Perceived sensitivity

[CaptaPraelium]

5 hours ago, vrnvorona said:

Though, i kinda can't really understand how to do this. I made some screenshots and got pictures ~2.65 size in scope while fov is 1/2. If i just devide 1/2.65 i get 37.75..... popular 0% MM value. if i divide 51/103 i get 100% MM 50 RSS. Can't get why they differ and why just having 38 OR 50 can be 1:1.

 

You can not explaing this but eh, is solution even exist?

This has been covered in other threads, I think we're pretty far off topic here sorry.

[CaptaPraelium]

Updated version of the geogebra demo. https://www.geogebra.org/m/adAFrUq3

I need to re-make this from scratch with three, linked targets with positions relative to each of the three focal points. It's a lot of work, it's gonna take a while. Meantime, this version allows us to reposition the target to be off-screen, to demonstrate the way the angles scale beyond the visible parts of the projection. It's *far* from perfect but it gives some idea.

Funny story. You know how people learn calculus in school and they're like "I'm never gonna need this after school this is stupid"? Well, not me, I was like, 'I need this for my career'. Decades later, different career, and I'm now having to relearn differentials just for this nerdery. The irony. lol.

Edited March 28, 2018 by CaptaPraelium

[vrnvorona]

8 minutes ago, CaptaPraelium said:

Updated version of the geogebra demo. https://www.geogebra.org/m/adAFrUq3

I need to re-make this from scratch with three, linked targets with positions relative to each of the three focal points. It's a lot of work, it's gonna take a while. Meantime, this version allows us to reposition the target to be off-screen, to demonstrate the way the angles scale beyond the visible parts of the projection. It's *far* from perfect but it gives some idea.

Funny story. You know how people learn calculus in school and they're like "I'm never gonna need this after school this is stupid"? Well, not me, I was like, 'I need this for my career'. Decades later, different career, and I'm now having to relearn differentials just for this nerdery. The irony. lol.

Differentials are ez, use wolfram if don't remember. ez pz

[CaptaPraelium]

Oh trust me, I'll be using a computer to solve them. I'm no sadist. But I need to MAKE them first

[vrnvorona]

Just now, CaptaPraelium said:

Oh trust me, I'll be using a computer to solve them. I'm no sadist. But I need to MAKE them first

Well that's another story, glgl

[SpecialOpsPepe]

I'm extremely new to this topic but in essences wouldn't a perceived viewpoint, (here i believe we are talking about sensitivity relative to fov?), differentiate between each persons matter of perception as not everyone perceives exactly the same? For example with ingame sensitivities there is no "perfect sensitivity", only standards through which have been deemed most suitable for optimal performance. Perhaps what I've wrote out is maybe a bit cliche or rhetorical, but i'm curious as to what exactly the goal in mind here is.

[Bryjoe]

11 hours ago, SpecialOpsPepe said:

I'm extremely new to this topic but in essences wouldn't a perceived viewpoint, (here i believe we are talking about sensitivity relative to fov?), differentiate between each persons matter of perception as not everyone perceives exactly the same? For example with ingame sensitivities there is no "perfect sensitivity", only standards through which have been deemed most suitable for optimal performance. Perhaps what I've wrote out is maybe a bit cliche or rhetorical, but i'm curious as to what exactly the goal in mind here is.

This is a difficult question really. You will have to make a compromise somewhere as unfortunately not every game has the same options. 3rd person game? I match on 360 and on Monitor distance for aimed/scoped view if there is one. If I can't change FOV, I personally think it's usually better to match based on 360 for HIPFIRE ONLY. As precision accuracy is usually not important from hipfire. It's best to just try to match FOVS between your games rather than covert based on monitor match, IMO. Scopes and aimed views have to be converted, though. 

An interesting case is actually in Battlefield with the vehicles. I am truly torn as to what is ideal, to match on 360 or match based on monitor distance. Lots of aiming in the vehicles, with turrets and machine gun nests it actually is pretty precision based but with the tank cannon, not so much. If I use the matching distance of 0%, the aim feels good but controlling the vehicle feels slow, which is more important? That is the question you have to ask. If you look at good players, I doubt they think about it nearly as long or as hard as we do, they just pick a sensitivity that feels good and run with it, which I personally think matches up with Distance/360 more often than monitor match.

