Planet with > 0.1 ATM
- Thread starter Exocet20
- Start date
Massively so.
To do it they'd need a way to rapidly generate (at least vaguely) believable dynamic weather systems consistent in space and time from planetary to surface scale over arbitrary planetary landscapes, within the few seconds of time that they have to generate the existing planetary surfaces, on basic gaming hardware.
(And then have a bunch of terrain, plants, animals, etc. that are believable for a planet with surface liquids, too, because the current sparse growths plants would really look out of place if the weather was all that changed)
I'm not saying it's impossible, but Flimley's "seven years" guess might be optimistic.
Actually drawing the clouds and weather would be the easy bit...
Well, because it wouldn't really change anything substantial and it would be a lot of effort. Good looking clouds are not easy, and the middleware that everyone buys and pretends is their own cloud/water implementation doesn't work on Cobra, so they'd have to write all that themselves.
I think the next step could be persistent large volcanic activity with local "weather systems" to introduce the concept or precipitation and cloud systems synchronized for multiplayer in specific locations. Even if it's just dust clouds and ash rain. Could also introduce larger scale hazardous ground cover and liquid systems with lava flows/seas.
Next step after that could be weather maps on planets, namely gas giants, which don't have too much changing weather systems. This can then include aerodynamics as influence on ship handling and hazardous/terminal pressure as a concept. Latter is needed if we ever want to drop under water with our ships.
Only after that could we expect them to go for dynamically changing weather maps and local weather patterns, seasons, storm fronts, etc. Then we should have the technology to simulate complex planets, which opens up the next frontier of filling all of that with life.
Next step after that could be weather maps on planets, namely gas giants, which don't have too much changing weather systems. This can then include aerodynamics as influence on ship handling and hazardous/terminal pressure as a concept. Latter is needed if we ever want to drop under water with our ships.
Only after that could we expect them to go for dynamically changing weather maps and local weather patterns, seasons, storm fronts, etc. Then we should have the technology to simulate complex planets, which opens up the next frontier of filling all of that with life.
the problem isn't generating weather on procedural planets, no man sky has been doing that for years.
the problem fdev will have is making that weather the same for people across client instances in the same place at the same time. So you dont have situations where clients join instances between players where one is having different weather than the other and it just spontaneously changes for the secondary client to match the primary.
The weather has to be tied to the procedural function that generates the planet so that clients get the same weather at the same place and time ...while also allowing the weather to vary over time in a way that is not obviously repetitive and fake. Maybe the original stellar forge procedural functions dont have such flexibility to incorporate that.
anything that creates persistent change is almost certainly out of the picture. There doesn't appear to be anything in place that would be able to save such persistent state changes in the stellar forge since the stellar forge generates the galaxy on your local computer as you need it, and so a persistent state change would have to be saved on the server as an override to that body and recalled in some way in the same way during generation with the state variables that had created the effect (since it wouldn't be like a static base mesh or something) ... i dont see that happening ever.
the problem fdev will have is making that weather the same for people across client instances in the same place at the same time. So you dont have situations where clients join instances between players where one is having different weather than the other and it just spontaneously changes for the secondary client to match the primary.
The weather has to be tied to the procedural function that generates the planet so that clients get the same weather at the same place and time ...while also allowing the weather to vary over time in a way that is not obviously repetitive and fake. Maybe the original stellar forge procedural functions dont have such flexibility to incorporate that.
anything that creates persistent change is almost certainly out of the picture. There doesn't appear to be anything in place that would be able to save such persistent state changes in the stellar forge since the stellar forge generates the galaxy on your local computer as you need it, and so a persistent state change would have to be saved on the server as an override to that body and recalled in some way in the same way during generation with the state variables that had created the effect (since it wouldn't be like a static base mesh or something) ... i dont see that happening ever.
I agree with your description of the main challenge but I think asteroid belts have that same persistent randomness, so does terrain generation. A long but repeating pattern of weather effects would simulate the effect without requiring actual atmospheric effects & weather to be modelled from first principles in-engine.
