turn-rate for that position of the screen in real-time.  In other words, an "impossible with current technology" variable speed turn-rate, which wouldn't work anyway because it would fly in the face of everything "consistent" gamers like about mouse movement. 

Having a formula which in-effect works to move the match position to a different part of the screen (depending on the starting and ending FOV) could be mathematically sound in theory, but will still always have points on the screen that are not synchronised, and if you happen to be aiming at one of those points in ADS, then the aim will feel always off compared to aiming at the same point in hipfire.

As far as I can see, it's always going to be a case of "pick your poison".

[DNAMTE]

5 hours ago, Skwuruhl said:

 

I also realized this is literally just 0% match distance but 100 times more complicated.

tan(90°/2)/tan(45°/2)=2.41

tan(125°/2)/tan(45°/2)=4.64

Just like this image way earlier in the thread

180° does not work using this, but it also doesn't work you how you applied it in your image. The chords have to be lying on top of eachother. If you tried to do this then the chord would tend towards infinity as implied by lim x->180[tan(x/2)]

I think given that the outer circle remains fixed in dimension as the smaller circle expands/ contracts inside it, the diameter could be used rather then a trigonometric for 180 calculations?

What this dimension is remains up for debate.

 

TheNoobPolice, I don't think anyone is trying to match the entire screen. 

[DPI Wizard]

I think I have a solution to this. Haven't read through everything here yet, it might be something similar to what is already suggested here.

[Skwuruhl]

13 hours ago, TheNoobPolice said:

I don't really see how what you guys are trying to achieve can exist. I say this because as we all know, If any distance on the monitor IS synchronised, then all other positions are, unavoidably, NOT synchronised as one FOV transitions to another.

This is literally what I've been saying this entire time. The only way to achieve this would be to have a monitor that was curved specifically for the FOV you use and the distance you sit from it. (good luck with this, as it needs to be curved in all directions, not just horizontally like current curved monitors). The only place I see this remotely possible is with VR.

 

11 hours ago, DNAMTE said:

I think given that the outer circle remains fixed in dimension as the smaller circle expands/ contracts inside it, the diameter could be used rather then a trigonometric for 180 calculations?

What this dimension is remains up for debate.

As you expanded the outer FOV towards 180° it would approach infinity. By the time you're at 179° the zoom level to 103° is 91.15x zoom. By 179.9° it's 911.5x zoom.

[TheNoobPolice]

Basically, there's nothing inherently wrong with the existing Monitor Matching to a specific percentage or Viewspeed (in practice, currently equivalent to 70-75% distance depending on FOV with the existing formula). Here's the best options for any user.

For games in which "tracking" movement / targets with a longer time to kill is more important - use a distance close to 0%

For "twitchy" games, in which the meta is about flicking or "snapping" to targets quickly, use the distance on the monitor where you find you are identifying new targets most often.

For games with a bit of both such as BF1, anywhere from 25%-75% monitor distance (i.e BF1 coefficient of 44% -133%) will work best depending on the player, and their chosen playstyle / ADS FOV level.

For less aim-critical games, (example, Dishonored 2) where you simply want a "sensation" of equal speed across the screen / FOV. It is perfect to use Viewspeed, since it is much more straight-forward to use with the calculator, especially for new users that don't want to have to think about where they match a distance on the screen.

Nothing wrong with tweaking the Viewspeed formula to scale differently across FOV changes, but it's still going to be down to user preference and playstyle, and one calculation is never going to work perfectly for everyone.

Edited June 28, 2017 by TheNoobPolice

[Drimzi]

Edited January 28, 2018 by Drimzi

[Skwuruhl]

4 hours ago, Drimzi said:

Assuming desktop can be the diameter (180° chord length), half the desktop distance (resolution width / dpi) is the radius (r), 2 r is the 180° chord length (a), π r is the 180° arc length (s), 2 π r is the 180° circumference (C), the central angle will be the game's FOV (θ).

You can create a circle doing either of the two:

C (1/tan(1/2 θ°)) = Circumference with zoom ratio

C = Circumference without zoom ratio

 

Scaling the circumference without the zoom ratio can be done like so:

r = 3.2, s = π r, θ = 45, 1/2 s×360/θ

This is 100% monitor match. However with this formula, the desktop to game conversion at actual 90 FOV is identical to 0% monitor match, compared to the calculator's version that just ends up with 360/θ × C.