[SpecialOpsPepe]

10 minutes ago, Bryjoe said:

This is a difficult question really. You will have to make a compromise somewhere as unfortunately not every game has the same options. 3rd person game? I match on 360 and on Monitor distance for aimed/scoped view if there is one. If I can't change FOV, I personally think it's usually better to match based on 360 for HIPFIRE ONLY. As precision accuracy is usually not important from hipfire. It's best to just try to match FOVS between your games rather than covert based on monitor match, IMO. Scopes and aimed views have to be converted, though. 

An interesting case is actually in Battlefield with the vehicles. I am truly torn as to what is ideal, to match on 360 or match based on monitor distance. Lots of aiming in the vehicles, with turrets and machine gun nests it actually is pretty precision based but with the tank cannon, not so much. If I use the matching distance of 0%, the aim feels good but controlling the vehicle feels slow, which is more important? That is the question you have to ask. If you look at good players, I doubt they think about it nearly as long or as hard as we do, they just pick a sensitivity that feels good and run with it, which I personally think matches up with Distance/360 more often than monitor match.

This is all quite confusing indeed lol, but so far from my experience it seems best to just choose whichever feels most natural and build off of that, there will always be a way to calculate the most optimal settings to optimize learning curves and performance but nothing beats experience. The most confusing part for me currently is understanding how FOV effects the degree of moment on the same cm/360, relatively it doesn't, you still turn the same amount of degrees with the same cm/360 with an fov of 90 as you would on an fov of 70, but it feels faster. I'd suppose thats because the difference in pixel density causing you to perceive your degree of movement as faster, which becomes much more drastic per reduction of total field of view. So far i think depending on the drasticity of FOV reduction it's may be best to keep the same cm/360 with unzoomed optics even when they change total fov while aiming. (In pubgs case ADS with red dot/iron sights/holosight is locked to 70 even though they are unzoomed optics.) I've just started delving into this topic last night so excuse me if im totally wrong with my assumptions  .

[potato psoas]

11 hours ago, SpecialOpsPepe said:

The most confusing part for me currently is understanding how FOV effects the degree of moment on the same cm/360

When you change FOV you must change the cm/360 with respect to the FOV. Here is a diagram to help you visualize why this is the case:

As you can see, a change in the FOV increases the radius and the cm/360 of the circle.

But if you were to imagine that both of these circles shared the same cm/360 then the lower FOV will feel faster because the same angular movement in-game would translate to a position further along the monitor, or even outside the monitor - totally not what you expected.

Edited March 30, 2018 by potato psoas

[SpecialOpsPepe]

13 hours ago, potato psoas said:

When you change FOV you must change the cm/360 with respect to the FOV. Here is a diagram to help you visualize why this is the case:

As you can see, a change in the FOV increases the radius and the cm/360 of the circle.

But if you were to imagine that both of these circles shared the same cm/360 then the lower FOV will feel faster because the same angular movement in-game would translate to a position further along the monitor, or even outside the monitor - totally not what you expected.

I understand, sorta  , i'm horrible at anything that has to do with mathematical calculations, but in this case if you are to match your cm/360 to accommodate for change in fov how exactly would that be done? So far to me that seems to be the controversy. 

[CaptaPraelium]

23 hours ago, Bryjoe said:

An interesting case is actually in Battlefield with the vehicles. I am truly torn as to what is ideal, to match on 360 or match based on monitor distance. Lots of aiming in the vehicles, with turrets and machine gun nests it actually is pretty precision based but with the tank cannon, not so much.

I don't know where you got this idea. Aim is critical especially in tanks. You should use the same formula that you use for infantry.
As for movement, you do that with the keyboard..... The movement speed is limited by the vehicle, and not effected by mouse sensitivity.

 

21 hours ago, SpecialOpsPepe said:

there will always be a way to calculate the most optimal settings to optimize learning curves and performance but nothing beats experience.