I'm sure FDev will want to do it the best way they can, linking it into stellar forge and I think the way sunsets have been done in Odyssey are great but as far as actually playing the game is concerned as long as there are enough constraints for the wind, rain or whatever to make sense (as is done in lots of games already) I don't think it needs to be generated from first principles or take years to re-invent the wheel.
asteroid fields are not random. they are procedural. Terrain is not random, it is procedural. The textures used on the terrain are no longer fully procedurally generated (in odyssey all the time/everywhere) because they wanted more variety that such generation wasn't giving them. But they are still procedurally determined. None of that is random or it would be different for every instance.
The complexity of the weather is surmountable - as is the making it appear non-repetitive enough and non-fakish to be simulated on current hardware. But i think what gives them further hesitation other than the fact that this work probably hasn't even begun is that the atmospheres of planets we can visit vary so much ...and what that means for weather would vary extremely depending on elements that make up the atmosphere and the temperature and whether the planet is tidally locked with star. It is trivial to generate realistic type weather for earth likes. But would the weather generation be good enough to simulate all of the other kinds of worlds within safe temp ranges that we would be able to land on? And can you have weather without simulating oceans and rivers/lakes? Can you simulate weather and not have trees/plants and wildlife?
it becomes a slippery slope of features that they're probably not prepared to implement. and the playerbase would appreciate the partially implemented nature of it even less than they do "tenuous atmosphere". And god forbid there be any inconsistencies in how things behave and their realistic chemical reactions and behavior. There would be a riot.
All that for what? doing the same exact things we can do on non-atmospheric worlds where we are already bored of doing those things there? Adding more places to do things you're tired of doing isn't going to fill that hole. The game would benefit more from working on growing the existing mechanics up...than spreading the existing mechanics over more environments.
You have identified my premise - the game doesn't have to actually make it rain or be windy any more than it has to create accurate asteroids in a belt - it just has procedural criteria (reproducibly random) that tells the game to put one of a dozen or so pre-generated effects here. I see weather as a similar effect, with a string of likely (pre-generated) weather effects.
If "The complexity of the weather is surmountable" the answer is obviously that it can't be done. Yet plenty of other games do have simulated weather effects, ED can too, and it already does similar things with rings & terrain in a similar way. It won't be 'real' weather any more than terrain or the rocks in a ring are truly generated from first principles, but it could have enough variation to maintain the illusion.
I'm not talking about what the player can do on those planets, or whether that is boring or grindy, I assume it will not be all that different from what we already have. Scan stuff, shoot stuff, get from A-B as quickly as possible then go back to A at your leisure.
except you think asteroids are some separate procedural function. They're not. the stellar forge has a starting amount of values in various variables that are associated with mass and elements and various other attributes and it uses a procedural function to create systems and stars and planets within those systems and what's left over gets put into rings and asteroid belts. The rings makeup is determined by the procedural function the rest of the system is. It's composition determined from the material leftover in the system, the rules on what elements would be leftover in the ring/belt.
none of it is random. it's all determined and interconnected. That's why it's so hard for fdev to make changes to commodities associated with elements in rings without screwing up everything else.
Weather would likewise want to be generated the same way. From the procedural function and process that builds the planet. Ensuring that the behavior players see matches the expectation based on all of the variables associated with that body.
surmountable means it can be done, not that it can't. I dont doubt there are simpler options to implement weather than trying to procedurally generate it to the same level than the stellar forge generates everything else. But that leads into the other effects that whether would imply. Weather implies bodies of liquid in places. Weather implies life in places. Weather implies hazards of types that dont exist elsewhere yet in the game. All things that would have to be implemented along with weather to varying degrees that they probably dont see the real return of value in. it would be quite costly ...but would it draw the sales with all of the limitations that would go along with those required features? What would it look like to have things like oceans, but not be able to go into them? NMS can go inside them. Do you think fdev would have that working at launch of full atmo planets?
