What the new secret feature should be

People should also lower their expectations for ELW in ED. ED does not even have global illumination nor proper anti-aliasing yet.
What is global illumination within a full scale model of the milky way? There is not so much for light to bounce off of.
Last I heard the illumination model used for the cosmos hints that the astronomical model, that which is used in modern astronomy, may not be predicting the correct amount of dust; More dust was needed, than is believed to exist, to generate the night sky in the SOL system in a way that made it recognisable, as that which we see from here on earth.

Now I know that this is not what you meant by 'global illumination', it is just that I felt a little context ought to be added to be fair.
 
Last edited:
Oh boy, this weather thought has got me to pondering resolution within randomness: Now this is the essence of fractal geometry, that which defines something as being fractal, is its self similarity at different scales. But what if you want to get a handle upon the randomness at specific scales; How do you do that? Given the nature of fractal generation, it's essence of very simple algorithms, how then do you go about integrating such a bottom up fractal generation or creation algorithm with a top down 'law of physics' restriction? ... Awesome stuff, and so very profound!

I love this game, fail to grok how folk don't see this side of it, I'm kind of in awe of the things I find generated by it, in a similar way that I'm in awe of naturally beautiful places.

Addendum: It could well be that the game is waiting upon advances in 'Constructor Theory' in order to progress in a way that is befitting of its reputation.
 
Last edited:
There's a list of graphics features that Unreal Engine 5 has which looks amazing yet are absent in ED.
Oh, can UE5 render a whole planet in the level of detail that ED/O is able, all at once?
If it can't do that, it isn't worth even considering, is it?

Funny how the latest 'State of the Art" game engines are compared to an engine that is, in gaming terms, ancient, so as to find the older engine lacking, yet the modern one has massive limitations, in comparison...
 
People keep saying that, if it was very simple it would be already done!
Producing a single static weather system is certainly - if not "simple" in terms of implementation - likely well within the level of things Frontier has already done elsewhere.

But procedural generation isn't magic and certainly isn't easy. The Elite series is renowned for its use of procedural generation not because it was the first to do it or even the first computer game to make good use it, but because each release they managed to do it far better than any contemporary competitor was. Claiming that to be simple massively understates their achievements - if it was that simple, everyone would be doing it.
There was a misunderstanding. I didn't say creating weather is easy. I only meant that synchronization between players is easy. I don't see synchronization as the main problem here. The weather itself, the rules by which it is rendered, its effect on the environment, on the player and the ship, the effect of the ship on the environment, landscape erosion, the wind and the new rules for rendering mountains in connection with it, optimizing the whole process, etc. - that's really hard to do. Very hard.

As an example of synchronizing animated noises between players, we can take the already existing asteroid belts. Of course this is not a comparable example in terms of shader complexity, but in terms of animation it works.
 
Last edited:
Oh, can UE5 render a whole planet in the level of detail that ED/O is able, all at once?
If it can't do that, it isn't worth even considering, is it?

Funny how the latest 'State of the Art" game engines are compared to an engine that is, in gaming terms, ancient, so as to find the older engine lacking, yet the modern one has massive limitations, in comparison...

Apparently was possible in UE4. I remember reading something about how UE5 makes it even better/easier.

Source: https://www.youtube.com/watch?v=otOPdOHWqWY


Not an argument to say FD should switch to UE5, but one day, someone is going to use UE5 (or 6 or 7) to make something like ED/SC/NMS/whatever.

I guess the issue with ED using something like UE is the fact that planets are generated on the fly based on procgen values, which wouldn't be possible with UE. It would work for Store Citizen though, where each planet is a fixed asset.
 
I guess the issue with ED using something like UE is the fact that planets are generated on the fly based on procgen values, which wouldn't be possible with UE
Thus proving my point, despite its immense power it remains suitable only for creating vast arenas, in its current form, which makes it not a candidate to replace the ageing COBRA engine, quite yet!
 
Thus proving my point, despite its immense power it remains suitable only for creating vast arenas, in its current form, which makes it not a candidate to replace the ageing COBRA engine, quite yet!

