Reply To: Star Citizen – Musings
PROCEDURAL GENERATION OF PLANETS
This became a buzzword again with No Man’s Sky. We know how that turned out. But enough of that.
So anyway, I decided to visit the denizens of my very own sub-Reddit to enlighten them on the wanton obfuscation that’s going on regarding this nonsense. Seeing as those morons over there are just going to down vote it to oblivion, I decided to leave here as well for posterity. It all started like this:
“Given what your source has said you are arguing semantics, similar to your thoughts and implementation of seamless transitions (no I am not poking fun here you have stated your definition of it and are entitled to it). Take for example the gentleman in Minnesota who runs a business called Drive a Tank. The business is called Drive a Tank but only two of his four packages includes driving an actual tank. The other two have you driving a British FV433 Abbott SPG (self propelled gun, or self propelled artillery piece) and a British FV432 APC (armored personnel carrier). Neither of those vehicles are tanks by any definition of the term tank. Is he misleading customers? No, he states what his vehicles are. Should his business be called Drive a Tank? Who cares, in the end you get to drive something cool. Should CIG be calling it “procedurally generated planets”? Who cares, as long as the end result is good.”
Well at least you actually did the research, instead of just spewing rubbish like others tend to do.
I concede the point that right now – especially after yesterday’s AtV broadcast which shows what they are doing and how, that it’s probably all down to semantics. Why is that? Simple, because CIG themselves are to blame for yet another fiasco. If you search here on Reddit, Google, RSI forums etc, it is clear that the insinuation of procedurally generated planets implies the pure sense of the term as used by us devs over the years, and which isn’t open to interpretation.
That’s how the media, gamers – and everyone took it. Then they started thinking NMS, Battlescape, Dual Universe etc. All of which use the term and tech correctly.
Right now, this is what you see on Google when you search for Star Citizen procedural planets.
The minute Chris went on stage and called it “procedural generation of planets” – which btw my sources (and recently confirmed by Goon, TheAgent) say devs have asked Chris to stop using – he created expectations that the planets would be created as such.
Also, try reading these…
Don’t think Chris is the problem? Fine. Read this and this.
Then read the transcript of this interview. Specifically this part: “Q: A game like No Man’s Sky is using procedural generation of planets, how will Star Citizen be different in that aspect?”
So….
All this time, and even with the unclear methods which started all the way back to August, everyone (myself included) was under the impression that they were in fact doing procgen planets.
Then finally, the Oct 27th broadcast of AtV 3:11 made it clear that what they were in fact doing, is what most of us with experience building terrain tech, have been doing all this time. And in their StarEngine editor is no different from what you see in the likes of Grome, World Machine or with tools like Gaia (1, 2), Terrain Composer (1, 2) etc on Unity. It’s just a height map. In an editor. In which artists hand-craft most of the elements, while use the editor’s toolset to generate repeating (e.g. grass) assets, manipulating a 3D object (the Sun or star in the scene) with an attached light source etc.
All basic rudimentary stuff. What’s shocking is that as per the funding stretch goals:
They got $1m (at the $20m stretch goal) for:
“First person combat on select lawless planets. Don’t just battle on space stations and platforms… take the fight to the ground!”
And another $1m (at the $41m stretch goal) for:
“Procedural Generation R&D Team – This stretch goal will allocate funding for Cloud Imperium to develop procedural generation technology for future iterations of Star Citizen. Advanced procedural generation will be necessary for creating entire planets worth of exploration and development content. A special strike team of procedural generation-oriented developers will be assembled to make this technology a reality.”
And it’s going to take a LOT of time for ANY of that to be in the SQ42 (which has planetary missions, according to sources), let alone the PU.
The biggest problem is going to be performance. You think the game’s performance sucks now? Just wait until they finish doing the first planetary scene with all their high fidelity graphics – then you’ll see.
Then you have data storage. Storing those height maps on disk is going to be a major problem. We’re talking terrabytes of data if they even think of doing any scene that’s more than 12 sq km large.
