Planetary illumination

Status
Thread Closed: Not open for further replies.
Kristov is quite right... it's not so much "experience" (and I believe the OP has been more than "nasty" in his presentation), but the more lights you add to a scene, the more computationally expensive it is. FD could probably easily do it 100% realistically. But if they did, they have to justify the extra work that will be available to a probably small % of people with the hardware that will run the game at a decent speed. You can fake *some* things, but FD (rightly in my opinion) try and get things as realistic as they can.

Add in some binary/trinary star systems and things get interesting really fast.

Multiple light sources are normally easy. But probably not on the scale ELITE is trying to do them. Common shortcuts stop working when you're doing the maths for projecting shadows through a planets rings onto the wheels of an SRV. At least I assume that's the issue.

It'd be curious to read what the technical issue is.
 
Multiple light sources are normally easy. But probably not on the scale ELITE is trying to do them. Common shortcuts stop working when you're doing the maths for projecting shadows through a planets rings onto the wheels of an SRV. At least I assume that's the issue.

It'd be curious to read what the technical issue is.

Sorry, perhaps my wording was bad - on a local scale, you're dead on - but on a solar system scale, probably not easy at all (or rather, easy conceptually - not easy real-time on various lines of graphics cards).
 
fdevs have to promise us they will fix this along with multiple light sources implementation. Apparently a year long rant from multiple commanders do make devs make some changes (SCBs are finally getting some change (not much though, if I understood them right)) to how game works :D
 
I was the one who pointed out some personal thoughts about the nightsky and the visible stars...
but then i realized that..

daaam... it's just a game after all.. it's not a definitive simulation of the real universe.
 
I was the one who pointed out some personal thoughts about the nightsky and the visible stars...
but then i realized that..

daaam... it's just a game after all.. it's not a definitive simulation of the real universe.


Daaam... it's just a movie after all.

[video=youtube;8B6jSfRuptY]https://www.youtube.com/watch?v=8B6jSfRuptY[/video]

;)
 
OP is correct. SpaceEngine does it correctly. ED doesn't, at least, not yet. This has been raised before. I imagine it's on the bug list.

Cheers,

Drew.
 
OP is correct. SpaceEngine does it correctly. ED doesn't, at least, not yet. This has been raised before. I imagine it's on the bug list.

Cheers,

Drew.

Hate to break it to you, but no. While Elite might overdo it slightly, OP is still wrong. And yes Space Engine does do it right, as in just like Elite, but slightly better:

It is all a matter of perspective, OP. Each of these screenshots were taken from the same timeframe. But each one was taken from a different perspective. And as you can see... There's a variety of ways to render the terminator depending on the angle you are observing the world from. This was done in Space Engine, by the way.

http://i121.photobucket.com/albums/o232/Mr_Blastman/scr00092_zpsdosqw7gf.jpg

http://i121.photobucket.com/albums/o232/Mr_Blastman/scr00094_zps9wenrzk2.jpg

http://i121.photobucket.com/albums/o232/Mr_Blastman/scr00093_zpsklcqvob9.jpg

I don't think Elite is doing a bad job.

As you can see here Space Engine also illuminates more than half the planet, how it should be and ALWAYS must be (except for neutron stars and white dwarfs).
It is literally physically and mathematically impossible for any other star to illuminate only half of a planet, even if it is just by an immeasurably small amount. Stars are always bigger than planets, therefor will always cover more than half the surface, no matter the distance.
Obviously disregarding weird potato planets and as said before neutron stars and white dwarfs, since these can be smaller than planets.
 
Last edited:
I would imagine the rocky airless worlds in Horizons will be lit correctly (or correctly-ish in a single source sort of way!) from a distance, if only because it will look rubbish when you get closer otherwise.
 
Last edited:
Multiple light sources are normally easy. But probably not on the scale ELITE is trying to do them. Common shortcuts stop working when you're doing the maths for projecting shadows through a planets rings onto the wheels of an SRV. At least I assume that's the issue.

It'd be curious to read what the technical issue is.

One of the most oft cited thing you can do to increase frames per second in video games - kill shadows. Lighting and shadows are very resource intensive, and dynamically altering them in real time compounds that cost.

Want to see this for yourselves? Pick a decent game you have that has dynamic lighting and shadows. Bump the lighting/shadows options to their max settings, enter the game and check your frames per second. Now, kill the shadows totally, see the difference in your frames per second. Witcher III is good for this, as is just about any game released in the past few years that has good dynamic lighting and shadows in it. Keep in mind, MOST of these games, even though they have dynamic lighting and shadows and use multiple light sources, they do NOT create dynamic shadows for anything but the character/avatar of the player, everything else is typically baked in. It's a great way to make it APPEAR that you have multiple dynamic light sources without actually using them on everything in the game space in real time, basic trick that works good in enclosed environments, combined with various LoD and PoV tricks to cull things from being rendered.

Elite, no baked shadows, you can see this by going into a big station that has the mailslot facing a star and watching the light/shadow interplay as the station rotates, some good videos on Youtube showing this. Single light source, dynamic shadows, and it's not too hard on our systems, provided we've got decent hardware, if you crank up the details. Add another light source and things change rapidly and for the worse. And we've got systems with 3 or more major light sources in them, I've seen 5 G and F class stars in a single system all within 1000 LS of the central star, that's a LOT of dynamic lighting and shadows to be calculating and rendering in real time.

