Viewspeed v2

[CaptaPraelium]

sick mspaint skillz

If anyone already knows (or is smart enough to figure out quickly) the way to find the 'Perspective VFOV' I would be very grateful!

 

Edited July 27, 2017 by CaptaPraelium
More food for thought

[potato psoas]

9 hours ago, CaptaPraelium said:

I think you'd do really well to completely separate the issues of matching 2D to 3D, and 3D to 3D. In all of your excellent work on this subject, I've never seen 2D to 3D conversion introduce anything but trouble.

But it is possible to do 2D to 3D conversion. As I explained in my post:

10 hours ago, potato psoas said:

Now, we can say that 2D is essentially 0 FOV. As you approach 0 FOV the "gears" get bigger and bigger and the arc length becomes flatter and flatter (until it is essentially flat/2D). If 0 FOV were to have a Circumference (which is how we determine 3D sensitivity), it would be infinitely long and would turn infinitely slow. But we can't convert with infinity, so this idea fails. However, infinitely slow doesn't mean the gear is not moving. 0 FOV is still in sync with every other FOV. So we know that it is possible to convert 2D-3D.

 

8 hours ago, CaptaPraelium said:

I'm not so sure. Given that the vertical chord and horizontal chord ratio is always the same as our aspect ratio, but the horizontal and vertical arc ratios are not......

We know that 0% MM is = zoom ratio is = chord ratio, maybe this is why 0% feels odd. Has anyone tried using arc ratios?

Well if the game world is not spherical then what we need is a sensitivity for the horizontal plane and a sensitivity for the vertical plane. And you figure out the vertical sensitivity the same way you find the horizontal sensitivity.

[CaptaPraelium]

35 minutes ago, potato psoas said:

But it is possible to do 2D to 3D conversion. As I explained in my post:

I did not question that it is possible, I pointed out that it is a separate issue and should be treated as such. I understood your post and I agree that treating it as 0FOV seems a logical way to handle it, and if a working 3D to 3D conversion is found, applying that to 0FOV may (should?) work. Obviously that would depend on the nature of the formula - we might get a division by zero or something and the formula could then be adjusted accordingly (example, BF's USA - details quoted above) for use in 2D conversion. Heck, it may just be a great way to test a formula at it's extremes (as has been said, the correct formula should work all the way from 0 to 180 degrees).... However, observation shows that it has not served well to conflate the two problems.

 

Quote

Well if the game world is not spherical then what we need is a sensitivity for the horizontal plane and a sensitivity for the vertical plane. And you figure out the vertical sensitivity the same way you find the horizontal sensitivity.

I am not saying that the game world is not 'spherical' (it's not, it's an infinite space in 3 dimensions but I know what you mean by spherical, it's a good word so let's use it) but yes, because the sensitivity is a matter of degrees per count, the horizontal and vertical sensitivity should be the same....Which would not work if using the chord length. For example let's say we have a 'gear' ratio of 2, one gear is twice the circumference of the other. If we have calculated these gear sizes from HFOV and horizontal chord length, then we use the same formula but instead use VFOV and vertical chord, the ratio will no longer be 2.

Edited July 27, 2017 by CaptaPraelium

[potato psoas]

1 hour ago, CaptaPraelium said:

I did not question that it is possible, I pointed out that it is a separate issue and should be treated as such. I understood your post and I agree that treating it as 0FOV seems a logical way to handle it, and if a working 3D to 3D conversion is found, applying that to 0FOV may (should?) work. Obviously that would depend on the nature of the formula - we might get a division by zero or something and the formula could then be adjusted accordingly (example, BF's USA - details quoted above) for use in 2D conversion. Heck, it may just be a great way to test a formula at it's extremes (as has been said, the correct formula should work all the way from 0 to 180 degrees).... However, observation shows that it has not served well to conflate the two problems.

