Make It To Break It: Planetary Annihilation’s Creation Tech

By Craig Pearson on March 27th, 2013 at 8:00 pm.

I will call you... planet!
Biomes! Double Craters! Height ranges! I apologise for shouting! This is totally cool, even for someone as RTS-unsavvy as I am. The Planetary Annihilation developers are very serious about the planets that make up the game’s maps, and have created the longest video ever to show you all the attributes players will have control over when they make their own planets. You know how some trailers and videos have production values? Not this one? It’s just two guys with a game engine that makes painterly planets, and it’s really rather fascinating.

I’ll admit I haven’t watched it all, because it is OVER AN HOUR LONG! When was the last time a developer allowed gamers to look at an hour of pre-alpha footage of their game? The answer is never, you uncouth liar. But every time I skipped to a section to make sure it wasn’t filled with adult situations and mild peril, I lost ten minutes to what they were demoing. The unit lighting, the twin craters under the ocean, the flow fields, and lets not forget the classy wipes taken directly from Windows Movie Maker.

And now you’ve watched it for an hour, go to bed and dream of planets. And someone post in the comments to confirm I haven’t accidentally posted a terrorist manifesto. I worry about that sort of thing.

__________________

« | »

, , .

43 Comments »

  1. luukdeman111 says:

    I don’t regret backing yet! I do really wonder if they will be able to live up to their pitch though… It just seems too good to be true

    • deaomen says:

      I believe the basic idea of Kickstarter is that some of the games might fail (in a terms of not living up to the hype), but the backers are willing to take this risk. Many publishers are not willing to do so and I can understand it (it’s about the money after all), but without taking risks we never get new kind of gameplay ideas or revive old genres. I believe strongly in this project, but I try to keep my expectation realistic.

    • Teovald says:

      I don’t expect PA to have the same production value as SupCom. Its budget is probably way lower : supcom’s budget has been estimated to $10 millions, PA got $2 millions from Kickstarter & I doubt that Über has invested 8 millions in that game..
      It seems like there will be only one faction, which could be great, or extremely boring.
      I don’t regret my pledge either for the moment (well, that first commander looks totally ridiculous, but it will be possible to choose between several models).

      • oxykottin says:

        Never underestimate a group of people working on a labor of love vs a studio working under the eye of a publisher.

        Kickstarter adds a level of peer pressure that is not felt in traditional game production. I would compare it to some of the things modders do that blow us away, but with the added value they get paid to do it.

        • Teovald says:

          I think it is more that indies have way more freedom to try new things. Wolfire shaps wrote a great article recently on this. If you look at Halo 1->4, most of the innovation took place in the graphics of the game, not at the core of the gameplay. Which is the complete opposite of Minecraft or Spelunky where exploring new techs was necessary to write the core of the game.
          PA seems to be trying out some interesting things as well, such as this planet level fights or procedural map generation.
          Their budget might be tight compared to supcom in areas such as play testing, ie balancing. I hope it will not be the case, but I try to stay only cautiously optimistic. Not having a publisher can also help in order to do not be tied to bullshit 10 steps dev plans or other madness but the other side of the coin is that trying out new things will have unexpected, and sometimes pretty bad side effects. These proceduraly generated planets could for example be a pain to balance.

          • P.Funk says:

            I grow weary of your counter point as I wonder how in god’s name you brought Halo into a conversation about Kickstarter and Indie developers versus Mod developers versus Chris Taylor mega projects.

            Look at MW:LL. Now imagine they owned the IP and actually got paid to make it instead of doing it for free.

            Modders is the right angle to approach this from. They have done astounding things for free. For 2 million bucks I can’t imagine how far some of the greatest alpha-progenitors of modern RTS games can go.

          • MiKHEILL says:

            Interestingly, Bungie was an indie studio until Halo was almost completed!

  2. PatrickSwayze says:

    DEM CRATERS.

    DEM LIGHTS.

    DAT AMAZING NEXT GEN NAVIGATION.

    Cannot wait for this game and happy I poured a truck load of cash in there.