Experience will trump even a 'perfect' formula... But both, will trump experience alone

 

21 hours ago, SpecialOpsPepe said:

The most confusing part for me currently is understanding how FOV effects the degree of moment on the same cm/360, relatively it doesn't, you still turn the same amount of degrees with the same cm/360 with an fov of 90 as you would on an fov of 70, but it feels faster.

I did cover this earlier in the thread but it's long AF, I know TL;DR is that our brain measures distance by angle between points, so when your FOV changes, the angles change, and it messes with your perception of distance. Distance and speed are linked, so it changes your perception of speed.

 

21 hours ago, SpecialOpsPepe said:

I'd suppose thats because the difference in pixel density causing you to perceive your degree of movement as faster, which becomes much more drastic per reduction of total field of view.

If you sit far enough away, pixels will be indistinguishable, but the effect remains.

 

21 hours ago, SpecialOpsPepe said:

I've just started delving into this topic last night so excuse me if im totally wrong with my assumptions  .

This thread is about doing things as scientifically soundly as possible. If we don't propose and test hypotheses, we're doing it wrong! Don't be afraid to speak your mind and make suggestions, if you're wrong it's just a step toward being right, and if you're right, you'll be helping a LOT

 

12 minutes ago, SpecialOpsPepe said:

how exactly would that be done? So far to me that seems to be the controversy. 

I don't know if I'd call it a controversy, but it's definitely the 64,000 dollar question XD

 

Edited March 30, 2018 by CaptaPraelium

[potato psoas]

6 hours ago, SpecialOpsPepe said:

I understand, sorta  , i'm horrible at anything that has to do with mathematical calculations, but in this case if you are to match your cm/360 to accommodate for change in fov how exactly would that be done? So far to me that seems to be the controversy. 

Well you accommodate by converting your sensitivity with one of the conversion methods, which are currently Gear Ratio, Monitor Matching and Viewspeed... and CaptaPraelium is trying to figure out another one as well. While they each follow a similar pattern, as in the diagram, the results are slightly different. Which method you use is still up for debate, mainly because the conversions will never be perfect, because of distortion on the monitor. Each method has its strengths and flaws, so you have to decide which one you think will give you the advantage.

[potato psoas]

I was thinking about something I said before about how our eyes are constantly evaluating the screen as we move our mouse... well, in-game, when we are moving our crosshair to acquire a target, the distance we expect to move and the distance we actually move is constantly changing.

If you think about how this applies to monitor matching, then with 0%MM the distance we expect to move and the distance we actually move gets closer to being the same. We feel more in control as we move to acquire the target - which is what we want. We're not just over/undershooting everything, our eyes constantly adjust. Whereas with 100%MM the difference becomes greater and we feel less in control as we approach the crosshair - wish is something we don't want.

If you ask me, knowing this makes a huge difference and makes it so much easier to decide which method to pick. There's no doubt 0%MM is far less flawed than 100%MM and every other monitor match percentage.

[vrnvorona]

2 minutes ago, potato psoas said:

I was thinking about something I said before about how our eyes are constantly evaluating the screen as we move our mouse... well, in-game, when we are moving our crosshair to acquire a target, the distance we expect to move and the distance we actually move is constantly changing.

If you think about how this applies to monitor matching, then with 0%MM the distance we expect to move and the distance we actually move gets closer to being the same. We feel more in control as we move to acquire the target - which is what we want. We're not just over/undershooting everything, our eyes constantly adjust. Whereas with 100%MM the difference becomes greater and we feel less in control as we approach the crosshair - wish is something we don't want.

If you ask me, knowing this makes a huge difference and makes it so much easier to decide which method to pick. There's no doubt 0%MM is far less flawed than 100%MM and every other monitor match percentage.

It is kinda matching my view of effective aiming zone. Like, closer to crosshair easier it to hit. And the less flawed this zone the better our best aiming zone will work. And to use it effectively we need just to have good crosshair placement

[Drimzi]

11 hours ago, potato psoas said:

I was thinking about something I said before about how our eyes are constantly evaluating the screen as we move our mouse... well, in-game, when we are moving our crosshair to acquire a target, the distance we expect to move and the distance we actually move is constantly changing.

