Viewspeed v2

[potato psoas]

On 12/8/2017 at 3:51 AM, Bryjoe said:

Why would sitting farther away help with the FOV? No matter how far or close you are to the screen the 3D space that is shown is the same. Actually, when you are farther away a lower FOV should be more comfortable, hence why a lot of console games have low FOV (also better performance). I think matching based on a ratio like viewspeed works just fine. In fact, it has been a revelation in how I match sensitivities across multiple FOVS. Game to game, my aim feels the same, it is not challenging to jump right into a new shooter anymore.

Here, I made a little diagram that shows the distance you expect to move after shifting is not always the distance you actually move.

Keep in mind that as you move forward the monitor grows (why there is a bigger circle) and when you move backwards the monitor shrinks

Here is another way to look at it:

Obviously your brain adjusts to sitting position but this will affect the expected sensitivity.

Edited December 9, 2017 by potato psoas

[potato psoas]

On 12/8/2017 at 3:51 AM, Bryjoe said:

I think matching based on a ratio like viewspeed works just fine. In fact, it has been a revelation in how I match sensitivities across multiple FOVS. Game to game, my aim feels the same, it is not challenging to jump right into a new shooter anymore.

If you think Viewspeed feels better and makes more sense then you use that as your conversion method, but you have to convert from your sitting position otherwise it will always feel just a little bit off.

Edited December 9, 2017 by potato psoas

[potato psoas]

23 hours ago, Drimzi said:

It's slower.

Input: Horizontal 4:3 Based

Input: Horizontal 16:9 Based

Input: Vertical

f = fov, c = resolution height, m = mouse dpi.

Since the formula uses vertical fov, use the correct link, the fov you set will get converted to vertical, so it can output the correct 360 distance.

Do you follow the same procedure as this: ?

On 12/2/2017 at 2:18 PM, Drimzi said:

The easiest way to use the updated Viewspeed formula is to do the following:

1. Use this formula
2. Set x = resolution height / mouse dpi
3. Press enter to calculate the 360 distance
4. Press approximate form, and click more digits until you reach the max amount.
5. Copy the large number you get
6. Open mouse-sensitivity.com calculator
7. Change to Distance mode and inches
8. Set Input game to Quake Champions (it has the highest precision, sensitivity with 8 decimal places)
9. Change the FOV type to Vertical Degrees, put in 90 FOV, put in your mouse DPI
10. Copy the sensitivity result
11. Change it from Distance mode to Sensitivity mode, change inches back to centimeters if you prefer
12. Paste in the sensitivity result for Quake Champions at 90 vertical FOV
13. The Quake Champions input will act like Windows/Desktop
14. Set the conversion method to Viewspeed v2
15. Save it as a preset
16. Now set the output to the game you want to convert to

 

[Drimzi]

Edited January 28, 2018 by Drimzi

[Drimzi]

Edited January 28, 2018 by Drimzi

[WhoCares?]

Just now, Drimzi said:

Just a warning, this formula is still not final. I am not satisfied with low FOVs at all.

Just to make so (so I dont mess with new sensitivitys and thus my muscle memory all the time), both Windows to 3D AND Viewspeed need some improvement at the moment right?

I would relly like to help with bringing this forward, but unfortunately my mathmatic skills arn't good enough  

 

[Drimzi]

Edited January 28, 2018 by Drimzi

[WhoCares?]

Looks pretty good! Which version it is?  is it a modification of your last recent posted formular?

[CaptaPraelium]

Looks solid!
Would be interesting to see this same test with two circles, one small like this and one very large.
I think it's best with some kind of crosshair too, so we have a more obvious point of aim.

[Drimzi]

Edited January 28, 2018 by Drimzi
Forgot to include the desktop conversion

[KandiVan]

Latest iteration feels too fast to me. Anyone else?

[WhoCares?]

5 hours ago, KandiVan said:

Latest iteration feels too fast to me. Anyone else?

Did you use 90 or 73.74 (vfov) in the formula? Drimzi said that scaling to 90 is probably wrong!

73.74 its slower just looking at the numbers 

 

Edited December 9, 2017 by WhoCares?

