[vtsrh]

Till now, I have seen about 10 different planets, from various trailers and presentations.

Clearly the game doesn't have 18*10^18 unique planets, since this is impossible. The problem this game has is, does it have enough variance in its planet generation, so that the average player will not notice a pattern.

The planets I have seen all looked mostly the same, with slightly different color schemes and animals made from generic pre-built parts. Unless the devs are only showing a couple of percent of the game, getting the 10 random samples from the game, all of which show a very similar planet is almost impossible.

[Negative1]

Indeed. Most of the tricks they use come from the holy bible of procedural rendering books (Texturing and Modeling, Musgrave et al) and are relatively superficial (and really not so difficult).

I don't argue there exists a tremendous number of configurations but I don't think they are not substantive enough to make a difference. Look at Spore if you want a case in point.

The hype train is very strong with this game because the media loves a good indie darling. I'm not convinced yet but have high hopes!

[joosters]

I agree - the 2015 e3 demo video was a disappointment to me. The whole "Let's pick a random planet to fly to" sounds awesome, but then it's a big letdown when it looks remarkably similar to the ones in the other videos.

The problem is you can have an unimaginably large number of unique planets but yet they still all end up looking the same. Sure, the coastlines might vary or the colouring may be mildly different, but if the gameplay effect is negligible, then it's not an advantage.

I was (unrealistically) thinking that there would be planets with bustling cities, unique architecture, races, cultures and technologies.

[hacker_9]

Yeah I was thinking it would be something like Mass Effect with cities generated by overlaying algorithms. Something like: 1. Generate buildings 2. Generate transport links 3. Generate mission objectives 4. Refactor city layout to make objectives harder, and so on until a relatively stable (and fun to play!) environment is created.

They could even import races from nearby planets that have already been generated if it made sense that they would have colonised the new planet also.

Obviously those steps are very high-level and the genius would be in making it all actually work together... but yeh, just seems like the hype machine was in overdrive and really this is just Spore 1.1

[pavel_lishin]

There's no better way to spend a half hour online than to look at people's projects that generate cities. If they do a decent write-up along with some videos and demos, it really shows you just how difficult it can be for a simple fly-through 3D video. I can't imagine generating whole cities and cultures that you can interact with.

[hacker_9]

Yeah it can get complex, I even know someone who got a job because they showed off a city they created with procedural generation in an interview! But that was just layout of buildings and roads, the next part of bringing it to life with people I imagine is even more complicated.

[obstinate]

> sounds awesome

I think that's why people keep on making games like this, but all past attempts have utterly failed at being interesting, except where human construction and ingenuity impinges on the random.

[Osmium]

> Clearly the game doesn't have 18*10^15 unique planets, since this is impossible.

I don't think this is impossible, since most of those planets will never be visited and hence will never have to be generated. It's possible to have that many planets in principle with just a handful of parameters, but if no player ever has the time to visit more than a few...

[vtsrh]

I was talking about having unique planets. I never said they don't have some large-number of them.

[Osmium]

Right; that's what I'm talking about too. They can have that many unique planets in principle which (due to lazy evaluation, you could say) they only need to 'assemble' a small number of them. But they can all be unique. I agree with you that it's a problem of how to make them seem sufficiently different to each other though.

[vtsrh]

If you are a computer then they are in-fact unique, since every hash has a different value. But that in not what is the definition of unique for humans. Since we are humans talking in English, then we have to agree on some definitions:

Unique - 1. existing as the only one or as the sole example; single; solitary in type or characteristics: a unique copy of an ancient manuscript. 2. having no like or equal; unparalleled; incomparable: Bach was unique in his handling of counterpoint. 3. limited in occurrence to a given class, situation, or area: a species unique to Australia. 4. limited to a single outcome or result; without alternative possibilities: Certain types of problems have unique solutions. 5. not typical; unusual: She has a very unique smile.

By those definitions, their planets are not unique. If your definition of unique differs from those established by the language we are using, then you have to specify this.

[Osmium]

Ah, I see. We were talking on cross-purposes :) I would favour 'similar' (or 'dissimilar') to 'unique' in that context, but each to their own.

[titanomachy]

I think both senses of unique are used in common parlance... for example, when people say "every snowflake is unique". Sure, they are similar... but not exactly the same, hence unique.

[cousin_it]

Your comment has made me realize just how easy it is for people to misunderstand each other, due to attaching different meanings to words. Of course "unique" means "distinctive" to most people. I had just managed to completely forget about it, and assumed that everyone shared the computer science meaning of "unique".

I wonder how many arguments ultimately boil down to this kind of misunderstandings over words. And it doesn't help that marketers will happily use loaded words like "unique" to create hype.

[vtsrh]

If I start using the phrase "let's define this" before starting an argument, people usually start to stare.

I think it could be done online since there is no strict text limit and it helps with understanding.

Will try it next time.

[chasing]

Yeah, I'm a little concerned (from what I've seen) that the focus might be on exploring this planet-generation algorithm and not on making a compelling game experience.

Spore suffered greatly from this: It was conceptually fascinating but really not much fun to actually play. Hopefully team No Man's Land have taken a lesson!

[glxybstr]

this is the true problem with procedurally generated content in games. portrayed as infinite content, but you end up getting an infinite amount of the mostly-the-same content. would you rather play 10 different levels or one level 10 times?