    Think this might be the best kicker since FTL

    • KlaxXxon says:

      Yeah, they seem to be doing really well on the tech front. I personally think the flow field navigation should be obligatory for all RTS games from now on. Pathfinding was the one aspect with which Supreme Commander 2 really impressed me (it had the best pathfinding I have seen to this date, and now the same guy works on Planetary Annihilation).

      The lightining was also wondreful

      • Parge says:

        Yeah, they have him AND Sorian – how can this fail?

      • MajorManiac says:

        Its the first time I’ve heard of flow fields. They are a ingenious concept.

        I also look forward to seeing how Elijah Emerson uses them as he is clearly a genius.

        • The First Door says:

          I think they are one of those things which have been touted in research areas for ages as a solution for all sorts of things but haven’t really been exploited much. It’s nice to see a clever programmer type pulling them into gaming!

  3. El_MUERkO says:

    Glad to see they’re sticking away from REALISTIC BROWN and with the cell-shaded look, Sup-Com/Homeworld meets TF2 :)

  4. dillinger88 says:

    I really like how in depth these guys have been going with the tech they’re making. It’s awesome to see what the underlying mechanisms are and why the’re needed for this particular RTS.

    What frustrated me was the audience. Looked like the devs were getting pretty annoyed by some of the stupid questions. Like how many people seemed to think the pathing ‘cost’ was a game mechanic they’d personally use in the game!

    The fact that they’re considering stopping these tech discussions, because so many people in the chat were idiots (standard twitch chat, I know), saddens me.

    • JusticePS says:

      There was a lot of positive feedback about that stream afterwards, so I think we’ll probably do more of this. Possibly not quite as long-winded; Mavor said he thinks that would’ve been better if each of the two segments were more like twenty minutes.

      And no, there’s no terrorist manifesto. Just some snowcrash frames here and there.

      • buffer says:

        You should definitly keep those up !! This was fascinating.

        I understand that many people asked silly questions and did not quite got what was going on, but this was because they were expecting gameplay stuff and lots of explosion and graphics.
        On the other hand, everybody interested in computer science and particulary AI must be fascinated, I know I was.

      • Lim-Dul says:

        I watched the Youtube video with my girlfriend (not an IT person – background in art history =) and she also observed that John Mavor started looking really pissed off somewhere in the last third of the stream. I am really wondering what was going on there.

        My girlfriend just remarked that she didn’t know why the flowfield explanation went on for such a long time since to her it was quite easy to understand from the get go and the length might have been just a result of trying to cater to the people who weren’t (and wouldn’t be!) getting it in the first place.

        I for my part am gobbling up dev-logs like these, no matter how technical, without hesitation and find them immensely interesting since I’m coming from the video game industry myself.

        Keep it up and don’t be discouraged but perhaps make the segments a bit shorter indeed and stop trying to bash your heads against stupidity – just accept it.

      • epmode says:

        I wasn’t watching this live so I don’t understand why so many people in the stream had problems understanding what was going on. Personally, I loved getting such an early look at the tech side. It’s so rare to see something like that!

      • crinkles esq. says:

        From a development perspective, the pathfinding looks pretty amazing. Very excited to see the final game! I’ll probably be picking up that AI book as well, sounds like a good read.

      • darkath says:

        From my understanding people being in the stream are not listening more than they are trying to ask their question in hope they get a reply.
        People that watch just watch, people that chat just chat…

        It’s like during a conference, the two guys at the back row that chatted during the whole speech asking the dumbest question because they didn’t listen.

    • Strangerator says:

      I think that might have just been a badly worded question. Technically, players will be modifying cost because when you “park” a number of units in an area, that area’s cost will increase causing other moving units to avoid that area. I think the question is more about, “will there be ways in the game to modify the cost of areas?” Deforming terrain and unit congestion are both ways players are confirmed to be able to modify the battlefield areas in terms of movement. I’d like to see things like being able to “sludge bomb” an area to cause enemies to want to divert around (thus breaking their formation).

      Watched the whole video, and I appreciated the detail and the passion evident on the part of the developers. Glad I backed this one!