This procgen fiasco is no different from all the other instances whereby reality takes a backseat to obfuscation. Some recent examples:
FPS headbob: they disabled headbob in fps, and all of a sudden it has a nonsensical name, “visual stabilization” to make sound like it’s something else – or some newly discovered tech.
Persistence: Same rubbish. There’s nothing persistent about Star Citizen. The ability to save and restore player data from a dB isn’t new tech, and nobody ever accused it of being “persistent” in the sense of the term as recognized in gaming tech. The game is 100% instanced, and that will never change. So no matter what is being told to backers, even if they invent some fantastic networking tech (hint: they won’t) that’s going to solve their multiplayer issues, there’s never – ever – going to be an MMO coming from this. Why? Because the design was all wrong – right from the start. All MMO games are persistent. You login, game state changes, you logout, log back in, and the state you left is no longer there. So regardless of whether or not you are on the server, the world state is consistently updated and persists. Even when servers are taken off-line, most save the entire state so that when it is brought back up, it is restored “in-place”. e.g. in Line Of Defense, you can login at a certain time, e.g. 1pm and the Sun is up, logout, come back later and it’s night time and it’s pitch dark and anything that was around during the day, is still there.
Seamless space<->planet transition: Rubbish. They’re using trigger points on the planet, in much the same way they do jump targets. Not new. Not revolutionary. Which is why in the Homestead tech-demo from the CitizenCon2016 presentation which I wrote about in my Shattered Dreams blog, you can see the area on the planetary sphere where the base is located. They jumped to it. They couldn’t pick an arbitrary point on the planet and jumped to that because it’s just a textured sphere that bears no relation to the planetary height map below. How do I know? Well watch this Universal Combat CE video (start at 12:05). I select and jump to the planet, pick a spot on the sphere, engage, and my ship appears (I use an external camera transition sequence) exactly at that spot. Why? because it’s procgen and the data for the planet, corresponds to what’s mapped to that sphere of the planet. Right now, you can download the UCCE or GALCOM Echo Squad demo on Steam and try it.
I could go on and on and on, but I’m sure that you get the idea of what I’m going on about.
So this “procedural planet” nonsense that’s now a huge bone of contention, is just more of the same. The disappointment that’s coming is when backers realize that by creating these planets the way they have, it limits the surface coverage for any planet. So instead of having large planetary surfaces – assuming they ever finish and implement it – you will end up with small planetary surfaces, with key points of interest. No different from how space has locations like ArcCorp, GrimHex etc.
The end result? Well, that’s precisely how it’s done in Line Of Defense and most games. Create a height map to define the terrain area (in LoD it’s 256 sq. km edge-to-edge). Have the artists and modelers build the assets into it. Then use scripts to add other dynamic content that’s not static (e.g. rivers, canals, bridges) in the scene. Use rendering tech for water (we use Triton), atmosphere (we use Silverlining), weather, dynamic day/night cycles etc – and a trigger point (in LoD it’s the jump gates or the location selection on the map if you are using the HAIS to go to the planet from space) to get to those locations.
That’s not procgen planets. Which, btw is what’s implemented in every single Battlecruiser and Universal Combat game since 1989. That’s why the smaller 2009 games, All Aspect Warfare & Angle Of Attack, use a different terrain tech that’s similar to what is described above.
And I opted not to do procgen in LoD because it’s a different kind of game and wouldn’t have benefited from it, seeing as I already designed the entire game’s world scope from the onset.
In conclusion, the bottom line is that, once again, backers are being sold a bill of goods that’s not representative, nor indicative of what they were promised. The $20 million stretch goal promised “First person combat on select lawless planets“. The $41 million stretch goal promised “Advanced procedural generation will be necessary for creating entire planets worth of exploration and development content“. Nothing they have shown thus far, is any of that. And my guess is that even the rudimentary FPS on planets, is a long way off, and because doing a proof-of-concept demo (as in Homestead) with a single player in a heavily scripted environment, is a lot different from the actual implementation for multiplayer. And that was promised in the 3.0 patch which was promised “end of this year“.