[Drimzi]

Edited January 28, 2018 by Drimzi

[potato psoas]

On 12/8/2017 at 5:29 PM, Drimzi said:

I believe any formula to do with sitting position is unnecessary. You will only use it to scale relative to what you already have. Sitting at your desk would require no extra layer of scaling. You would only need to scale it relative to your normal sitting position if you want the same feel from a drastically different, temporary position, like a far away couch, or playing on a large projector, or something like that.

Okay I thought about this and you are right. You only have to adjust after shifting position. Since your eyes are perceiving 2D as a projection of 3D not 3D as I have shown in my posts.

 

In that case I would like to go back to my other conversion method that I have been talking about for ages... I've always found something was off with Viewspeed and Monitor Matching. This whole thing hasn't been thought about in a methodical way.

Let's assume that your monitor distance doesn't change between FOVs and the circumference goes around the edges of the monitor, like in this diagram:

As you can see, the lower the FOV, the greater the circumference and radius becomes. Now knowing this I think it's pointless that we came up with all these arc ratios and chord ratios and circle ratios and gear ratios when conceptually it's so simple. If you are given the monitor distance and the FOV then you can calculate the cm/360 using trig.

First you find the radius using trig

Then you find the circumference with the radius.

And there you have it, the cm/360 for that particular FOV at your given monitor distance.

 

Here is an example at 2400 DPI 3/11 1920x1080:

Edited December 9, 2017 by potato psoas

[Cocyx Skeleton]

On 12/7/2017 at 2:43 AM, Drimzi said:

It's slower.

Input: Horizontal 4:3 Based

Input: Horizontal 16:9 Based

Input: Vertical

f = fov, c = resolution height, m = mouse dpi.

Since the formula uses vertical fov, use the correct link, the fov you set will get converted to vertical, so it can output the correct 360 distance.

Out of all of the new formulas so far, this one feels the most "right" to me, personally. (I believe this is the one you made the Osu video with)

[Drimzi]

Edited January 28, 2018 by Drimzi

[Drimzi]

Edited January 28, 2018 by Drimzi

[potato psoas]

13 hours ago, Drimzi said:

 

That is viewspeed v2. You are only getting different results because you are using the horizontal fov and full desktop width, whereas I use the 1:1 values only, as I believe adding and cropping additional fov should not affect the sensitivity.

Example:
 

1/(sin((π x)/360)) = csc((π x)/360)

(π monitorDistance / DPI) * csc((π FOV)/360) = Viewspeed v2

 

 

Your result from doing aspect ratio dependent horizontals:

(π 1920 / 600) * csc((π 106.26)/360) * 2.54 = 31.91858 cm

My result doing aspect ratio independent verticals (for widescreens):

(π 1080 / 600) * csc((π 73.74)/360) * 2.54 = 23.9389 cm

 

 

My latest formula is doing the above, but also scaling by the Chord/Arc Ratio (Viewspeed v1).

(π MonitorDistance / DPI) * (1/((sin((π FOV)/360)) (360 sin((π FOV)/360))/(π FOV)))

It is definitely a lot more elegant this way, and your picture really explains what is happening. Cheers.

 

Here is another picture:

(Example: 16:10, 10.668cm/monitor width, 90 FOV in 4:3 Base)

I realise that it's exactly like the one DNAMTE made. Though I never understood what it meant... I'm so out of the loop.

I guess we at least came to the same conclusion.

Edited December 10, 2017 by potato psoas

[KandiVan]

All i need is science to show me that its the best option. So youre saying that 2D -> 3D will more than likely never be perfect? Regardless, if V2 is the best option then ill stick with it. 

[potato psoas]

4 hours ago, Drimzi said:

For 3rd person perspectives, I'm not sure on how to tackle that, or if it even needs to be compensated for.

So I looked at Minecraft and PUBG and came up with this:

The monitor is the camera and it is behind a point of rotation (usually the player model) with radius as the distance between camera and point. The camera rotates around the point as the tangent to the circle. Two things can happen: the player model turns around with the camera or the camera just moves around the player model. Sometimes the radius between the point and the camera decreases when you bump into a wall like in Minecraft. Depending on the game the radius might affect the FOV or cm/360. DPI Wizard would have to take that into account for each game.

I'm not sure how you convert as a sensitivity, but this is what is going on between the camera and player model.

Edited December 10, 2017 by potato psoas

[CaptaPraelium]

18 hours ago, Drimzi said:

 

 I believe adding and cropping additional fov should not affect the sensitivity.