I think they see a mountain of work for a result that just adds a new environment to do what can already be done and not enough potential sales to justify it and doing it poorly would be worse than not doing it at all.
I mis-read, my apologies
Nevertheless I stand by my view that weather doesn't need to be generated from first principles, just as an asteroid belt or planet terrain are not generated from first principles, they use pre-generated elements placed according to criteria determined by the stellar forge. Weather can too.
Obviously - like the rest of the generation - it would need to be cleverly faked.
But it still needs to look largely plausible. People don't notice too much that the basic planetary orbits are implemented as simple nested two-body gravity, or ring systems are implemented as constant angular velocity, because on the timescales we're looking at it's good enough, and it's not like anyone's seen a real one anyway.
For weather the "plausibility" requirements are higher, so actually making a convincing fake it still going to be difficult. (Possible, I expect. But not quick.)
But it still needs to look largely plausible. People don't notice too much that the basic planetary orbits are implemented as simple nested two-body gravity, or ring systems are implemented as constant angular velocity, because on the timescales we're looking at it's good enough, and it's not like anyone's seen a real one anyway.
For weather the "plausibility" requirements are higher, so actually making a convincing fake it still going to be difficult. (Possible, I expect. But not quick.)
even faking it ...weather creates required features due to the effect of weather. bodies of liquid, damage due to weather effects, flammable nature of some atmospheres (not really weather but something that will need to be considered for full atmo), clouds (not all will be comfy sparse droplets of dusty water).
That's not even considering being allowed on earth likes or other planets with life. many of these things require really fake limitations or significant changes to what we experience in the game. We only get tenuous atmo because they could ignore all of these things while providing the eye candy.
That's not even considering being allowed on earth likes or other planets with life. many of these things require really fake limitations or significant changes to what we experience in the game. We only get tenuous atmo because they could ignore all of these things while providing the eye candy.
I agree with you, except for the timescale. Assuming there has been a long standing intention to do this, they have had years to prepare.
Just a fancy word to say the same thing, really. Multiplying a random number by an offset in a billow noise texture formed from a random lattice can be part of making a randomly generated cloud or a procedurally generated cloud if you're looking to sling a five dollar word. In the days of yore before it had such value, it was usually called random generation. In the rare case people talked about technology at all in old games (gameplay used to be what mattered in games), you'll see "random".
Small steps. You could do weather on Mars-class bodies without having to model surface liquids or life more complex than what we have now. Of course, mostly what that tells you is that the first step would probably be adding some procedural clouds/haze/dust storms to existing landable atmospheric worlds, since atmospheres we can land in today span roughly the range from Pluto to Mars.
I don't think that quite pinpoints the problem. Procedural turbulent noise models exist that can create plausible-looking, non-repeating cloud patterns on a sphere. Doing that much is certainly easier than what they've already done with procedural terrain generation. It gets somewhat harder when you consider that you need to be able to generate detail down to at least kilometer scale in 3D so you have something to fly through. Something like fields of fluffy cumulus clouds is harder still, but you wouldn't physically expect that until you get to worlds with much more active weather than what they'd likely start with. But overall it's tractable, and a team that developed the ED terrain generation should certainly be capable of doing this, if it was their priority.
Hey, speak for yourself, this one bugs me on a regular basis! Static hotspots as revealed by the mapping probes are actually wildly implausible, since they should be shredded by orbital shear in just a few orbits.
They did it in 1993.
Why not in 2022 ?..
Last edited:
![[VR] wi03016](https://forums-cdn.zaonce.net/data/avatars/m/172/172660.jpg?1591551126){kind=avatar}
thicker atmospheres, basic weather system and high altitude accessable gas giants would be a great addon to the game.
FDEV is lost in details they should bring more major features like this and fine tune it afterwards.
FDEV is lost in details they should bring more major features like this and fine tune it afterwards.