Scale of planets in E: D

Zieman

Zieman
This post (and the following discussion)
https://forums.frontier.co.uk/showthread.php?t=175391&page=4&p=2696544&viewfull=1#post2696544
in the general forums got me curious enough to do some actual testing, trying to find out if the planets in E: D really are represented in 1:1 scale.

Simple, I thought, just compare actual in game size to the size reported in system map.
Then I saw someone already had beaten me to it:
https://forums.frontier.co.uk/showthread.php?t=175391&page=11&p=2698064&viewfull=1#post2698064
But something didn't seem quite right. Oh yes, IIRC the FOV setting value tells the vertical FOV, not horizontal. FDEV posting confirming that: https://forums.frontier.co.uk/showthread.php?t=168931&p=2582229&viewfull=1#post2582229

So, I took a couple of screenies of planets, wrote down the distance I took the screenies at and system map reported radiuses, calculated predicted planet diameters in pixels and compared that to the screenies. Results of 9 measurements showed a variable scale, from 0.908:1 to 1.13:1 :eek:

Started thinking - are my measurements, maths or basic assumption wrong, or is the variable scale really a fact?

First measurements. I tested some High Metal Content Planets and Rocky Planets, radiuses between 1216 km and 4100 km, from distances between 4960 km (4.96 Mm) and 26500 km (25.6 Mm). Rounding error in having the distance (in most cases) at 100 km accuracy proved to be negligible. I also trust the pixel counts Photoshop Elements gives from zoomed-in, unedited full screen screenshot. Conclusion: measurements are fine.

Then maths.
Let's denote:
half of the viewing angle of the planet = α
FOV = vertical field of view in degrees (= 60 if FOV slider set at max)
calculated planet diameter in pixels = X
Vres = vertical screen resolution in pixels
We get:
tan α = radius/distance
X/Vres = 2α/FOV <=> X= 2α*Vres/FOV
Looks legit to me. :)

We're left with the basic assumptions.

When drawing the following diagram (slightly different version), I noticed that the assumption "tan α = radius/distance" is actually false. A few test calculations showed that once distance becomes greater than 10*radius, the error in calculated diameter falls below 1% though.
View attachment 55769
From the pic we see that when using r/D as tan α, we actually calculate angle β. Correct value for angle α comes from sin α = r/D .
This correction reduced the observed variation of scale to be from 0.903:1 to 0.996:1 .

Weird thing here is that of the 9 measurements I made, 3 were of the same planet (r = 2531 km), at 3 different distances. Resulting scales were:
D = 26500 km => 0.903:1
D = 8290 km => 0.928:1
D = 4960 km => 0.996:1

How come that the closer to the planet you go, the closer the scale appears to 1:1 ?

Is the fisheye effect the culprit (objects do become distorted near screen edges)?
And has that been taken into account by making all planets actually to be 90% of their real size, so they look like 100% when close enough, because of the fisheye effect?

Additional point: I have set my FOV at 64.00000 in the config file, but surely the difference in fisheye effect strenght vs. FOV of 60.00000 is negligible?
 
Last edited:

Janichsan

J
Have you taken your increased FOV into account when calculating the radius? It's almost 7% larger than the normal 60 degree, so the viewing angles of all objects is thus also significantly smaller. That could attribute for the roughly 10% smaller radii you find.
 

Porina

Porina
What you call fisheye distortion is kinda the opposite. I think the game engine uses a rectilinear projection. This has the feature that straight lines remain straight, but particularly as the angle gets wider, the further out you go, the more stretched things become. For a narrow field of view this effect isn't significant.

The assumptions made so far sound ok, but would only hold true if the exact centre of the field of view was centred on the centre of the object being measured. As the object moves away from centre, it will appear bigger with increasing distance. Does that help at all?
 

Zieman

Zieman
Janichsan said:
Have you taken your increased FOV into account when calculating the radius? It's almost 7% larger than the normal 60 degree, so the viewing angles of all objects is thus also significantly smaller. That could attribute for the roughly 10% smaller radii you find.
Click to expand...
I'm not sure I get your point... :S

Because 1080px on my screen represents angle of 64°, I of course use 64 as the divisor in the formula to calculate expected planet radius in pixels ( X= 2α*Vres/FOV ), so the X is also that ~7% smaller than with FOV of 60.