If you think about how this applies to monitor matching, then with 0%MM the distance we expect to move and the distance we actually move gets closer to being the same. We feel more in control as we move to acquire the target - which is what we want. We're not just over/undershooting everything, our eyes constantly adjust. Whereas with 100%MM the difference becomes greater and we feel less in control as we approach the crosshair - wish is something we don't want.

If you ask me, knowing this makes a huge difference and makes it so much easier to decide which method to pick. There's no doubt 0%MM is far less flawed than 100%MM and every other monitor match percentage.

 

In theory it seems that way with 0%, which does make it seem like the best solution, but in practice, 0% just does not feel like it preserves the speed. When aiming at a target with travel time involved (not flick), the perceived speed of the panning is the most important variable. In my findings, this was something closer to 1:1 (Gear Ratio) or 4:3 monitor match. I tested the feeling of speed by scripting CS:GO to zoom in and out whilst changing the sensitivity on the fly. 0% gave me terrible results whilst panning the camera around and travelling to a target. 0% also does not preserve flicks at any point on the monitor, whereas the higher percentages do. In my mind, 0% is more flawed. For my current opinion on Viewspeed, I think the two viewspeed ideas (v1, v2) are precursors to monitor match. Combine the two ideas, it results in 1:1 monitor match.

[vrnvorona]

3 minutes ago, Drimzi said:

 

In theory it seems that way with 0%, which does make it seem like the best solution, but in practice, 0% just does not feel like it preserves the speed. When aiming at a target with travel time involved (not flick), the perceived speed of the panning is the most important variable. In my findings, this was something closer to 1:1 (Gear Ratio) or 4:3 monitor match. I tested the feeling of speed by scripting CS:GO to zoom in and out whilst changing the sensitivity on the fly. 0% gave me terrible results whilst panning the camera around and travelling to a target. 0% also does not preserve flicks at any point on the monitor, whereas the higher percentages do. In my mind, 0% is more flawed. For my current opinion on Viewspeed, I think the two viewspeed ideas (v1, v2) are precursors to monitor match. Combine the two ideas, it results in 1:1 monitor match.

I don't believe that it is possible cause it is not sens flawed but fov difference itself. Like, whatever sens you set, if fov is different it will feel bad at edges.

[CaptaPraelium]

16 hours ago, potato psoas said:

I was thinking about something I said before about how our eyes are constantly evaluating the screen as we move our mouse... well, in-game, when we are moving our crosshair to acquire a target, the distance we expect to move and the distance we actually move is constantly changing.

If you think about how this applies to monitor matching, then with 0%MM the distance we expect to move and the distance we actually move gets closer to being the same. We feel more in control as we move to acquire the target - which is what we want. We're not just over/undershooting everything, our eyes constantly adjust. Whereas with 100%MM the difference becomes greater and we feel less in control as we approach the crosshair - wish is something we don't want.

If you ask me, knowing this makes a huge difference and makes it so much easier to decide which method to pick. There's no doubt 0%MM is far less flawed than 100%MM and every other monitor match percentage.

Yep. The reason 0% works like this is because it is at the extent of logical sensitivities - the lowest extent. Everything is too slow at the centre, and moves toward being too fast as it moves outward. Because of the way the distortion curves, the errors in sensitivity curve the same way. So most of the error is away from the centre.

note that the 'sweet spot' is closer to 0% than it is to the opposite extent. This is why, if you choose one extent over the other, 0% wins easily.

The trouble is that we don't continually re-assess our position. We make estimations and move, then re-assess it. Think of a 'flick shot'. You don't slowly move the crosshair over the target, you make one singular 'flick' of the wrist, and click a few buttons, all in one instant action like it's packaged up in a bow. If the system is that we constantly check the presence of the target under the crosshair, literally any sensitivity will do.

Edited April 1, 2018 by CaptaPraelium

[potato psoas]

4 hours ago, CaptaPraelium said:

Yep. The reason 0% works like this is because it is at the extent of logical sensitivities - the lowest extent. Everything is too slow at the centre, and moves toward being too fast as it moves outward. Because of the way the distortion curves, the errors in sensitivity curve the same way. So most of the error is away from the centre.

note that the 'sweet spot' is closer to 0% than it is to the opposite extent. This is why, if you choose one extent over the other, 0% wins easily.