Well I also explained how it can be done:

12 hours ago, potato psoas said:

The solution to this is that the arc length for 0 FOV and the chord length that each Field of View shares is exactly the same distance. Therefore, we don't need to calculate the circumference of 0 FOV like we would when converting 3D to 3D, we can just use its chord length/arc length (which is what the calculator uses anyway *- the 2D edge-to-edge distance). If we are converting from 3D to 2D we use the circumference to calculate the chord length/2D sensitivity and if we are converting from 2D to 3D we use the chord length/2D sensitivity to calculate circumference/3D sensitivity.

 

1 hour ago, CaptaPraelium said:

I am not saying that the game world is not 'spherical' (it's not, it's an infinite space in 3 dimensions but I know what you mean by spherical, it's a good word so let's use it) but yes, because the sensitivity is a matter of degrees per count, the horizontal and vertical sensitivity should be the same....Which would not work if using the chord length. For example let's say we have a 'gear' ratio of 2, one gear is twice the circumference of the other. If we have calculated these gear sizes from HFOV and horizontal chord length, then we use the same formula but instead use VFOV and vertical chord, the ratio will no longer be 2.

Well it's either spherical or like a "football". The shape it creates is how the view rotates around the player in game. If it's spherical then you don't need to worry about the vertical arc because everything is 1:1. It's just like having black bars - you don't see the entire 180 degrees of FOV. But if the game isn't 1:1 and the vertical is squished or something, like when you play with a 16:10 aspect ratio on a 16:9 resolution, then you need to adjust the vertical sensitivity to compensate.

[CaptaPraelium]

I'm going to focus on 3D to 3D conversion, for the time being. These other issues are other issues, I'd rather not be distracted.

Edited July 27, 2017 by CaptaPraelium

[potato psoas]

9 hours ago, CaptaPraelium said:

The horizontal distortion gives this kind of 'wrap around' effect, like I'm looking at it all projected on a cylinder wrapped around me, or the inside of a tube cut in half. But the vertical distortion doesn't feel that way. It just feels kinda sorta like perspective. Like this:


Am I the only one that sees it that way? I mean I know it's not quite right, but when you look at that wolfram image, do you see a half-tube kind of shape? (and I mean kind of. It sure af doesn't look like a sphere to me)

What you are noticing with the cylindrical kind of look is that the vertical fov has a flatter arc. The horizontal arc is much more pronounced because the FOV is greater. And the HFOV is greater because the aspect ratio requires it to be. If the aspect ratio was 1:1 you would notice it looking a lot more even, like a sphere because the FOV and arc would be the same.

Edited July 27, 2017 by potato psoas

[CaptaPraelium]

10 minutes ago, potato psoas said:

What you are noticing with the cylindrical kind of look is that the vertical fov has a flatter arc. The horizontal arc is much more pronounced because the FOV is greater. And the HFOV is greater because the aspect ratio requires it to be. If the aspect ratio was 1:1 you would notice it looking a lot more even, like a sphere because the FOV and arc would be the same.

The lines of longitude will always be straight in rectilinear projection, and the lines of longitude always arced. This is what gives the 'cylindrical' look.
Edit: Good example, Follow the link to the wolfram site in my post. It has a greater vertical FOV than horizontal. It still looks like a 'cylinder'

Edited July 27, 2017 by CaptaPraelium

[potato psoas]

10 minutes ago, CaptaPraelium said:

The lines of longitude will always be straight in rectilinear projection, and the lines of longitude always arced. This is what gives the 'cylindrical' look.
Edit: Good example, Follow the link to the wolfram site in my post. It has a greater vertical FOV than horizontal. It still looks like a 'cylinder'

Ok nevermind then. I don't know that much about the ways you project onto the screen.

[CaptaPraelium]

I came across this video today while searching for research into the perception of the projection. Sadly he reached the point of perception (aka muh feels) and said it was too complex to discuss hahaha (at 8:09). Regardless, as for the projection itself, this is a really excellent visualisation of it all.

 

[potato psoas]

27 minutes ago, CaptaPraelium said:

I came across this video today while searching for research into the perception of the projection. Sadly he reached the point of perception (aka muh feels) and said it was too complex to discuss hahaha (at 8:09). Regardless, as for the projection itself, this is a really excellent visualisation of it all.