    • Runty McTall says:

      I actually think that question raised an interesting idea for gameplay. Say I want my bombers to swoop from my base towards a target, drop their bombs and return to my lines. Sure I can do that with waypoints or something but to be able to mark general areas as being “avoid this bit” because of AA guns or whatever would be very convenient – the pathing would just take them around it. Plus you could add degrees of cost to it so, for example they might avoid that area if they get a clean run but if a high damage mobile unit is shooting at them then that cost might be higher than the “avoid” cost I put on the area and they might choose to divert into it anyway as the lesser of two evils.

      It might be a UI or useability nightmare but I don’t think it is so stupid as to not be worth raising.

      • darkChozo says:

        It’d actually be interesting if they could do something like have units autopath away from threats. Like, if you know that there’s a dedicated anti-X emplacement there, then the pathfinding will automatically up the cost for an X being within the anti-X’s firing range. That way, bombers would try to skirt any anti-air and you could even have some cool stuff where they start to scatter when a hidden AA uncloaks, or something like that.

        Might be a bit costly to put together, and it’s a bit of a macro vs. micro thing, but it would be rather cool none the less.

  5. Arkh says:

    Thanks for this video, Pearson! Now I see the wrong-doings of the western imperialism and the evil empire the USA made, and I will not tolerate this anymore. The part about manufacturing homemade plastic explosives was very informative, and will help me when I will join Al-Qaeda tomorrow. Allahu Akbar!

  6. mickygor says:

    Gonna use this as a soundboard since my comments on the youtube video (I didn’t watch it live) were drowned out.

    It looks like the flat, default terrain has a cost of 0. Surely this should be a cost of 1 (or whatever the smallest increment of cost would be)? If it’s a cost of 0 there’s no way to account for the fact that it takes time to cross it regardless of how relatively easy it is to actually traverse. With a cost of 0 a unit would choose to go right around the planet rather than jump over a slight ridge that also happens to go right around the planet.

    • Shark says:

      Surely, the algorithm considers distance too.

      • mickygor says:

        One would hope, but check 46:52. It just looks like grey has 0 cost, and when you compare the green to the red of a sheer cliff, I feel like if grey’s cost were more than 0 the units would’ve just crossed the green.

        • LonZealot says:

          At the end he talks about the integration field, which considers the distance and the flowfield is then created from the integration field.

        • crinkles esq. says:

          They have some of the best pathfinding programmers in the industry working on the game…I’m pretty sure they know what the appropriate cost values should be.

    • The First Door says:

      As far as I understand it, LonZealot is right. The cost field is not all that is used to generate the flow field.

      You create a distance field which is 0 at the point you want to path to and then smoothly increases the further you are away from that point. The cost field is then added to the distance field to make an integration field. This means those cliffs end up being effectively further away from the goal than the points around them. Then you create a flow field by iterating through each point on your integration field and creating a vector pointing to the lowest scoring local square. That way, each unit only has to query the vector for the square it is on to work out the direction to go and they will ‘flow’ towards the goal no matter where they start from.

      That distance field is why you have areas with zero cost, because the vectors will then just point towards your goal thanks to the scores getting smaller towards it.

      • darkChozo says:

        Hmm, I haven’t been following PA too closely, but from the video and this discussion it seems that their “flow field” is some sort of potential field-ish pathfinding algorithm. Have they discussed at all about how they plan on avoiding local minima? Potential/vector fields always sound good, but they tend to have issues when you have deceptively low-cost areas surrounded by high-cost areas that ultimately lead to a dead end. I think the pathfinding bug in that demo looked like a unit getting caught in a minimum. And they didn’t really address the finer details, though that seems to be more them suffering from tech-people-talking-to-laymen syndrome than anything else.

        It’d be cool if they continued doing tech talk, probably separate from more gameplay-focused stuff with an explicit disclaimer that it’s all under the hood. It’s not too often that you get to see so much detail into the development process, usually it’s more looking at the finished product and saying how cool the tech powering it is.

        • The First Door says:

          Actually, they didn’t discuss how they were planning to avoid local minima. I believe there was a question about it on the live stream, but it wasn’t answered. I hope they do talk about solving some of the problems this technique presents.

