New Planet Tech is KILLER of Exploration (all terrain is tiling/repeating/not procedural/random)

[Valorin]

I like how some people claiming there is going to be less variety because of new tech like to cite Dr. Ross , but tend to avoid some fragments.

"Q: What type of planet is this new tech going to be applied to?
Every planet you could land on before, and the new ones opening up, will be using this approach. The old surfaces can't be represented in this new approach and you're going to get a

larger variety

using this new tech."

The truth is, we have absolutely no idea what the new tech is really capable of.
It's not like FD ever managed to get anything near intended state when they were introducing major changes.

Three days doing nothing else but landing on planets and looking at stuff I can say I got at least a rough idea what the new tech is capable of in the variety department, and it's a lot.
Worlds rarely look the same and on every world there's different areas often coming with different features, like volcanism for example, which you can see from your ship when approaching. Ice planets have a lot of different terrain textures, scatter rocks are vastly different from each other and so on and so forth.
Yesterday I landed on a planet full of canyons and found a depression with volcanic features, with a fat rock in the middle. Going there I found out that this rock was indeed the biggest fattest fumerole I have ever seen in the game.
And that doesn't even take atmospherics in account, where lighting, colours and all are hugely different.
I didn't count on how many worlds I landed so far, but it's three days. Roughly 25 hours or so in total? I don't know. I have free time at the moment and beside eating and sleeping and so on I don't do anything else.
I can't say too much about other features in Odyssey or their state, but the new planets, yeah.

Discplaimer: I know there are glitches and bugs and the LOD and the duplicate patterns and so on. Saying positive things doesn't mean I don't acknowledge the negative ones.

 

[Navigare Necesse Est]

I've been on every day, new planet every day, and not yet seen this kind of repetition.

The one thing I did note is that they have added telemetry to the game to report back what is happening. That may be the cause of the FPS drop that some people are seeing. Saying they are doing nothing is obviously wrong. A certain percentage of planets are exhibiting this, we know it's not every planet. If you really wanted to do something useful, science it out. Go out there and do a survey. What type of planet (Metal Rich, Rocky, Icy) - inhabited space or not - is it every planet - is it moons - is it planets at a certain size?

I leave that rather to the devs, it's their job after all and you and me basically pay them for it - not the other way round. The first part of your post seems to be directed at someone else. I'm sure they'll tackle the performance first and likely will improve it too.

 

[fourrunes]

Still, no assets should be used more then once on any planet.
There are not that many types of planet, it just does not makes any sense.

Don't you guys see where this is going?

When you start talking about assets in an almost infinite galaxy - there is no escaping what we are seeing here. Or at least, it's going to be very very very hard to get away with it without introducing some chaos into the assets shapes themselves.

 

[Just]

Does your work involves a 400 billion stars galaxy?

OK, so now it's time for some math.

Task is assuming tech FIXED state will be NOT repeating the same assets on the same planet, which is the only implementation making sense IMO.

If you had 1000 "assets" that can be used as "stamp" on a specific type of planet,
how exactly is
probability
of finding the same asset on two different planet in the same number of probes
changing
for three cases of the total number of planets:
a) 10.000
b) 100.000
c) 400.000.000
?


A small tip:
with all those number sets encountering a stamp you have already seen is more dependent on number of probes (explored planets) then on total number of planets.

Even then, realizing this fact depends on number of assets - but this is a non-mathematical, "human factor" part of the problem.

 

[Amy Pendergast]

Don't you guys see where this is going?

When you start talking about assets in an almost infinite galaxy - there is no escaping what we are seeing here. Or at least, it's going to be very very very hard to get away with it without introducing some chaos into the assets shapes themselves.

That's not necessarily true - you could make prefab tiles and multiply them on proc gen surfaces, voila, indefinite variation.

Edit: Multiply in the sense of calculate them as modifier not repeating

 

[fourrunes]

That's not necessarily true - you could make prefab tiles and multiply them on proc gen surfaces, voila, indefinite variation.

But it's not what they did. They must have ran into a major roadblock in connecting Horizons tech and EDO, if that's what you mean.

 

[Gwallt]

Don't you guys see where this is going?

When you start talking about assets in an almost infinite galaxy - there is no escaping what we are seeing here. Or at least, it's going to be very very very hard to get away with it without introducing some chaos into the assets shapes themselves.

That's exactly my concern - if they don't, we're stuck with the tiling. My expectation is that they'll find a way to balance this one, and we'll then only notice the lack of chaos when we look really hard... I can live with that. But as you say, the mathematics are a challenge on this!

 

[Amy Pendergast]

But it's not what they did. They must have ran into a major roadblock in connecting Horizons tech and EDO, if that's what you mean.

I just wanted to point out, that prefabs can be used to increase variety, not that anyone has done so

 

[Just]

That's exactly my concern - if they don't, we're stuck with the tiling. My expectation is that they'll find a way to balance this one, and we'll then only notice the lack of chaos when we look really hard... I can live with that. But as you say, the mathematics are a challenge on this!

You like to refer to mathematics, so I guess you understand exactly the math behind the general idea of current tech.
There is a simple mathematical task in this subject above, solving this it would really assure me that you know what you are actually writing about

 

[dannysquid]

Does your work involves a 400 billion stars galaxy?