If the zoom ratio doesn't need to be accounted for at all, then 100% monitor match is probably the way to go, since it will be like pulleys or gears. If the zoom ratio does need to be accounted for, like 45° not being half of 90°, and shouldn't have double the circumference, then a multiplier is needed to solve this. I know 53.13° has a zoom ratio of 2 with 90° but I'm unsure of how to shift the match % over to 53.13° instead of 45°. Anyone know how?

 

Literally nobody uses 180° FOV anyway so it hardly matters. Even so, if you were to make a circle with a diameter of the resolution then the "inner circle" or lower fov circle or whatever you call it would have to be infinitely small to make the chords on top of eachother.

100% monitor match works for curved screens only . And again, it has to be properly curved for your FOV and sitting distance. So unless you're some super rich shit who can have custom monitors built or something... Otherwise all you're doing is matching movements at the very edge of your screen. Not very useful.

The zoom ratio is tan(largerFOV / 2)/tan(smallerFOV / 2). Not sure if this is what you're asking though.

[DNAMTE]

As you say, the result for very high fov should be extreme. Depending where your sitting, your 'natural' view if the monitor were a window would never even approach 90'. Excluding some super wide super monitor.

180' is simply distorted too much when projected on to a flat plane.

[DPI Wizard]

I'm still working on videos etc, but I'll pop inn to say that ANY calculation on this site is tuned to horizontal movement only. Any vertical movement fill just have to follow whatever is set horizontally.

And I know there's a lot of smart guys here, so maybe we can come up with something better than Monitor Distance and Viewspeed. 

Monitor Distance is  good as it can be for what it is, but for viewspeed we might be able to do some tuning.

I''ll start working on this soon

[Drimzi]

Edited January 28, 2018 by Drimzi

[Skwuruhl]

1 hour ago, Drimzi said:

Well I figured out how to make it so that half the fov gets double the circumference. All I had to do was compare the zoom ratio between the fov and 106.26 instead of 90. The 100% formula probably works correctly with a 4:3 monitor, but not 16:9. Now 53.13 has double the circumference of 90. Before 45 had double the circumference of 90, regardless of the ratio being '2.414' instead of the expected '90/45=2'.

Now this analogy applies to the correct FOV based on the zoom ratio.

r = radius = half desktop distance (inches)

s = arc length of 180

θ = actual hfov

 

r = 3.2, s = π r, θ = 90, (2 π r (tan(1/2 106.26°)/tan(1/2 θ°))) =

 

With this formula, while flicking between hipfire and scope, the sensitivity feels unaffected and it is just the size of everything growing and shrinking. If it is just the game world growing and shrinking, then muscle memory remains perfect regardless of fov. You just have bigger or smaller targets. Pressing ESC to open the menu, the 2D cursor follows the same trajectory at any fov so it is matched to 2D perfectly too.

This is basically still 0% match distance. Between 106.26 and 90 your equation is 20.11 and 26.81 inches. When considering rounding wolfram calculates virtually 0%. Pretty much only thing 'different' is that you're missing cot(θ/2) your hipfire 360° distance:

d=1920/300, θ=103, d π cot(θ/2)=

is all that's needed to match desktop (at 0%, which is what you're doing).

I already said 0% match is also the same as matching distance to the scaling factor of the zoom. Regardless it still has the same (unavoidable) issue: the farther you get from the match location the more off you get. Just a biproduct of using 2D screens, no method can perfectly accurate at more than one 'circle' on the screen.

Now on the bright side, as long as you're doing your big aiming movements in hipfire and then only entering ADS once you've aquired a target to do finer aim adjustments then 0% will be great. It's very accurate in the general area of the crosshair, and under the crosshair is perfect (hence 0%).

75% is more useful when you're sitting in ADS and want to be able to flick to targets towards the edges of your screen or generally just doing larger turns while ADS. But the accuracy near crosshair will be lower because of this. Example of a 70% circle from earlier in the thread. The farther your target from the green circle the more off your sensitivity will be.

What you choose is entirely dependent on your playstyle. Since I spend most of my time in hipfire and only ADS after putting my target generally under my reticule, I use 0%. If you camp in ADS or something and need to accurately flick to shit way off to the sides then a higher % is better.

[Drimzi]

Edited January 28, 2018 by Drimzi

[potato psoas]

Can I point out, given that "viewspeed" is like gears synchronizing with a pulley, that 2D desktop sensitivity is the same as a gear representing 0 FOV, since the closer the FOV gets to 0, the flatter the arc of the circle. I know that 0 FOV is an infinitely large circle, but maybe it is possible to use limits to calculate the exact sensitivity.