        • karthink says:

          I thought he said the algorithm finds the cheapest path to the goal from the result of the integrator. This transfers the problem of getting stuck in local minima to the global (for each unit) optimizer. Assuming the unit’s path to the goal is mapped out before it begins to move (barring real time changes in the cost), there are ways to avoid this problem. Something like simulated annealing with a little more noise than you’d use in a regular optimizer should push it out of potholes in the potential field.

  7. sventoby says:

    This game is looking like it might be what I hoped Spore’s space stage would be.

  8. The First Door says:

    Being a massive programming geek that was one of my favourite development videos I’ve seen in ages! It’s a shame the presenter seemed so pissed off at the live stream questions by the end. I really hope they continue to show some of the techy stuff in the future.

  9. Arkhonist says:

    Yet SMNC is still dead… I will never forget Uber, never!

  10. Lemming says:

    I loved it, but he kept mentioning the ‘server’ which got my back up a little. Is this an online only game and I completely missed that part about it when I backed? I thought I was going to be singleplayer/LANing away TA-style.

    • Runty McTall says:

      I imagine you can run the server and client on the same machine if you wish. If you are playing a local game then it will transparent to you I expect, aside from the performance impact of one machine doing everything.

      • crinkles esq. says:

        I sure hope that’s the case, because if PA turns out to be always-online, it’ll change my purchase from Day One to End of Days.

        • Runty McTall says:

          I very much doubt they’d do this. It’s not their style at all, from what I can gather.

          I think “server” could be applied to most RTS games (and other games for that matter) in the past (seems to be a move towards peer to peer now but it’s probably still the case). It just means a software agent running on the host maching doing stuff, it doesn’t mean that you need a dedicated server machine (remote or otherwise) that cannot also run the game client and participate.

          If you set up a Half Life deathmatch game on the LAN you’re still creating a server (and it will show up in other people’s server browsers). You can run the server standalone or within the client, it’s up to you. I imagine PA will be the same (if you’re doing the persistant galactic war thing, with lots of people dropping in and out over a period of days or whatever then I imagine a standalone server would be a good idea, though).

        • P.Funk says:

          I’ll never cease to be amazed at how many prospective “day 1″ buyers have not even the cursory awareness of the products they’re allegedly so excited about.

          Its been pretty clearly explained in multiple places that they’re planning on a scalable game. This means you can have a game as big as the server you’re using can handle. It also means that you can have tiny lan parties. It means you can have AI games on a single planet hosted locally.

          RTS games live on multiplayer. Single player is nice, but its not where the bacon is. There will be single player skirmishing no doubt, but thats not what you sell people on.

          Frankly, if you want a big juicy single player experience out of a Kickstarter project then you should find a single player foused game. The downside to kickstarter is that design goals have to be strictly focused so as not to waste the very limited budget. PA is gonna be much less content bloated than a SupCom game, but the heart of it will still be the bit that counts: the multiplayer, but expanded beyond what most people ever dreamed up before.

          As for LAN, they’ve clearly and emphatically said they believe strongly in the concept of LAN parties and modding communities. No equivocation. Thats in some other podcast/livestream. Little bit of research would reveal this easily.

    • Jethren says:

      It is coded as a client-server setup, but when you do a lan the person hosting the game has their machine automatically acting as the server. For online play you also don’t have to use the servers they run themselves, anyone will be able to run their own server. No internet connection will be required to play the game, as it is one of the things they promised on the kickstarter.

  11. Excelle says:

    Really enjoyed that livestream. Although one of the annoying bits was a question they disregarded when I’m sure the poser meant something difficult. In relation to pathfinding, someone asked if it would work with ferrying and teleporting. Of course, said the chap, you just navigate your people to the teleport – it would work the same way.

    Noooo – I think he meant, will the flowfield pathfinding take into account the use of teleports and ferries? Can you click on the other side of the planet and then your units will decide the quickest way is to jump through a teleport gate to get there, or navigate an impassible sea by calling in a transport vessel?

    I guess I should ask this in the backer forums :)