Marian: From 3D to 2D

Hello ladies and gents. I'm the co-creator of ye olde indie game "Aquaria" and I wanted to take this opportunity to update you on the progress of my follow-up to that game: "Marian and the Fantastic World of Dreams".

While I've been involved in some neat projects, like the Winnitron 1000, I haven't released a big, commercial game since "Aquaria". For the last two years I've been working on "Marian". Since I started the project, I haven't taken a substantial break from it. I work on it almost every single day, in one way or another. It's been a huge part of my life.

Whether you've been following this project for a while or not, I feel that this is the right time to pull back the veil a little bit - for a peak of what was going on and what may happen in the near future. We've put together a collection of screenshots that nobody outside the team has seen yet. But first, here's some music from the game to get you in the mood...

The Dream

MarianIntro

Marian looks up, as something enters her room...

My brother was the one to first come up with the idea of a game about a puppet. I was really attracted to the thematic potential of a marionette main character. I was also in an emotionally rough time and the ideas of "being controlled" and "controlling others" really spoke to me. "Aquaria" also dealt with themes that I found to be emotionally important to the way I experience life.

I find that the emotional connection is what really pulls me into a new game concept. The game play, style, and world all stem from these initial inspirations. I knew that if we were going to take the game in that direction, it would become somewhat similar to "Aquaria" - very involved, hard to develop, emotionally and physically draining. Development proved to be even rougher than I anticipated, with lots of unexpected technical hurdles and an even larger doses of the usual fear and self-loathing.

TowerString

Marian's string was a very complex physics object, and she would use it to explore a 2.5D world.

Into the Third Dimension

 When we released the first 3D teaser for "Marian", there was some excitement and speculation about it being 3D. There was also some confusion from "Aquaria" fans about why the new game, unlike its predecessor, was not in 2D. Long after the release "Aquaria", I started getting into the Unity game engine, which we used to create "Paper Moon". I started to fall for the engine, because it allowed me to prototype new ideas very quickly and in visual way. It was also nice to not have to worry about any nit-picky technical details. I figured that Unity was a big step forward in terms of how fast we could develop new content. To me this seemed like a signal that the time was ripe for 3D indie games (and to some degree, that's true - lots of new indie 3D projects are popping up using Unity and they look awesome).

Cylinder

The cylinder level, from a design by Kyle Pulver.

I didn't want to do 3D just for the sake of doing 3D. Marian was always intended to use 2D game play, but since we were using a 3D game engine, we realized we could do a lot of cool things with that 2D game play, like wrap it around a 3D object or lock it to a path through 3D space. We could also use really interesting camera angles and moves to add a lot of depth to the atmosphere.

However, one of the other goals of the project was to make use of Unity's physics engine. I really wanted Marian's movement to be managed by PhysX, so that we could have the environments react to her in really dynamic ways. The combination of these two things, "2D game play in 3D space" and "physics-based game play" proved to be a technical nightmare. Although we did get most of what we wanted working with the use of a lot of creepy hacks, the levels that we had in 3D space , while they had cool design elements, never really controlled quite right.

CylinderPlant

Interior of the cylinder level.

Details, Details, Details

 Creating a world in 3D is a significant challenge, not only because there is an entire additional dimension that you have to define in detail, but realism quickly starts to intrude in tedious ways. Since a high res 3D world allows for more detail, the player expects to see it. If any of those details are off, the scene looks wrong. The level of realism has to be maintained throughout the entire project, or the immersion will be ruined.

GubbinsAttack

When "Gubbins" attack.

Creating the new Marian character model took ages. Thanks to a lot of hard work by Adam Mechtley, we finally got to a model that we were pretty happy with. But it still wasn't perfect. She'd look better from some angles more than others, and her hair never really felt real. Environment artist Justin Messner did a great job improving our asset workflow and figuring out how to build reusable 3D models that we could snap together to create areas that looked organic. But we still had a long, long way to go before we could call our look "polished". Given the amount of time it took to refine the art to the level required. I started to worry that even if we could hit that level, the game would take so long to develop that our models would look really dated by the time it was released.

At this point we had a team of about eight people who were all involved to varying degrees. Although everyone was being very kind by working for relatively low rates and trying to help out where they could, a team this large meant that I had to spend a significant portion of my time managing the team. While this was necessary, I found it really tedious. I really prefer to work with other people as a collaborator, not as a "boss" or a "manager". Some team members would take on more responsibility and discuss how their work fit into the overall game, while others were only really interested in working off a list of assets. It's very important to me to know that everyone working on a project understands the vision, and can bring part of themselves to the table to compliment and improve the end result. For a game that was to involve emotional themes, it felt wrong to have some parts of the team working in a very detached way. Overall, I feel like I was never really comfortable in the role of micromanaging people and that I never found quite the right way to direct a team of that size.

StringAndGears

Gears, gears, gears.

I was also unhappy with the design of the game, in the sense that it never really hit the right thematic tone that I was looking for. We were trying to do something really new and interesting with the string mechanics, and while we did have some cool physics-based mechanics working, it wasn't easy enough for players to pick up and play. I could probably write an entire book on the piles of different control schemes we tried out with the string mechanics. While prototyping and testing completely different control schemes was a great learning experience, it was a very complicated problem that never really had a perfect solution.

What's the Soul?