No, but the principle is the same. Enough textures arranged in an interesting way won't be noticeable unless you're being really, really OCD about it. Procedural noise isn't some sort of perfect answer for everything. It's main flaw is that it looks pretty repetitive in it's own way. If you look at height maps of the Earth they don't look always look like a load of procedural noise functions.

But I don't really understand why these particular splodges need to be hand-made as they just look like mathematical noise anyway.

 

[Gwallt]

No, but the principle is the same. Enough textures arranged in an interesting way won't be noticeable unless you're being really, really OCD about it. Procedural noise isn't some sort of perfect answer for everything. It's main flaw is that it looks pretty repetitive in it's own way. If you look at height maps of the Earth they don't look always look like a load of procedural noise functions.

But I don't really understand why these particular splodges need to be hand-made as they just look like mathematical noise anyway.

From what's come out of today's very interesting discussion, I think the splodges are generated - perhaps to higher standard of quality - and then saved to be used as assets. So your interpretation is correct - they look like procedural generation, because in a way they are. But they've been saved and are then deployed as assets...

 

[Gwallt]

You like to refer to mathematics, so I guess you understand exactly the math behind the general idea of current tech.
There is a simple mathematical task in this subject above, solving this it would really assure me that you know what you are actually writing about

Don't know what mathematical problem you're referring to, but I have degrees in Mechanical Engineering and Economics, so my maths is decent, but focused on those fields. I know enough to know broad principles, and in some special cases, I know a lot. ProcGen is not one of those special cases, but it's statistical in nature, as is economics, so I understand the principles.

What's your mathematical background, as a matter of interest?

 

[Amy Pendergast]

From what's come out of today's very interesting discussion, I think the splodges are generated - perhaps to higher standard of quality - and then saved to be used as assets. So your interpretation is correct - they look like procedural generation, because in a way they are. But they've been saved and are then deployed as assets...

Seemingly to save time and computing power on the server side of things

 

[Gwallt]

Seemingly to save time and computing power on the server side of things

Looks that way - enhanced loading times etc.

 

[Just]

From what's come out of today's very interesting discussion, I think the splodges are generated - perhaps to higher standard of quality - and then saved to be used as assets. So your interpretation is correct - they look like procedural generation, because in a way they are. But they've been saved and are then deployed as assets...

Yes, but again - it does not really matter if those assets are proc gen of predefined in a case of "admiring" a SINGLE planet.
It's repeating of the same asset again and again next to each other that makes a SINGLE planet looking bad, artificial, unrealistic.

Of course when we are focusing on visiting multiple planets, the diifference between assets being generated "locally" (so every planet will be different) of predefined set (worse case scenario) is fundamental.

But it does not impact the "single planet" problem.

 

[Allin]

Funny that most people here defending the new tile system argue WHAT COULD BE vs what IS.

It's not how it works. That system CAN work but it DOESN'T. And the argument is Horizons didn't have that issue to such extent.

IF, as the defenders correctly suggest, there is no way to avoid it - then it shuld NOT be done in this way but instead the old tech should be expanded as opposed to being remade.

That's about it.

 

[Amy Pendergast]

Funny that most people here defending the new tile system argue WHAT COULD BE vs what IS.

It's not how it works. That system CAN work but it DOESN'T. And the argument is Horizons didn't have that issue to such extent.

That's about it.

Defending? Who? Where?

Discussing? Yep...

 

[Allin]

Defending? Who? Where?

Discussing? Yep...

Not talking about you specifically, but there are tons of posts that say there is nothing wrong with the new tech and that in the galaxy of 400 billions you can't avoid repetition.

Which is nonsense, because contrary to the use of pre-existing stencils (that are the same by definition) even if the procedural generation inevitably would reach some point of similarity in the results due to the limitation of the script, chances of it being EXACTLY the same on a sample of Elite's planet surface size are infinitesimally smaller than 1:400.000.000.000

 

[Amy Pendergast]

Not talking about you specifically, but there are tons of posts that say there is nothing wrong with the new tech and that in the galaxy of 400 billions you can't avoid repetition.

Which is nonsense, because contrary to the use of pre-existing stencils (that are the same by definition) even if the procedural generation inevitably would reach some point of similarity in the results due to the limitation of the script, chances of it being EXACTLY the same on a sample of Elite's planet surface size are infinitesimally smaller than 1:400.000.000.000

Totally with you on the second parts, that's why we discussed this and I think in the end everyone in this thread acknowledges that there is something serious amiss here - even Mr "Potato machine" (sry couldn't resist )

 

[Arthur Spooner]

I honestly don't care how many repeating patterns they come up with. I prefer the random generation for one reason. It makes for interesting exploration. You get to see strange things occasionally when the RNG cooks up something odd. Rather than something from a subset of strange things that you may have seen before on another planet... literally.

I don't find the argument of having enough patterns to pass a visual test as being relevant to exploration. It really only matters for getting screenshots. Maybe if you had millions of them, but I doubt anyone at fdev is going to craft that many. To me the RNG system is the better way to go. Yes, you'll get some funky stuff for on foot sometimes, but it is far more interesting and engaging than what we have currently, which is bland porridge compared to what it was in horizons.

Perhaps fdev can find a way to make it a setting in the game? I doubt, other than in very specific cases, the terrain being one way or the other matters a whole lot.

There is no RNG in Elites planet tech. This is why every planet - also in Horizons - looks exactly the same to every player. Procedural generation is NOT RNG.