Ignore me if you think I don't know what I'm talking about lol (most of this thread is going over my head).

Edited June 29, 2017 by potato psoas

[TheNoobPolice]

That could work in matching to desktop but not with an FOV of 0 course! You could in theory start with 0.001 FOV (or less) and work it out back from that, but there is going to be a 0.1% margin of error compared to desktop, not really an issue because there is a small percentage of sensor drift on mice anyway, (although the pixart 3366/3360 sensor does a much better job than any preceeding it)

It still doesn't really matter, since there is no objective improvement that can be made for every single game and FOV transtion, because it is literally impossible for two different FOV's to have the exact same sensitivity. I'm not even convinced there is an optimal monitor distance for every FOV transtiion. It depends what you are doing in the game, where you are looking at the screen, and how you're own brain percieves sensitivity which has not at all been studied with any degree of satisfacotry data. Anecdotally, I've known some players who loved the Medal of Honor Warfight model, which if I remember rightly seemed to match degrees turned of 1:1 regardless of FOV / zoom, which I thought was bonkers and couldn't even play that game cause of it, but whatever.

If the guys want to work out a formula that they think works better then all good, but there will still be some types of games where the optimum is simply just 0% monitor distance.

[DNAMTE]

I don't mind tinkering around. If we come p with something that benefits someone, like a clearer choice for % match (like viewspeed) then that's great.

Back to chasing my tail.... I still feel the circle ratio (1.3 for 90 - 45FOV) coupled with a multiplier is where the correlation should be made. What the multiplier is I've not 100% decided on yet. I have results and ideas but not a solid theory.

[TheNoobPolice]

5 hours ago, Skwuruhl said:

Now on the bright side, as long as you're doing your big aiming movements in hipfire and then only entering ADS once you've aquired a target to do finer aim adjustments then 0% will be great. It's very accurate in the general area of the crosshair, and under the crosshair is perfect (hence 0%).

Just a point about this - sounds good in theory but people don't turn THEN aim, they turn AND aim simultaneously, and rightly so, otherwise you're taking too long to acquire a target and at higher levels get owned.

One of the key overlooked advantages of a proceedural turn-rate adjustment like BF's USA, rather than a turn rate mutilpier (either static or configurable by a slider), is that it adjusts the turn rate during the FOV transition also. Meaning the turn-rate gradually slows as you are zooming in. When you run USA off, the game "switches" to the ADS turn-rate as soon as click the ADS button, which means there is a split second of turning as the FOV is transitioning where the sensitivity feels far too slow, even if you use the calcualtor on this site to otherwise match the ADS sensitivity to be the same.

[Drimzi]

Edited January 28, 2018 by Drimzi

[TheNoobPolice]

1 hour ago, Drimzi said:

Ok I'm pretty sure I just found the solution. The point on the monitor where it matches has to be dynamic. It can't be static. I made a formula that starts at 0 for 90, goes to -1 at 0 and 1 at 180. Use that as a multiplier for 0% formula. Literally every single FOV feels the exact same.

Formula

d = desktop distance

f = 4:3 based hfov value

Interesting - How did you test this? Can you give some examples of soldier sensitivity modifiers to try this out in BF1 using this formula?

Example: If I currently play on a hipfire 4:3 FOV of 106.26 (which I do, i.e 90 VFOV) and I use the 1.25x ADS scaled zoom (i.e 93.67 4:3 FOV) using USA with a coefficient of 0%, what 1.25x zoom slider percentage would I need to use to arrive at the correct sensitivity as per your formula between hipfire and ADS?

Edited June 29, 2017 by TheNoobPolice

[fenriquez]

3 hours ago, Drimzi said:

Ok I'm pretty sure I just found the solution. The point on the monitor where it matches has to be dynamic. It can't be static. I made a formula that starts at 0 for 90, goes to -1 at 0 and 1 at 180. Use that as a multiplier for 0% formula. Literally every single FOV feels the exact same.

Formula

d = desktop distance

f = 4:3 based hfov value

With the brief testing I have done, it does feel that scaling between objects seems equal when hip-firing and zooming in. Thank you drimzi, it feels much better than the current viewspeed implementation.

[TheNoobPolice]

6 minutes ago, fenriquez said:

With the brief testing I have done, it does feel that scaling between objects seems equal when hip-firing and zooming in. Thank you drimzi, it feels much better than the current viewspeed implementation.

What settings in game did you use to test this?

[Drimzi]

Edited January 28, 2018 by Drimzi