 

Yeah that explains a bit

[CaptaPraelium]

It sure might explain some of muh feels, and least of all, the reason why we all use HFOV even thought practically none of the game engines do.

I've also encountered another explanation for some muh feels. Incoming more 1337 mspaint skillz (These are gonna get more and more ghetto until I know of a better tool for the job lol)

.... So, apparently the idea of rectilinear projection is that, if your FOV matches your angle to the projection, when looking at the centre of the image,you don't perceive the distortion:



Accordingly, depending on an individual's FOV and distance from their monitor, they will have their own unique perception of the distortion. This explains why different people have different experiences of the same formula.

So, this is all very nice because I'm managing to pin down the reasons for the biggest culprits of 'muh feels'... However, while I can do the math to make formula that account for all of this, it's going to take a long time. Any math wizards who'd like to step in to help would be my BFF

Edited July 27, 2017 by CaptaPraelium

[Drimzi]

Edited January 28, 2018 by Drimzi

[Kilroy]

Why do we need to take into account the distance between the eye and the monitor?

[Drimzi]

Edited January 28, 2018 by Drimzi

[CaptaPraelium]

I'm of the opinion that while it's optimal, our brain can compensate for it, and it's not needed. Pretty easy to test by simply sitting further from your monitor. But y'know. Muh feels. I don't always buy into them.

To answer the question though, because as distance from the monitor increases, the angles to the screen are reduced in proportion with one another, while the angles from the projection to the screen, which are influenced by distortion, stay the same. Thus, the distortion at the eye changes as the eye's distance from the screen changes.

I do however believe that the top post on this page deserves attention. A formula to calculate the VFOV at Actual HFOV would be very interesting. The perception of perspective which results from rectilinear's straight vertical lines and curved horizontal lines increasing with verticality, is real, both on paper and in everyone's feels (and on paper of how feels works), and afaik, has not been accounted for in any of our formula as yet.

Edited July 28, 2017 by CaptaPraelium

[Drimzi]

Edited January 28, 2018 by Drimzi

[CaptaPraelium]

I found that 100% was inconsistent, it felt too fast when making smaller movements near the centre of the screen (including tracking) but for long flick shots it was awesome (hit one that was invisible off the edge of the monitor. Shadowplay didn't record. Gutted). Kinda the opposite of 0%

[Drimzi]

Edited January 28, 2018 by Drimzi

[KandiVan]

Ok so I use (90 π d)/θ to convert from 2D->3D (hipfire) and then 100% monitor match that hipfire sensitivity to my ADS sensitivity?

[Drimzi]

Edited January 28, 2018 by Drimzi

[KandiVan]

2 minutes ago, Drimzi said:

Yeah for simplicity just convert to 90 degrees so you don't have to use wolfram for everything. Save the input and pretend that it is Windows/Desktop, like I have done here 

https://gyazo.com/305bc8dba668770220cc2d9168c2076a

 

https://gyazo.com/dbc865b2bde64f3fc0548ce2f361218e

 

Anything wrong with what I've done there?

[Drimzi]

Edited January 28, 2018 by Drimzi

[KandiVan]

9 minutes ago, Drimzi said:

Well it's not EXACTLY 99.28, just do what I did. Use wolfram http://www.wolframalpha.com/input/?i=π×(1600%2F200), click more digits until it won't let you do anymore, put that into the calculator using distance mode at 90 degrees, get the sensitivity value, switch to sensitivity mode, convert from that to your games.

So like this then? 

https://gyazo.com/d24e796f295ef528b4f4c9942eadd176

https://gyazo.com/f8e0a353a5b1da82e230a9dc822952af

 

 

Edited July 28, 2017 by KandiVan

[Drimzi]

Edited January 28, 2018 by Drimzi

[KandiVan]

6 minutes ago, Drimzi said:

No the input needs to be 90 degrees, you have it at 106.26. Change the fov type to resolution base and enter 90.

Woops, must've reset without me noticing, definitely tried switching it. So this should be correct then, https://gyazo.com/65ea15f1d9a8d8d912a52f702a0cda90

Edited July 28, 2017 by KandiVan