IGN recently posted a video showing an eighteen-minute gameplay demo, which might be useful to anyone interested in the game:

[edit] The video-link seems to have been converted into a large black box in my browser, which I presume should be an embedded player. In case it's failing for others, I've additionally edited the link into the first line of text above.

[edit 2] It seems that the problem was indeed on my end--videos opened on their Youtube pages also failed, and restarting the browser seems to have fixed the issue. I'll leave the link in the text above anyway, as it might be useful for some.

I'm just waiting to see.

This thing started out on the hype train, which means that there's no way to get someone to give a realistic perspective on it until it's out and being experienced for a while. Honestly it reeks of paid tech-demo, but I'm not passing judgement on it until I have a chance to try it for myself.

I've played some Elite Dangerous which also uses procedural generation for its galaxy, systems, and planets (but you can't land on said planets).

On the one hand, it's pretty awesome to scroll out and see all the places you could visit (but never will without a million lifetimes). On the other hand, it's pretty 'lifeless' as others have stated. I'm mostly interested in how they're going to handle planetary landings in Elite, cause that means their generation will have to generate interesting landscape and potentially cities, which is a lot harder then placing a few suns and orbits.

What worries me most about No Man's Sky is it looks procedural. Elite Dangerous doesn't look procedural at a glance. Sure, you quickly realize there isn't a whole lot there, but NMS's planets look like My First Terrain Generator and their "galaxy" is very uniform. With Elite you get the beautiful spiral armed galaxy with tight star clusters, nebulae, and expanses of empty space, giving it a kind of "terrain" based on how far you can jump in your ship. But the brief part of NMS that I saw in Sony's presentation was just "here's a bunch of uniformly distributed stars with colored mist in the background".

So hey, I respect their achievement, but I think they need to apply some more limitations on their algorithms to try to get them to be less... 'uniform' for lack of a better word.

As to whether the game will be boring or not - that's up to the player. Some people love exploring, some people don't. There's a guy in Elite who made a name for himself by being the first to cross the entire milkyway and he documented his trip with a series of videos, so obviously he was enjoying it

... Some people love exploring...

That's really my primary point of interest. If there's fun and interesting things to find by exploring then I could get into that. One of the things that upset me about EvE was that everywhere had been explored, charted, and subjected to systematic exploitation. The feeling of legitimate exploration was hard to find. The wormhole space provided some when it was introduced, but it wasn't too long before people sussed out the system behind w-space and it became more of the same.

Ill just wait for it to be on sale - by then the reviews will tell me if its worth getting and most of the bugs will be fixed :D Getting in first is a sucker's game!

Ill just wait for it to be on sale - by then the reviews will tell me if its worth getting and most of the bugs will be fixed Getting in first is a sucker's game!

A lot of the value of playing a game at launch, to my mind (of console games in particular), is the sense of community. The ability to compare the experience with everyone else who is playing at the same time.

Particularly for an exploration-based game, playing it after all your friends have stopped strikes me as kind of a downer.

IGN recently posted a video showing an eighteen-minute gameplay demo, which might be useful to anyone interested in the game

The landing starting at 12:30 demonstrates the problem with local noise functions pretty well. That doesn't look like any planet we know of, though the clouds help hide the lack of global features to some extent.

The sad part is that it *is* avoidable, if you spend some time and effort on it. I'm not saying build an entire planet up from scratch - that's computationally prohibitive. But with layering of functions at different scales, some hierarchical/recursive features... One can effectively hide the small-scale repetition.

To me these kind of procedural generation won't ever be really fun to play unless they hit Minecraft peak (which seems to be a very tiny window).

Honestly I'd prefer if they tried to develop a procedural generation method that works offline and for a single system. I'd take one fun big playthrough through a detailed and limited set of systems rather than a big infinite blob of "meh".

That's essentially the Dwarf Fortress route to procedural generation. Let a complex finite world generation step work first, then let the world evolve in itself by certain rules.

If ever get to the point of worrying about this kind of stuff for my project, I'll definitively go with an offline size limited generation step instead of generating things on the fly forever. Yeah you might spend one hour generating a map, but I'm convinced it has the potential of be a better experience overall.

Procedural generation alone doesn't make a game fun; it's all the stuff you're allowed to do once the generation is done.

Minecraft and Terraria just use it for terrain, and then have a lot of interesting things to do layered on top of it.

StarBound and Elite:Dangerous went overboard with the procedural generation and forgot to (or were technically prevented from) add any interesting depth to their gameplay.


Whether No Man's Sky will be boring or not will have nothing to do with its procedural generation, and have everything to do with whatever else they manage to implement.


My concern with ridiculously large scale procedural generation is that it can influence how easy or difficult it is to implement other features of a game. Randomly generated worlds which the player can modify need space to store those modifications. This is Elite:Dangerous's primary problem: 100 billion star systems with a handful of stars/planets each... and the only thing you can permanently change outside of the tiny region of human-populated space is put your name on a star or planet when you're the first one to explore it. Even that single piece of data is probably going to take a ridiculous amount of space on a server farm somewhere. Even in human-populated territory, every single thing you can do is basically a randomly-generated encounter with no human polish behind it.

Games that have that kind of overextended ambition will very quickly run out of things that are actually possible to implement.

I personally think it's going to be boring.

The reason I say that is because almost every planet we've seen so far, is a pretty close representation of every other planet we've seen. The idea that it's procedurally generated is fine for the fanboys on forums. But when you show me 10 different planets that pretty much all look like they could be different colored versions of itself, then the scope of the game loses it's value.

I'd much rather have just a few dozen solar systems, but make each one an entirely different type of environment, with different vegetation, different life-forms and so on. Scope means little when I just look at screenshots and footage and see simple variations.

Why does every planet have grass colored differently? The fish in the latest 18-minute IGN video look EXACTLY like fish here on Earth. We're millions of galaxies away and for some reason, every planet looks like a colorful and abstract version of Earth for the most part.

Don't get me wrong, some of the components look cool, like being able to dogfight for no reason. Trade materials and stuff might be cool. The meta-game stuff outside NMS on Wiki Guides will be cool to read. Seeing how a million different people all upload data and learn how the galaxy works is awesome. My FAVORITE thing they are doing is building their own Table of Elements. I think that's brilliant and very cool for their own lore.

But in the end, seeing almost a dozen planets now, with only color variation to distinguish which planet is which, does not bode well for the game, in my opinion.