- - - Updated - - -

Porina said:
What you call fisheye distortion is kinda the opposite. I think the game engine uses a rectilinear projection. This has the feature that straight lines remain straight, but particularly as the angle gets wider, the further out you go, the more stretched things become. For a narrow field of view this effect isn't significant.

The assumptions made so far sound ok, but would only hold true if the exact centre of the field of view was centred on the centre of the object being measured. As the object moves away from centre, it will appear bigger with increasing distance. Does that help at all?
Click to expand...
If you mean by "it will appear bigger with increasing distance" the distance from screen center, I agree - that's what we see on screen. Maybe the fisheye effect was wrong term to describe the phenomenom. :)

That's also what I meant when I speculated "the closer to the planet you go, the closer the scale appears to 1:1" - screen center right at the center of planet, but when you're close (in game), the planet fills most of the screen and the distortion stretches the object's extremities thus making it appear bigger than it is.
 
Last edited:

Janichsan

J
Zieman said:
I'm not sure I get your point... :S

Because 1080px on my screen represents angle of 64°, I of course use 64 as the divisor in the formula to calculate expected planet radius in pixels ( X= 2α*Vres/FOV ), so the X is also that ~7% smaller than with FOV of 60.
Click to expand...
Yes, that was the question. From your earlier description it wasn't clear whether you actually calculated with 64 degrees.
 

Zieman

Zieman
Cody said:
My brain hurts!
Click to expand...
I don't believe you... :p

But, just for the sake of it - TL;DR:
- suspicions abound that the planets in E: D are not as big as FDEV claim them to be
- I made some tests, results vary a bit, but it looks like the planets in game are about 90% of the size they should be
- because of the projection E: D uses, the closer to the planets you go, the bigger they look like
=> planets are (or at least appear to be) big enough

- - - Updated - - -

Parmo said:
It's those mushrooms I tells ya!
[]
Click to expand...
Depends on the mushrooms! :D

There are mushrooms and then there are mushrooms... ;)
 

fruitbat

fruitbat
Don't know about the size of the planets, but fisheyes definitely taste worse than mushrooms, although depending on the fish, may or may not be psychedelic....

On a more serious note, be interesting to do the same test in a rift, where people say the sense of scale is very different and 1 to 1.
 
Last edited:

EUS

EUS
This thread makes me feel like this fellow:
 

Attachments

  • pat_star.GIF
    pat_star.GIF
    518.6 KB · Views: 468

newman1702

newman1702
A 3d engine, depending on how it renders stuff, fov, etc, will distort objects, based on distance and angle you're viewing them from. There's also LOD-ing, which can play a role in throwing off accuracy, depending on how it's implemented. Add to that the fact that planets usually aren't perfect spheres, and that, again depending on angle and distance on-screen measurement isn't always completely accurate, I'd say that the numbers you're getting are close enough to 1:1 to attribute any differences to measuring errors caused by any or all of the stuff I outlined above.
 

Zieman

Zieman
newman1702 said:
A 3d engine, depending on how it renders stuff, fov, etc, will distort objects, based on distance and angle you're viewing them from. There's also LOD-ing, which can play a role in throwing off accuracy, depending on how it's implemented. Add to that the fact that planets usually aren't perfect spheres, and that, again depending on angle and distance on-screen measurement isn't always completely accurate, I'd say that the numbers you're getting are close enough to 1:1 to attribute any differences to measuring errors caused by any or all of the stuff I outlined above.
Click to expand...
I'd say the scale appears to be 0.9:1, all your points considered and accepted. :)
 

newman1702

newman1702
Yea, I guess my point was that these numbers would need to be way further from 1:1 than 0.9:1 is, to support a conclusion that they're not realistically scaled, such as the theory raised before that they're actually just 20-25km in diameter, which is obviously not the case :)
 

Zieman

Zieman
newman1702 said:
Yea, I guess my point was that these numbers would need to be way further from 1:1 than 0.9:1 is, to support a conclusion that they're not realistically scaled, such as the theory raised before that they're actually just 20-25km in diameter, which is obviously not the case :)
Click to expand...
Oh yes, you're right.
 
You must log in or register to reply here.
Back
Top Bottom