I'm not sure what you are trying to get at here? I thought the sweet spot was at the crosshair for 0%MM?

4 hours ago, CaptaPraelium said:

The trouble is that we don't continually re-assess our position. We make estimations and move, then re-assess it. Think of a 'flick shot'. You don't slowly move the crosshair over the target, you make one singular 'flick' of the wrist, and click a few buttons, all in one instant action like it's packaged up in a bow. If the system is that we constantly check the presence of the target under the crosshair, literally any sensitivity will do.

Well you probably can't flick shot without developing muscle memory, but I don't doubt that you will experience the sensitivity becoming more in control as you approach the target. To what extent we can distinguish the change in perceived sensitivity with respect to speed of movement I don't know. Although one thing I realized is that a higher refresh rate monitor would help when you are making micro-adjustments. Try playing on a low framerate - you are going to be over/undershooting everything. You are definitely constantly evaluating the entire screen as you turn around. The refresh rate plays a big part in how good we can become before needing to develop muscle memory for flicks. But even then our muscle memory is never going to be perfect so it helps to be able to make those micro-adjustments.

[potato psoas]

9 hours ago, Drimzi said:

In theory it seems that way with 0%, which does make it seem like the best solution, but in practice, 0% just does not feel like it preserves the speed. When aiming at a target with travel time involved (not flick), the perceived speed of the panning is the most important variable. In my findings, this was something closer to 1:1 (Gear Ratio) or 4:3 monitor match. I tested the feeling of speed by scripting CS:GO to zoom in and out whilst changing the sensitivity on the fly. 0% gave me terrible results whilst panning the camera around and travelling to a target. 0% also does not preserve flicks at any point on the monitor, whereas the higher percentages do. In my mind, 0% is more flawed. For my current opinion on Viewspeed, I think the two viewspeed ideas (v1, v2) are precursors to monitor match. Combine the two ideas, it results in 1:1 monitor match.

I know what you're talking about, it does feel like it's panning. But what I said before about the flaws and strengths makes a lot of sense to me too - and I can notice that with 100%MM the cursor feels too fast at lower FOV and too slow at higher FOV. I feel like distortion still plays a part even with the Gear Ratio method.

[Drimzi]

If you want to test the perceived movement speed of the crosshair at different FOVs for different formulas, I have attached a script for Logitech Gaming Software (LGS) to use in CS:GO.

 

The script essentially makes the game zoom in and out. It presses 1 repeatedly, which will be bound ingame to change FOV, and then presses 2 or 3 to change the zoom direction. Someone can create a more accessible format that doesn't depend on LGS if they wish.

 

To get this to work, you first need to define how large the zoom spread is by creating a config file to bind 2 and 3 to the lowest and highest zoom levels. Example:

fovscale30+77.cfg

bind 2 "exec up_30"
bind 3 "exec down_107"

Then in the LGS script specify how many times 1 needs to be pressed before changing direction. This is the variable 'fovIncrements'.

 

After, you need to create 2 config files for every FOV. These config files will change the FOV, the sensitivity, and rebind 1 to execute the next FOV config file. An example looks like this:

up_90.cfg

fov_cs_debug 90
sensitivity 0.3879408422332074
bind 1 "exec up_91"

down_90.cfg

fov_cs_debug 90
sensitivity 0.3879408422332074
bind 1 "exec down_89"

 

Once all the config files are created, and you have created a game profile for CS:GO in LGS and imported the attached script, play CS:GO on a workshop map and execute the cfg that binds 2 and 3. Then enable 'Scroll Lock' on your keyboard and click your middle mouse button. The script should activate. The script will press 1 repeatedly, executing the down cfg files until 2 is pressed, and then execute all the up cfg files until 3 is pressed, and the cycle will continue. The game will zoom in and out seamlessly and you will be able to test how the sensitivity formula feels as it progresses through the FOVs. Alternatively, you can press one of the side buttons on your mouse, it will change FOV every x number of left clicks.

