Ars Technica War Stories: Homeworld

- [choral music]

- [Rob Cunningham] Looking back the creative and technical challenge of Homeworld seemed so impossible at the time. Luckily we were all young and had no lives. We literally lived at the office. Everybody's girlfriend left them. We were all in. 

My name is Rob Cunningham. I'm the CEO of Blackbird Interactive. This is how we made Homeworld, the first 3D real time strategy game. [upbeat action music] 

I first became interested in video games in the early 90's. It would have been about 93, 94, but my first ambition was to get into film and you know, make movies. I got a job teaching drawing to some video game design students here in Vancouver. I got into video games professionally because a friend of mine Alex Garden, who was actually one of my students at the time in my drawing class, just asked me one day, hey you know, do you want to start a video game company with me? And I was like, sure nothing better to do lets do it. 

So me and Alex and like four other people got together and started Relic Entertainment. So when we started Relic, it was the summer of 1997, and we were all punks. I was 26. Alex it think was 20. We all got together at Alex's house one night to eat pizza and talk about this game called Homeworld. He described it as a merger of Command and Conquer meets Battlestar Galactica. The story is you're lost civilization going back to your home world, and the gameplay is basically Combine and Conquer, but in space. So there'd be combat, production, resourcing set in this 3D environment and that's what was going to be new about it. Sounded super intriguing and super cool and you know, and I loved the RTS's of the day and thought that this would be a really cool new thing that no one had really seem before. 

And it was like a week later that we actually went down to Seattle to pitch it to Scott Lynch at Sierra. We had just barely met each other. We really had very little in the way of a pitch, but Alex was super enthusiastic and contagious with his energy. After we left, we stood around in the parking lot. It was like an empty parking lot. It was a Saturday. There was no one there and Scott phoned Alex on his cell and said, hey you guys are green lit. All of our lives pretty much changed that day. I think it was like a month later, we had a whole bunch of money and started making Homeworld. Homeworld was a real time strategy game set in space, but what was different about it was that it was actually in 3D. At that time in the mid 90's, RTS games were all top down camera. Essentially 2D worlds with sprites of the various assets and vehicles and units moving around. Those sprites were all rendered rotations of the 3D model. So it looked kind of 3D, but it was basically 2D, it was like a slide show of 2D assets. The idea behind Homeworld was that we would not do that. We were going to do actual 3D and at the time 3D was like a new thing that was just coming out. 

So we had like doom and a couple of other games that were actually in 3D, but this was going to be the first RTS. So the plan was to set it in space, so that you would never have to render the terrane. So the terrane would have these huge geometry and texture sync and we wouldn't have to worry about that because we were in space, that was the idea. So therefore we could pile all of our budget into the ships and the textures and the effects. Because of the 3D nature of the game, we were going to do, as close to a realistic scaling as we possible could. So in RTS's at the time, the vehicle that was manufactured by the factory was like almost the same size as the factory. A tank would come out of a tank building and the tank was like 85% of the size of the building it came out of, which of course is completely unrealistic. So we thought you know how great would it be if the fighters and frigates and ships being produced were being produced from these much, much bigger ships. 

None of us had undertaken something of this magnitude and of this level of technical innovation, and to be honest even when we pitched it and started production on it we had no idea what we were getting into. [laughs] that beast just made itself and we were along for the ride. The initial budget for Homeworld was $1 million and it was supposed to take one year. That turned out to be three years and $3 million, but every time we started running out of money Alex would rush down to Sierra and convince Scott to, you know free up another million and we were all amazed that this magic was happening. 

Technically there was really just one problem on Homeworld, and that was that it was going to be a 3D game and we were going to be set in space, and the camera was going to be this mad orbital camera that you could rotate around, and zoom in and out with unprecedented flexibility. No game at that time had that kind of camera flexibility. Not even close. You could zoom right into a fighter and it would fill the screen and then zoom right out until it became three pixels, and that fighter had to fly in front of a gigantic mothership and texture had to hold up. That was a really big deal. There was hardly any graphics cards. There was the software version of the game when we put it out. No one had solved these problems in hardly any games, much less a strategy game with so many assets on the board at once. We had to come up with all sorts of solutions for polygon count, for you know, texture budget, render space. Computers were totally crap in 97 and 98, 99. I was kind of amazed anyone could even run the game when we released in it 99. 