The technical issue, it's expensive in overhead, takes a lot to do the math and more to do the rendering, with the sheer size of the game space in Elite Dangerous complicating that by a huge amount. Seems like it'd be easy to most people, it's just light and shadow after all right? And that's why I said most people don't have experience with this because it's NOT easy, anyone who's done this stuff knows this and understands why Elite does it the way they do it currently. David said they are working on making it better, more light sources being used, rings on planetary surfaces, but he won't give a time scale because he can't. Trying to make it work nicely with DX9...I wouldn't want to be tasked with that, no way in hell, and forget making it work nicely on the Mac, not going to happen. DX11 or 12, yeah, that's doable without requiring multiple 970Ti's or better, but that also cuts out a lot of your playerbase...however, since Horizons is pure DX11, it would appear that they are going that way, and that's good.

We'll have to see what Vulkan brings to the table, but since neither Apple NOR MS is supporting Vulkan, I'm losing my faith in that happening...*sigh*
 
Er, guys? Guys?

You all do realise that with Horizons, the planet generation engine will be totally rebuilt from the ground up? Whatever problems or bugs exist with the current planets, they will be gone after the 25th with the old planet generator.
 
Why don't you all look at this video again:

https://www.youtube.com/watch?v=0jE7UhIyRnA&feature=youtu.be

Look at the planet's shadow. Look how its edge perfectly encircles the equator, and how it appears straight or curved depending on our point of view relative to the direction of the light source.

/end thread.

That looks an awful lot like a white dwarf, therefor being one of the only exceptions to the rule. Just saying.

Anyways, just found a perfect example in Space Engine while messing around, where it's exactly as drastic as in Elite, and absolutely correct at that:
1.jpg2.jpg3.jpg

So I guess we have proven more than enough now that Elite is right and OP wrong, both with real pics and simulations. So yeah I can at least agree to this:
/end thread
 
Last edited:
From millions of miles away? Same difference.

Care to elaborate? The fact whether or not the star is smaller than the planet (which can only be the case with white dwarfs and neutron stars) is the very basis to whether this effect arises, no matter the distance. That's simple geometry. This even still matters at an infinite distance, believe it or not. Obviously at this distance the effect is infinitely small, but still there, that's the whole point.

So:
- All stars except neutron/white dwarf: Always more than half is illuminated. ALWAYS
- Neutron/white dwarf:
If star>planet: More than half illuminated
If star=planet: Exactly half illuminated
If star<planet: Less than half illuminated
 
Last edited:
The smallest known white dwarf is about the size of Earth. The planetoid in the video is smaller than the Earth. But that's irrelevant because geometry becomes less significant as distance increases. This is why the half moon looks exactly like a half moon, even though the sun is a lot larger.

half-moon_1024x768.png
 
Last edited:
I have ticketed this several times this year with no response. The engine for planetary illumination is all wrong. Look—
View attachment 75039

A planet partly in shadow is represented in 2D by a semicircle on the lit side joined to a semiellipse on the shadow side. That shouldn't be a problem for computer programmers. It's basic stuff. Atmospheres do not diffuse light to cause it to wrap around the night side of a planet as in these wretched in-game depictions:
View attachment 75040View attachment 75041View attachment 75042

In this one the shadow is actually behaving like the illuminated part of the planet with the terminator curving the wrong way altogether:
View attachment 75043

Now see how it works in real life, first with the Earth:
View attachment 75044

And now even with planets that are practically all atmosphere like Jupiter and Saturn:
View attachment 75045View attachment 75046

If FD still can't even work out where the terminator line should be between night and day on a planet from deep space, I fail to see how they're going to represent them accurately from the surface when planetary landings come out, and I foresee big problems in Horizons with it still being daylight on the surface with the sun way below the horizon. It's not like realistic illumination of a sphere is a complicated thing to achieve in computer programming. It's far more complicated to get it wrong as FD have done. Someone sort this out please.

Your point is not correct, it dpends on perspective, there are times night sides of planets had different shapes, shadow shape depends on where the observer is and how far away the planet is from the star, I think ED does a very good job and casting lights properly from the star of reference, stars don't cast light from one point but from the whole shape so FD did it right.
 
The smallest known white dwarf is about the size of Earth. The planetoid in the video is smaller than the Earth. But that's irrelevant because geometry becomes less significant as distance increases. This is why the half moon looks exactly like a half moon, even though the sun is a lot larger.

http://www.tiikoni.net/background/half-moon_1024x768.png

It only becomes less significant when the star is small (our Sun is a small star!) and the distances are huge, as in several AU. You keep forgetting that stars are not just a bit larger, but extremely large in comparison to their planets (from several hundred to many million times as big), so even at incredibly large distances this still matters a lot. Most planets you see this effect on are extremely small in comparison to their star and are fairly close. Therefor Elite is still right. Looking at large planets with small stars don't show this effect as prominent.
Either way, OP is still wrong and will always be wrong. Maybe the effect is too strong in Elite. To find out, we would have to do proper calculations, however OP is convinced that this effect shouldn't happen at all and that you basically shouldn't be able to see the terminator make a straight line (again matter of perspective) behind the equator, which is simply wrong and will always be wrong in all eternity. No matter how much anyone wants to discuss it. It's like trying to argue about whether 2 < 1. It just isn't.

This seriously starts to feel like the whole "acceleration stays constant"-discussion all over again.
 
Status
Thread Closed: Not open for further replies.
Back
Top Bottom