Designing Manifold Garden's believably unbelievable world and puzzles

Since video games began, part of their promise was in bringing us experiences that are not just dangerous but impossible in real life. Manifold Garden is like that; at the start, it appears to be a simple block-moving puzzle game, but before long there are impossible sights at play that demonstrate the strange implications of its world.

It’s one thing to just provide huge, yawning chasms; it’s another to have those chasms actually repeat infinitely in a kind of 3D version of the screen wrapping from games like Bubble Bobble, if the screens above and below were visible, and extending outward forever like one of those “ball machine” GIFs.

Manifold Garden is a creation lead by William Chyr, and takes its inspiration from art, film, and books as much as it does other mind bending video games. He very kindly agreed to talk to us about those inspirations, as well as the complexities of otherworldly puzzle design and more just below.

Who are you, and what is Manifold Garden?

I’m a Chicago-based artist and independent game developer. Prior to Manifold Garden, I focused mainly on installations (typically made out of balloons). I worked previously in advertising, as well as at a studio making science museum exhibits. I studied physics in college and spent a summer working at a linear accelerator lab in Italy. 

Manifold Garden is a game that reimagines the laws of physics. In the game, you explore a beautiful Escheresque world where the laws of physics are different. Geometry repeats infinitely in every direction, and falling down leads you back to where you started. You can also manipulate gravity to change your perspective and see the world in new ways. As you solve puzzles and progress throughout the game, you bring vegetation and life to a once barren world. 

Manifold Garden is inspired, of course, by the works of M.C. Escher. I'm personally brought to mind of a couple of other brain-twisting, perspective-important puzzle games: Fez, and The Witness. Have you had the opportunity to try those games? What kinds of artistic and/or gaming inspirations have helped you to conceptualize the game?

Definitely, Fez and The Witness are both some of my favorite games, and both have had a huge influence on Manifold Garden. Works that have influenced Manifold Garden include the architecture of Frank Lloyd Wright and Tadao Ando, books like House of Leaves by Mark Z. Danielewski, Blame! by Tsutomu Nihei, and of course films like Inception, Blade Runner, and 2001: A Space Odyssey. 

Looking at the work-in-progress video of the game, it's certainly fearsome! Falling off of a ledge and into what amounts to a bottomless expanse is the kind of thing that might feature in my nightmares. That's part of the fun of the game, since heights are harmless in it, and falling is essential to progress; do you figure that players might need some special incentive to "take the plunge," so to speak?

The first opportunity to take the plunge in the game is soon after the player makes it outside (usually happens around the first 30 minutes). The game doesn’t give any instructions. The player sees a chasm and a switch on the other side that they need to get to. If you’re using real world logic, it seems like an impossible situation. There’s no way to get across. Some players do walk around the building looking for a path for quite some time, before trying to jump. 

It was an intentional design decision to not give players any instructions for this part, and to let them make the decision to take the leap. In fact, once you do it, there’s an achievement called “Leap of Faith” that is unlocked. It needs to be the player’s agency that gets them to take the leap, and not an instruction from the designer. That’s all part of the sense of discovery and learning that’s critical to puzzle games. Letting the player come to their own conclusion on what they must do makes the moment more powerful. 

One thing that fearlessly inventive puzzle games frequently have in common is the possibility of weird implications or alternate solutions. Have there been any cases in the making of Manifold Garden when it turned out there was a way to solve a puzzle in an unintended way? Were you ever tempted to leave such a solution in, as an Easter egg or reward to the clever player?

We definitely saw a lot of unintended solutions to puzzles during playtesting. A large part of designing puzzles is eliminating the “bad solutions”, especially if they’re teaching the player the wrong thing. In general, the early puzzles tend to only have a single solution, because they’re designed to introduce concepts to players in a very particular order. The later puzzles in the game are more open. We do embrace alternative solutions, and in fact, many of the Easter eggs are based on these solutions that playtesters discovered. It’s always very rewarding to see what players come up with. Even after we launched, we’ve seen players solving puzzles in completely new ways. 

Puzzles must not be made impossible to solve, or else the game may become what the kids today are calling "softlocked," impossible to progress in. Did you have to pay special attention in constructing Manifold Garden's puzzles to avoid such states?

Yes, Manifold Garden doesn’t have a simple button that completely resets the puzzle state. Early on, there were puzzles in the game where this could happen, but those were removed. It was our intention to carefully avoid these situations. All the levels in the game are connected as well, so this was actually a big challenge for us as designers to eliminate “softlock” situations. 