 

Quite right. And by that logic, your formula should work equally well for 16:9, 21:9, 1:1 ...... and 0.5:1, 0.2:1 , etc
This is the perspective stuff I was banging on about earlier in the thread. I still believe that not accounting for this is our greatest oversight and probably the reason we keep hitting walls.

[potato psoas]

18 hours ago, Drimzi said:

whereas I use the 1:1 values only, as I believe adding and cropping additional fov should not affect the sensitivity.

 

15 minutes ago, CaptaPraelium said:

Quite right. And by that logic, your formula should work equally well for 16:9, 21:9, 1:1 ...... and 0.5:1, 0.2:1 , etc
This is the perspective stuff I was banging on about earlier in the thread. I still believe that not accounting for this is our greatest oversight and probably the reason we keep hitting walls.

So I was thinking about it and I don't think it makes any sense trying to change a game's sensitivity when the vertical FOV scales with the horizontal FOV anyway. What you need to do is have a separate sensitivity for the vertical and the horizontal. That or you play a game in 1:1 aspect ratio. I've only tested it in CS:GO so far, but I don't think in any game developers actually convert the Vertical aspect properly. They do in fact just add and crop FOV depending on your aspect ratio. When I tested CS:GO, the vertical and horizontal both had the same cm/360 but they have different FOVs. So they would feel different.

Have a look at this:

(This is just an update on my previous diagram)

Orange is how it should be if you only have one sensitivity setting. If you are going to go the Green way then you need a different sensitivity for vertical and horizontal.

 

I tried using 1600 DPI for both x and y on the desktop but used 2000 DPI for y in CS:GO and it felt exactly the same. Then you go back and you realise how flat the vertical sensitivity feels.

Edited December 10, 2017 by potato psoas

[CaptaPraelium]

Hmm I don't know about that. IMO the vertical and horizontal sens should be the same since 1 degree up and 1 degree left are the same angle of 1 degree no matter how that 1 degree is represented on screen. Consider how your suggestion would effect a player who can rotate around the Z axis (example an airplane which can roll). Now that you bring it up, "would this work in a plane?" is probably a good checklist item for any formula...

 

53 minutes ago, potato psoas said:

They do in fact just add and crop FOV depending on your aspect ratio.

Right and this is what I am trying to say about perspective. We already know that simply accounting for the change in vertical FOV is not sufficient to account for the perception effect of the distortion in the horizontal plane, so we need to look for another constant. In viewspeed v2, that constant has been the diagonal assuming 1:1. ... which to be honest made complete sense to me. Until I though it through some more. Really, it's just taking another arbitrary horizontal measurement which is the same mistake we made when assuming 4:3 and 16:9 in the past. If we think about it, the horizontal distortion is a result of the formula used for the projection and should be derived from said formula.

Let us remember that at 180 VFOV (beyond which the projection fails entirely), there is NO distortion in the horizontal plane because 180 VFOV = 180 HFOV.

Note that if you block off the sides of this video so that you see a square (I just held up my hands), the lines of perspective to the vanishing point in the centre are perfect 45 degree angles to the corners of your square.

The distortion occurs relative to the reduction in VFOV and regardless of horizontal screen space.

Dude, how are you drawing those images? A picture tells a thousand words and I'd love to be able to communicate so clearly as you can with those images.

[potato psoas]

21 minutes ago, CaptaPraelium said:

Note that if you block off the sides of this video so that you see a square (I just held up my hands), the lines of perspective to the vanishing point in the centre are perfect 45 degree angles to the corners of your square.

Yeah I can see this in CS:GO by using the command mat_setvideomode 1000 1000 1 (Use 1000 x 1000 as the bottom gets cut off by your taskbar). Placed my ruler along the vanishing line and yeah it was 45 degrees.

 

21 minutes ago, CaptaPraelium said:

Dude, how are you drawing those images? A picture tells a thousand words and I'd love to be able to communicate so clearly as you can with those images.

Just throwing them together in Paint, but I make sure the dimensions are correct and all the points connect fairly accurately - helps heaps cause usually I guess with hand drawings. Have to write down formulas and use my calculator to do it.

Edited December 10, 2017 by potato psoas

[CaptaPraelium]

That's...mspaint.............. You're a God. Sorry I lack the skill and patience for that jazz lol