80.lv talked to Motion Twin and discussed the ideas and technology behind the recent pixel art roguelike action platformer Dead Cells.
Motion Twin is a zero hierarchy workers cooperative that’s been around for 16 or so years now. We started out as a bunch of guys making games in a little French apartment in bordeaux. This was back at the beginning of the modern web and we used to sell them to third parties to use as mini games. From there we made our own sites and well, the rest is history.
So if you think about that situation, a bunch of guys making games… There is no boss there is no regimented structure. There is just communication between individuals in order to organise in a way that allows for effective production. Once we started to make money we needed a legal structure for the government, but we wanted to maintain that freedom to do what we wanted together.
It’s been a continuous adventure and it demands a lot more structure, organisation and rules than a traditional hierarchical command structure would. But the results speak for themselves. We’re all invested 150% in our own company, we get to experiment with whacky ideas, like a real time text based cooperative zombie survival base defense game where 40 real people have to get organised and defend a town… I mean what producer in their right mind would let that happen?
So yes it can work, it takes lots of effort and passion, but that’s the point we’re only interested in having people who love games as much as we do get involved for the long haul.
Dead Cells started out long long ago as the spiritual successor to Die2Nite (the game I mentioned above). As such it was originally a F2P, multiplayer, web/mobile tower defense game… We made a solo proto so that we could show off the core concept at games shows and we quickly realised that it was more fun than the multi version. The input of multiple team members and several reboots finally made us accept the fact that we didn’t want to make mobile/web games anymore, but rather PC/Console titles for a more ‘core gamer’ audience (i.e. ourselves hehe).
Our starting point was a medieval universe in pixel art, so something we like a lot, but about as classic as you can get. So we knew that we’d need to have a very strong art direction in order to stand out from the crowd of pixel art games already out there. This need to stand out informed pretty much everything from the choice of colours, the lighting effects and the animations that we wanted. We were really aiming for that “neo-retro” feeling of a game with a classic base but very modern effects.
For the backgrounds the AD is based on three core elements: a saturated pallet, Celtic architecture/environments and alchemy. This is how we maintain coherence in between very different biomes.
It’s all done in photoshop by hand. The backgrounds and decorations are done in PA and then redrawn in normal maps. This allows us to use our dynamic 3D lighting system to create different tints and reliefs on our items. Thanks to the normal maps, a torch on the left of a statue will light up the statue from the correct direction while respecting its base colours. This is how we are able to change the entire ambience of our levels without having to redraw all of the elements in the decor in the colours that we want for that particular biome.
For the animation of the main character and the monsters we use 3D models that we render as 2D sprites with a little homebrew tool we whipped up. The models and animation are all done in 3DS Max by one guy. Therein lies the reason we chose to go down this path. This workflow allows one person to create a lot of animations very quickly and efficiently. However the most important advantage is that we can modify any animation easily in order to adapt it to the gameplay that we want, so if the developer wants to add weight to a weapon, you don’t have to spend hours redrawing the whole thing.
The generator is split up into the distinct tasks, which are based on the type of random that you find in Spelunky, Left4Dead or Diablo 3. We watched a stack of talks and post mortems from these guys in order to understand what they did right and where they went wrong. Basically we use about 50% procedurally generated and 50% hand-crafted content in the game. We keep a short leash on the the RNG so that you don’t find yourself in undesirable situations (not enough/missing loot, boring or useless rewards, impossible levels etc). Having said that we’ve still got a lot of work to do on the generator until we’ll be 100% satisfied with the result.
Well the references for this type of feedback are all fighting games like Street Fighter 4, Blue Blaze or Mark of the Wolves. We use a lot of particles, stop frames, slow downs and other techniques taken from these type of games. For example, the critical hits freeze the game for 1 frame, followed by a slow down of a few tenths of a second, a nice blood spray and a specific impact sound feedback. It’s the sum of all these things that create that feeling of weight when you introduce a big old sword to a zombie’s skull.
What are the next steps for Dead Cells?
Ohhhh tons of stuff! New weapons and skills to give you more variety and possibility when it comes to combos and interesting builds. More levels organised into more branches to give you more choice when it comes to exploring the world and choosing the direction of your run. More bosses, more challenge areas, speed running daily challenges and leaderboards… We’re not short on ideas, we’ve just got to figure out how to prioritize it all and get it into the game.