Subsets of the camera problem involved visual orientations so you didn't get lost in the game, making positive ID on ships so that you knew who was where. The scale involved in the different ships was so extreme and the distances were so extreme that we had to come up with solution for how do you keep track of everything in a strategic game. In addition to the 3D challenge at the time, memory budget was a big deal. If I'm not mistaken, the entire texture foot print was like 32 megs. We had to come up solutions for texturing all of the ships, all of the effects, and all of the backgrounds as well. In the mid, late 90's there was no off the shelf video game development software, so everything had to be built by hand. So we had to build all of the tools and technology that would run the game sim right there in house. 

Because of the 3D nature of the environment, the player had to control the camera. We couldn't do a fixed camera for this style of game. Would've just been impossible to stay in control of your units and know where things were on the strategic board. One of our big concerns was the players losing orientation, they wouldn't know which way was up. So Erin Daly designed this magnificent camera that too into account zoom levels so that you could zoom in and out on ships very quickly. You could refocus on different elements of the ships and the environment based on selection set and orbit, but then it was fixed at the poles. That way up was always up in the game. 

Aaron Kambeitz and I designed and of the art in the game from the ships, to the backgrounds, everything about the in game environment supported this terrifying fear we had of the player losing orientations. The famous banana mothership, the big vertical Kushan mothership, I designed that way because I couldn't think of a shape that was more obviously up, down that wasn't a vertical mothership because it was cool looking, though I thought it was cool. No one had done that before, but it was so that the player, no matter where they were in the world they could look at the mothership and know that way is up and so you're never going to get lost. Same deal for the Taiidan mothership, which was basically just a horizontal slap. 

Early in the game we knew we had issues with respect to the render distance of the computer. Like you just couldn't draw everything in the world, and as units became smaller and smaller, we've realized quickly that you just couldn't tell what they were. So we implemented three basic solutions to this problem. The firs was an LOD or Level of Detail system where as the unit became smaller on screen as the camera zoomed away from it, the geometry and texture of the unit would switch between different versions of itself at lower polygonal counts until it got basically supper tiny at LOD 4. You know, now we're talking about a thing literally three pixels across on screen, but because pretty much everything turns into a blob at three pixels, we had to hand exaggerate each ships profile at LOD 4 so that its silhouette signature from maximum zoom distance was still recognizable. So this sounds barking mad, but it actually worked. The LOD 4 of the Taiidan fighters for example, they had these little wings on the back that you know, the silhouette was very important that each ship had its own character and you could tell what everything was. At LOD 4 the wings were like three times the size of the ship. So we just basically did that with every ship in the fleet. The second thing we did was, at the very limits of the render list, even with LOD 4, it was really hard to tell you know, who those guys were. Especially when their like swarming around in a great big like spaghetti bowl. So we implemented what we can the tactical overlay system, which is inspired by the air traffic control radar screens. There were these little symbols that corresponded to each class of unit, so that you had feedback on what the hell was going on in the game. 

And the third and probably most difficult solution to this 3D render and visual identity game board problem that we had, was Erin Daily's solution for what later became the sensors manager. So in the early part of the game, the plan was you would have a render sphere and then when units got to the edge of the render sphere, they would create a new render sphere elsewhere. You would have to manually toggle between the two render spheres of the game and we called them away missions, kind of like you know, like in Star Trek. When the guys what down to the planet or whatever. That proved untenable almost immediately, but Daily came up with a great solution where all of the render balls were rendered in a single epic render sphere. This was inspired by the mini map that would usually happen in the bottom left hand corner of the screen in RTS's. 

We actually tried that and we had this like 3D globe thing, but you couldn't, it was impossible, you couldn't see anything there. It was like thousands of units fighting in like these, in this tiny little corner of the screen. So we abandoned the mini map and went with a full screen mini map. Which we later called the sensors manager and each of the units Fog of War, was rendered as a blue sphere in that environment, and then we made it like sound cool and Ruskay gave it that iconic, [imitates zoom noise] sound when it zooms in and out and it was like this awesome sort of 3D digital map environment, and it became a very important part of the strategy of the game. 