To be honest, I wasn't really following the conversation, so not sure whether it proved your point or not. Just pointing out what is possible based on my own understanding ;)
 
Apparently was possible in UE4.

What's shown in the video isn't remotely comparable to Elite Dangerous planets, it's more or less a bare minimum to do an animated zoom in and out of a particular spot using fixed library assets.

For example, you can see the actual terrain only exists for a small square patch (21:55)

terrain-patch.jpg


It's a nice tutorial but has no real relevance to a fully explorable planet, let alone 400 billion systems worth of procedurally generated ones..
 
There was a misunderstanding. I didn't say creating weather is easy. I only meant that synchronization between players is easy. I don't see synchronization as the main problem here. The weather itself, the rules by which it is rendered, its effect on the environment, on the player and the ship, the effect of the ship on the environment, landscape erosion, the wind and the new rules for rendering mountains in connection with it, optimizing the whole process, etc. - that's really hard to do. Very hard.
Got it - thanks. From my perspective, what I'm saying is that the synchronisation itself is easy once the generation problem is solved - but the requirement to have that easy synchronisation makes the generation problem so much harder in the first place.

(e.g. in a single-player game you could generate the weather on system arrival, then incrementally evolve it only in the bits the player can see, then discard it on system exit and start over on next arrival)

Not an argument to say FD should switch to UE5, but one day, someone is going to use UE5 (or 6 or 7) to make something like ED/SC/NMS/whatever.
UE5 still has serious performance issues reported with loading regions >20km across - not a problem at all for 99.9% of games, utterly fatal for even a relatively basic space sim.

I doubt that'll be "fixed" in UE6 or UE7 because while it's not all that difficult to support larger areas as a goal in itself (plenty of 90s-era games did, after all) it requires compromises which would make the engine less optimised for displaying the high-detail close-in environments that everyone else actually uses it for, to cover a use case that (to within measurement error) no-one needs.

The reason space sims tend to write their own engine is because the market is so small (and so split on what a "space sim" actually involves...) that no-one has produced a modern off-the-shelf engine optimised for space sims. Maybe one day someone will and there'll be a big increase as a result in the number of space sims being made - "first, make your own game engine" is a pretty large barrier to entry right now! - but I'd expect that to be something new rather than a descendant of the optimised-for-something-else Unreal.
 
...
The reason space sims tend to write their own engine is because the market is so small (and so split on what a "space sim" actually involves...) that no-one has produced a modern off-the-shelf engine optimised for space sims. Maybe one day someone will and there'll be a big increase as a result in the number of space sims being made - "first, make your own game engine" is a pretty large barrier to entry right now! - but I'd expect that to be something new rather than a descendant of the optimised-for-something-else Unreal.
What game besides Elite has a more or less real universe ? Well, not about the name of the systems, but the distance between bodies in the systems, the rotation of these bodies, gravity, etc. ?
 
What game besides Elite has a more or less real universe ? Well, not about the name of the systems, but the distance between bodies in the systems, the rotation of these bodies, gravity, etc. ?

Don't think there is one, most games don't even use full sized simulated planets, SC are one tenth scale, NMS planets scale down to the ludicrously small, basically walk around them in a reasonable amount of time, there's no orbital movement in either NMS the game or SC the tech demo, EVE Online, not as far as I am aware.......of course I haven't tried or researched all games, but of the 3 big ones NMS, EVE Online and ED, and the one tech demo SC, ED is the only one.
 
What game besides Elite has a more or less real universe ? Well, not about the name of the systems, but the distance between bodies in the systems, the rotation of these bodies, gravity, etc. ?
Kerbal Space Program, even though it's got only one solar system, but with mods it can be changed to simulate the physics of anything, fundamentally.
KSP2 had plans to add multiple systems, but it's pretty much dead, now.

Space Engine needs a mention, too.
While not really a game, but rather a universe simulator, it's pretty much accurate.
 
Back
Top Bottom