 Much like "Aquaria's" development, which started out as an action-RPG with branching plot, I realized that we had two major alternatives for where to take "Marian". On the one hand, we could continue in the direction it was going; a 3D, heavily physics-based game. On the other, we could abort the 3D version, and restart from scratch in 2D with a new outlook on game play.

Marian3DStares

3D Marian looks on...

As you can imagine, the decision was gut-wrenching and hard to make. Once you've made a significant time investment in a project (in this case, more than a year), it becomes difficult to let it go, even if you can see the difficultly in continuing. One of the problems with continuing down the 3D path was funding. I had funded the 3D prototype with a combination of my own money with matching funds from the government of Manitoba. The Manitoba funding went kaput soon after, so we wouldn't have been able to tap into that for full production. I didn't have the funds to make a 3D project of that scale happen.

For a time, we looked into publishers. While publishers would be a way to get a sizable production budget, they really didn't understand what the game was about. I feel that if I'd gone with any of the publisher options we were looking at, the final game would have ended up being a neutered, less dark and less compelling version of what I really wanted to make. If I wasn't going to go with a publisher and instead fund a 3D project myself, I would have had to greatly limit the scope of the game: we wouldn't have been able to create all the diverse environments that we planned for and would instead have had to focus the game on a few areas... and that was it. I believed that cutting those areas out of the game would destroy a big part of what could make its themes work.

TowerMockup

I think Derek really liked this area, even though it was a just quick, crappy mock-up.

I consulted with some other indie developers, including Spelunky's Derek Yu and Super Meat Boy's Edmund McMillen. Most developers I talked to agreed that 2D would suit the game well, but Derek and Edmund were more attached to the 3D version. I think they saw a lot of potential for a game that distinguished itself from other indie titles by using 3D models and environments. I disagreed with them. I didn't think that any one element would be the key to making the game work. To me, the success of "Marian" would be measured by how well the game came together as a whole. My outlook is that when someone sits down to play the game, the music, game play, level design and art should all work together to suck you into a totally unique world for a little while. That feeling wasn't coming through for me as strongly as I would have liked in the 3D prototype. I think Derek and Edmund were responding to what was there, not what could be.

The Living Storybook

While we were busy trying to build Marian in 3D, we were still developing loads of 2D concept art for all the different environments in the game. What I started to realize over time, was that the 2D art was way more successful at conveying the feel of the game than our 3D attempts.

Marian-bookingame

The Marian book, one of my favorite models from the 3D prototype.

In the 3D prototype, we had built a really nice book model for the menu system of the game. It would pop out of Marian's pocket, flip around, open up and fly towards the screen. Its pages would show you the map, details about items, etc. Around this time, we had also received a nice compliment from Ron Carmel (World of Goo) who compared the 2D concept art to that of a storybook.

These two ideas eventually fed into my feelings on restarting the game from scratch - with a new focus.

Throw it Out, Start Again

A few months ago, I decided to archive the 3D Marian prototype. I would rewrite the entire game from scratch, switching programming languages completely from UnityScript to C#. This was a good idea for a few reasons, but mainly it would force me to rethink how all the pieces of the game fit together. It would also allow me to focus on building the game from a blank slate, rather than trying to adapt existing systems that had become overly complex.

We also let go of the team at this point, eventually whittling it down to only those who were fully on board and willing to invest lots and lots of time in making "Marian" a detailed and unique world. We ended up with a team of two, myself and illustrator Ashley Dumonchelle.

I started by rebuilding the game play in a similar fashion to the original prototype. After a while, I realized that I still wasn't feeling it. The string was still too complicated to control and didn't immediately lead to enough interesting game play possibilities. I realized I could both radically simplify the way the string worked, while also removing most of the PhysX-based code that made it a pain to implement and refine. How did we do that exactly? You'll find out in the next few months... :)

Marian2DBigShot

The new 2D version, art by Ashley Dumonchelle

In addition to being able to fund it myself, free from the stress and limitations of publishers, this new version of the game would be able to run on many different platforms, would be something that non-gamers could pick up and play, thanks to the simplified control scheme, and would make use of a unique art style that was very effective at conveying the mood and themes of the game.

And that's what Ashley and I have been working on for the last few months. I wanted to make sure this was the right path before going into detail about it. Now that we've seen that this path works, and we have a good stretch of the game assembled and already being play-tested with positive results, I feel a lot more at peace with myself and this game than I ever did when working on the 3D version.

MarianSprite

Marian's new life as a 2D sprite

I think the goal of making your own "anything", whether it's a game, music, art, a book, etc is to add something special to the world. It should be unique in ways that are important to you. Whether that's a game that happens to be heavier on story or one that is more mechanics-driven or one that is kinda nuts - I believe you should be exploring your own style.

Although I certainly felt depressed while having to wrestle with difficult decisions back when I was stuck in the middle of them, I can see now that those experiences and many of the ideas that went into the old prototype will inform and improve everything in our new version of the game. The work wasn't lost, it is being converted and translated into a new form.

That gives me hope that one day we'll be able to release this game and you'll be able to enjoy it.

And then we'll feel like all this hard work, including the wrong turns and dead ends, has truly been worth it. :)

What's Next?

Discussion and updates about Marian continue on our newly minted forum. Stop by Ashley's Art Corner for regular updates on the art in "Marian". If you have any questions about development or the game in general, the forum is the place to go.

AshleysArtCorner

Check out Ashley's workspace on the forums.

You can also keep in touch by following the Facebook page for the game. We really appreciate you being along for the ride!