Those three things are more or less how we solved this huge challenge of creating a 3D environment for a space strategy game, but not completely being confuted. When we started the company and we started the game, because we were so young and inexperienced we had no idea what we were up against. So what that created was a production environment that was kind of insane, and the pressure was so high that the first solution that we had to come up with for each problem had to be the final solution, and there was no iterative process what so ever. These things would come out of no where and be final. There's kind of an amazing thing that happens though when a creative and technical endeavor is put under that kind of pressure with those constraints. Something intervenes and just makes you come up with something great, just because you have to. It's kind of hard to describe, but it's kind of like that whole thing where like the sketch is better then the final piece. It's like all of Homeworld was just one huge sketch. 

Looking back in the time since then, some things have changed and some haven't. The creative process hasn't really changed and you know, how you really make a game. Like you plan it, there's pre-production, production, post-production, bug fixing. However, the technology base totally has and the systems and the shear horse power of everything involved. When looking back at Homeworld 1 to Homeworld 3, which we're working on now here at Blackbird, it's sort of this amazing journey of this creative process. Which is basically kind of like this background rhythm that has just never really changes, but then running past it is this insane technological, slow motion explosion that's happening. 

So like for example, the things we're doing now routinely on Homeworld 3, like literally everyday, were, and this is going to sound flakey, but they were dreams that we were having about Homeworld 2 in 2001. For example, massive megalithic structures just covered in detail that like motherships would be casting shadows onto, and strike craft would be passing in and out of the nooks and cranny's and you know, atmospherics, and like terrane gameplay in 3D. Like these were dreams we were having in the early 2000's, which were absolutely impossible to execute on, and we tried and we failed, and they couldn't be done. [laughs] and Homeworld 2 got totally canceled, and then restarted and then we made Homeworld 2 essentially as kind of like a visual upgrade basically of Homeworld 1 because that was what was technically possible in 2003. 

Early in Blackbird's history while we were still in my garage, THQ went bankrupt and they were the owner of the Homeworld IP at the time. So I thought it would be great if we could get the IP. So we had our legal team unbundle the asset from the bankruptcy proceedings so that we could bid on it and buy the IP. Because I thought you know, it's going to be worth nothing. Like no ones done anything with this thing for years, but as soon as we unbundled it, it became you know, public knowledge and soon this kind of bidding war happened and we got outbid like immediately, like blown totally out of the water and it went to Gearbox for I believe 1.3, 1.35 million. We couldn't believe it and I congratulated Randy and I was like hey you know, why did you guys buy this thing like you guys are developers, you do like Borderlands and first person shooters and stuff? And he's like we're just fans, we you know, we just didn't want it to go to someone that was going to mess it up and I was like well that's great. You know thank you for that, what's the plan? And he's like we have no plan it was an impulse purchase you know Mattel wanted it and I wanted it. So I was like hey man you know, we've been building this thing in my garage called hardware and it kind of looks and feels like Homeworld, you should check it out. 

And so we showed it to Gearbox and the Gearbox team thought it would make a good prequel to Homeworld 1. Because it's set on a desert planet and so that's what we did and that became Deserts of Kharak and that's what launched Blackbird Interactive. So the Homeworld story is kind of bound up in the story of Blackbird Interactive and Relic and kind of weird actually, now that I think about it. I think the title has aged so well because of the uniqueness of the tone and the vibe that it captured.

- [Homeworld] No ones left, everything's gone. Kharak is burning.

- [Rob] There's some cinematic elements to the experience of playing Homeworld, with the scale and the elasticity of the visual experience that really transcend the artistic or technical achievement form 99, that still holds up today. And the emotional connection that people had with the title back then also kind of transcends the medium in which it kind of came through. Like people you know, looking back remember it better then you know, then it actually was because it touched them and that feeling you had was just so unique and iconic. It's not like anything else.