 

Example of the script in action:

 

The main problem with this method is that the zooming rate is going to be completely wrong since it scales linearly through the FOV values and not the actual zoom amount. As many are aware, the zoom ratio does not scale linearly when using basic multiplication / division on the FOV values. 179 vs 179.999 is a 1000x zoom, and 10 vs 20 is a 2x zoom for example. This very inconsistent zoom scaling may skew the results and make more bias towards certain match percentages.

 

drimzi-fov-changer.lua

 

Edited April 5, 2018 by Drimzi

[CaptaPraelium]

This is what I'm trying (badly, I know) to explain with the crappy picture above. You're actually both right.

If everything we were doing were a 2D image, which in fact it is because it's on screen, 0%MM would be correct. Heck, that's exactly how 0%MM works, it treats the game world only as it's 2D projection and tells us how much we zoomed it in on our flat screen. But, because of the distortion we experience by viewing it as a 3D scene (but not changing our focal point when we change FOV, ie, sitting in the same place all the time), the image becomes stretched, so the cursor feels like it is moving through the 3D space at a different speed. The nature of that 3D space is determined by the distortion, so the distortion is controlling the 'too fast/too slow' feeling we get. (and it's not necessarily just the distortion of our seating position - here, I am referring to the difference in distortion between two FOVs)

There's good reason why gear ratio/1:1/100%VFOV/4:3 feel closer and seem to move the image at a speed that feels right. 0%MM is always exactly right for anything directly under the crosshair. But for anything away from it, 0% is too slow, and the further it is from the crosshair, the more 0% is too slow. As we monitor match further and further from the crosshair, we get higher and higher sensitivity. If we 'monitor match' at 180degrees - even though that's not on the monitor - what we get is a sensitivity divider of 1 - as in, SAME sensitivity, same cm/360, for any FOV. Which is obviously too fast.

So, what we have here are two known sensitivities. At the centre of the crosshair, we have 0%MM. This is as slow as it gets. And at the very edge of the game world/our turning circle, we have =cm/360. This is as fast as it gets. And this makes sense. There's no reason we would ever want to go below or above these speeds respectively. So let's graph out a line between those. In my crappy mspaint, it's shaped like a tan curve (inverted at the Y axis because it's more intuitive to think of it as 'left X'' than 'negative X'). Somewhere along that line, is a place where it's a little too fast, but not too close to the =cm/360 extent; and a little too slow, but not too close to the 0%MM extent.. That's the sweet spot.... and those places where drimzi's muh feels and clever LGS script make it seem correct, they seem that way because they're close to that sweet spot.

The geogebra demo I did, shows hints of all of this, but it's not always immediately apparent. We can demonstrate it though. I'm still working on being able to show my working and give an exact formula that gives exact sensitivities to use.

Edited April 1, 2018 by CaptaPraelium

[sammymanny]

15 hours ago, Drimzi said:

 

In theory it seems that way with 0%, which does make it seem like the best solution, but in practice, 0% just does not feel like it preserves the speed. When aiming at a target with travel time involved (not flick), the perceived speed of the panning is the most important variable. In my findings, this was something closer to 1:1 (Gear Ratio) or 4:3 monitor match. I tested the feeling of speed by scripting CS:GO to zoom in and out whilst changing the sensitivity on the fly. 0% gave me terrible results whilst panning the camera around and travelling to a target. 0% also does not preserve flicks at any point on the monitor, whereas the higher percentages do. In my mind, 0% is more flawed. For my current opinion on Viewspeed, I think the two viewspeed ideas (v1, v2) are precursors to monitor match. Combine the two ideas, it results in 1:1 monitor match.

so, which mm was found to be most accurate? 56.25?

[Drimzi]

On 4/2/2018 at 1:41 AM, sammymanny said:

so, which mm was found to be most accurate? 56.25?

None, Viewspeed v2 felt most accurate for preserving mouse feel while the game zoomed in on the fly. If you move forward/back to counter the zoom, then 0% monitor distance match felt best.

Edited September 13, 2018 by Drimzi

[Skidushe]

Just a thought, but has mouse acceleration ever come to mind when doing this 'perfect formula'? 

[vrnvorona]

1 minute ago, Skidushe said:

Just a thought, but has mouse acceleration ever come to mind when doing this 'perfect formula'? 

It can't be solution cause it is inconsistent.