Kibbi Keeper Postmortem — From Wrangler to Keeper
Game: Kibbi Keeper
Release Date: 03/04/2022
Platforms: PC
Number of Developers: 16
Length of Development: ~ 6 months
Initial Concept
Our initial concept for Kibbi Keeper was a monster management game called “Monster Wrangler” it would have the player collect and keep a farm of monsters that a player could use to fight off various waves of other enemy monsters. Our Game Designer (GD) suggested that the game change from combat with the Kibbis to a focus on puzzles and have the Kibbis act as the tools the player would use to solve the puzzles. This shift left the team with a lot of room to work with as we would no longer need to implement a combat system and enemy AI. However, we had no idea how much that decision would influence the foreseeable future and change the game into something truly memorable.
Learning to be niche
When we made the switch, we realized that many of the games that we could try to pull from were extremely niche. From Slime Wrancher to Pikmin, we found that these games that fit the same genre were charming and eccentric in their style and execution. In addition, we realized that all these games nailed the idea of a unique and cute creature design that even if a person had never played the game, they could recognize the game from the monster design alone. This emphasis on creature design led our team to shift most of our focus early in production into the style and look of the Kibbis and the world they would inhabit.
Koncepting Kibbi’s
We went through many different options for the design of the Kibbis. Some are wildly different from others, but we knew we had to stress cute and memorable for each one. We narrowed the design down to a handful of creatures that our team felt fit our style by taking the designs outside the team, asking family and friends and other teams what they thought, and keeping track of which creature got the most “aww’s”. This is how we decided on the general style of the kibbi. And to take it one step further, we ran tests in the engine with the basic untextured model with variations to its design at various angles to find the one design that looked great and stood out no matter where the player saw the kibbis. I believe this helped our artists develop the best version of the kibbis instead of just going with the first design we had and running with it.
Making things Ghibli style
Our lead artist originally suggested that using a physically-based rendering (PBR) style would look good for our game which was the initial concept we ran with. Once we had created a version of the Kibbis, we threw them into VR and were able to look at all the angles and decide which style looked best. While PBR seemed cool, there would have been multiple performance concerns, and the style didn’t exactly fit what we wanted for the game. We went through various iterations and different styles, but we eventually got a look at the Ghibli style for our environment. We got a concept in-engine with some grass and a pillar rendered in the Ghibli style, and the whole team fell in love. Going back to the roots of the niche genre we were aiming for, getting the whole team to stare at the screen for a few seconds and smile, we knew we had to go with that style.
Open World Ambitions
We were warned to stay away from an open-world design structure during pre-production as the amount of time to test and get it working would be very intense. We realized, though, that for our game to have the right feeling that would match other niche games of the same concept, it would have to be an open world. The shift from combat and emphasis on visuals helped us not over-scoping our world. Our job was made easier by using level streaming in Unreal. There were still challenges for our team, especially concerning optimization. Creating a semi-persistent world is difficult as the timing of events and the loading of assets had to be considered. Everything down to the sightlines had to be considered. When a player walked into a different area but could see another area, we had to ensure that assets could be loaded and seen from there. This was both a blessing and a curse as we searched for different assets not appearing in a consistent place gave us a lot of exposure to the purely visual aspect of our game.
Kibbi Ai
Our kibbi’s AI was an essential part of the game. Anyone on the team would struggle to say which was more important for us: the visual design or the Ai. They were definitely the two sides of the same coin. Our initial AI system was an extremely complex mood system with around seven different states that the Kibbis could have. Once we started developing this system, however, we realized that getting the conveyance of each of the states would be very difficult to show and be more frustrating than cute. This led us to change the system entirely and simplify it. We shifted to the KISS method quickly, Keep It Simple Stupid. Seven became two; it’s surprising how much you can get done with only two states. I remember one of our programmers even shed a tear of joy when the change was decided.
Conclusion
We were fortunate with Kibbi Keeper’s development. Our team ran into a few problems, and it would be difficult to think of a time when development felt like it was losing momentum. I don’t think anyone expected Kibbi Keeper to turn out as it did, from the monsters and the world around them to dipping our feet into VR. This game never failed to be boring, and I am incredibly grateful that we could create this unique experience.