Looking at videos of the game, it feels like it might be possible for a player to lose track of their position in these gigantic areas. Have you considered it important to ensure players have some idea of where to look for puzzle elements within those vast regions?

This was definitely a major challenge for us in designing the game, and older versions of the game had this problem. A common feedback we got from playtesters early on was that when the game worked, it worked really well. But in between, there was a lot of tedium. They were constantly lost, and those moments clouded their experience. A major focus for us was minimize the time player spent feeling lost or confused. 

We utilize a number of techniques. When designing outdoor levels, it’s important that the layout be easily recognizable. We had a rule that if you looked at the thumbnail of a level’s layout, you need to be able to immediately recognize which level it was (eg “the diamond level” or “staircase level”). Otherwise, it was too complicated. 

There are also architectural ornaments that help guide the player. For example, Manifold Garden has a lot of indoor hallways - small enclosed corridors that twist and turn. One issue we observed early on was that players would often get lost here. There are six degrees of motion with gravity change, so it was easy to get turned around. Players would often end up traveling quite a bit before realizing they’ve been going in the wrong direction. To solve this, we use windows strategically. Windows are placed along one side of each corner, and point in the correct direction. So if you’re looking down a hallway the “right” way, you’ll be looking out a window and see light. If you’re looking down the “wrong” way, you’ll just see a wall. 

It’s a subtle element, and players might not even consciously register what’s happening. However, the game uses this everywhere and is very consistent. So if a player is going down a hallway the wrong way, they’ll feel like something is wrong. 

If you’re interested, I discuss these techniques in more detail in my GDC talk here.

Manifold Garden looks like the kind of game where, despite the lack of opponents, the engine must have some special considerations to handle the repeating terrain and large distances involved. Were there any special developmental challenges for constructing the engine, or the huge maps?

The size of the levels, and the repeating nature of the levels, definitely caused (and continue to cause) a lot of problems for us, especially when it comes to optimization. 

Each structure in Manifold Garden repeats into infinity. Jumping off a ledge and falling down results in the player landing back at the place of origin. Travel far enough in any one direction, and you’re back to where you started.  We achieved this effect by replicating objects in a grid to encompass all of the player's view frustum. In practice, this means that we have to render up to 500 times more geometry in our world. 

In addition, levels connect to other levels through portals, which are doorways that open into different worlds. These portals would be impossible in real life, as you can look through a portal to see an entire world, but the backside is just a solid surface. The space is no longer regular, but is in fact "non-euclidian". We can render arbitrarily deep recursions. This creates an effect where you can see a world inside a world repeated to infinity, like when you stand between two mirrors. Each recursion, in turn, is another full instance of an infinitely repeating world. 

Much of this was made possible thanks to the work of our graphics programmer Arthur Brussee. 

Did you create Manifold Garden by yourself? Are there any other people who have contributed to the game? And, being the prime mover of a project that's taken seven years so far to realize, have you had to face any difficulty regarding keeping yourself motivated and believing in the project during its production? How much playtesting has gone into its refinement?

While I did work on the game by myself for the first 3 years of development, the final work is definitely the product of a team. I focused mainly on design, but we had important contributions from Geert Nellen and Droqen. Towards the end of development, I was too close to the game to see what was important, and their perspectives really helped refine the project and identify what was important for players. 

As for coding, I actually did very little of it in the last 2 years. A lot of the credit goes to the engineering team - Arthur Brussee completely overhauled the graphics pipeline and made the visuals of the game possible. Aaron Mills, Tyler LaGrange, Sam Blye, and Garret Polk played key roles in the systems. We also had other teams like Akupara and GameCake helping with the Apple Arcade versions. There are still a few pieces from what I wrote in the game, but most of the systems have been rewritten and improved upon. And of course, the soundscape of the game is by Martin Kvale and our composer Laryssa Okada. 

The hardest part of development for me was probably the middle of the project - around 2015. At the time, I still had a “sprint” mentality, and was focused on having the game being done. There was no end in sight, and there constantly seemed like more to do, so it was very stressful and disheartening. It wasn’t until I started to think of the project as a marathon that I started to enjoy the process. 

We did a lot of playtesting. In terms of QA, we conducted almost 2000 hours of professional QA by the time we shipped. As for playtesting, we pushed out new versions several times a week, and would bring in a new group of testers every two weeks. At times, it was pretty challenging to keep up with the volume of feedback, but it really helped us refine the game.