In this candid and detailed postmortem, developer Dean Dodrill tells the tale of the challenging and largely solo development of Summer of Arcade hit Dust: An Elysian Tail, a title that took him over three years to create.
I had always loved video games, and thought making my own would be cool.
Late in 2008 I read an article in OXM detailing what was possible with a new programming language, cryptically called XNA. I had never programmed before, and besides a short contract doing cutscenes for Jazz Jackrabbit 2, had never worked on a video game (I'm actually an animator/illustrator by trade).
For years I had been feeling the call to make my own game -- something small, like an 8-bit NES title. The OXM article convinced me to download the free tools and just try it. Four years later, I released Dust: An Elysian Tail as the headliner of XBLA's Summer of Arcade 2012. Yeah, I can't believe it either.
This is the story of how Dust: AET came to be, along with what went right and wrong during production and launch. Admittedly it was hard to come up with a list, because when you are independent and your livelihood is at stake, you can't afford to let things go wrong.
After convincing my wife that I'd like to take "only three months" off to learn to program and release a small 8-bit-style indie game in early 2009, I downloaded Visual Studio and XNA, and joined Microsoft's XNA development community.
A week of rudimentary tutorials later, I quickly realized you have to make EVERYTHING from scratch! No level editor existed. No animation tool. No scripting or dialogue systems. No nothing -- just a bunch of otherworldly programming words. But that's what made it exciting: learning a new language (literally), slowly making my artwork interactive, and more importantly, growing my aspirations beyond the original 8-bit design.
Most of my education came from online tutorials, and while each one provided valuable insight into the process, very few were directly applicable to what I wanted to achieve. That's when I came across James Silva's book on developing for XNA, which he had written while working on The Dishwasher.
Up until that point, I had committed to working with a blocky tile editor that I had been cobbling together to create my game. His book offered concepts that allowed for much more organic world creation and animation tools. That was the eye-opening moment for me, and an image of a mysterious turquoise character running through a beautiful forest alongside a herd of deer cemented itself into my brain.
What Went Right
1. Crafting The World and Characters
First things first: I needed a protagonist. He needed a unique silhouette, easily readable from a distance and exaggerated in motion. He would be fully animated in the traditional Disney sense.
Animating "Dust", as the protagonist would be called, was time-consuming. I would rough out each motion with pencil on paper, and then do a digital pencil test. When I was happy with the motion, I would ink each frame on my light table, scan the artwork into Photoshop, and then separate the shadow layer out and do cleanup/color/shading (a technique I developed while working on my film). Finally, I would composite a CG animated weapon into each frame to create the final artwork.
Each frame would be downsampled from film res to a 300x300 block, and arranged neatly on a sprite sheet. In the end Dust himself comprised over 500 unique frames of animation, across four massive 4k sprite sheets. Dust alone used up most of the RAM available on the Xbox 360.
The character editor was created in tandem with the artwork. The editor was also customized to handle segmented characters, which ended up comprising the rest of the cast.
Dust (the early working title, later changed to Dust: An Elysian Tail to make it searchable, and to tie it into my existing IP) would be an open-world mix of genres, where the player could gain new abilities and take on side quests from non-playable characters, and feature a simple leveling-up and loot system. My largest inspirations were Blaster Master, Monster World, Cadash, Strider, and Castlevania.
Again, with the help of James' book, I crafted a level editor that would allow me to incorporate my skills as a painter. Small pieces of background art would be digitally painted with a Wacom using a custom set of brushes in Painter, then assembled into 4k sprite sheets. Segments could be scaled, rotated, tinted, and flagged for motion and particles through the editor.
The final game ended up featuring over 230 handcrafted maps -- individual gameplay areas assembled into a number of themed regions, all accessible from a central world map. Creating the background assets and maps ended up being the most time-consuming task through the entire production. I stuck with familiar themes for areas to reflect my preference for older games. A Metroid style mini-map was retroactively implemented later in development.
One of the first systems to be written (and rewritten, and rewritten) was the particle system. This might have been one of my favorite areas to work in, making everything spark, glow, or just come to life in some way. In Dust: AET, a particle could be as simple as a floating mote, or as complicated as an entire avalanche, spawning hundreds of new particles in its wake. A dynamic weather system was also implemented to bring the world to life.
Immediately after my first sketches of Dust, I created his flying companion, Fidget. I needed a character at his side to help the narrative along, and to compliment/contrast his personality. And really, that's all Fidget was good for, mechanically. At first, anyway.
I was showing my brother how maps were assembled, using hundreds of sprites. We joked how funny it would be if the world around you were sucked into a spin move I had animated, since it was technically possible, if not a bit ridiculous. It dawned on me that I could expand on the simple throw axe/knife/what-have-you of a Castlevania and do something flashier with projectiles.
I animated Dust throwing out a projectile and programmed what would become the Dust Storm. The problem was that you had to stop running as he threw the projectile from his hand, and it killed the flow of combat. Then I realized I had this floating character following you the whole time. I had Fidget cast the projectiles, created some animation and code, and the combat clicked.
Dust: AET finally had its signature mechanic, and it looked and felt very, very cool. The next thing I needed was for it to sound very, very cool.
2. Sound and Music
It was finally time to find a composer. While I had been handling the sound effects and foley work myself, I knew that I'd need a strong soundtrack to accompany the visuals.
I first reached out to an old friend, Alex Brandon (Funky Rustic), of Unreal/Jazz Jackrabbit/Deus Ex fame. While I couldn't bring him on full time, he did end up providing a number of wonderful tracks for the final game. Shortly after, I met a couple of guys out of Ireland, who would eventually define the sound of Dust: AET and bring it to a level beyond my wildest dreams.
Let's let Chris Geehan of HyperDuck Soundworks tell the story.
Chris Geehan: When I found Dean, I was extremely nervous about whether I could convince him to give me a chance, but once I spoke to him, I was relieved by his interest and my chance for an opportunity to impress him. This started what I consider to be more of a friendship than a work relationship, and one of my most cherished friendships at that.
The work process between Dean and me boiled down to this: Dean would present me with YouTube videos or music in Dropbox for references, pointing out specific moments in the music for any of the themes he wanted to sway influence on. In sound design, it would actually be a case of us talking about a sound, and Dean saying, "One sec...", coming back to me with a video yanked from YouTube's archives, and a specific minute and second in it which pointed out the right reference for what he was thinking. It was a bit mind-blowing -- not how Dean could find these references, but how he did it within 10 seconds, nearly every time. I'm not kidding.
So on Dean's end, there was never a problem with explaining what he was thinking of. What came over time, however, was Dean's willingness to let me follow through on musical and sound ideas that we hadn't really discussed -- and when I took risks, it really began to pay off. Jumping from the Ys I - IV style to cinematic orchestras, wild synths, and epic booming percussive beds was probably the biggest risk of all.
And it never stopped being risky; every song I re-did, or wrote in that style, got bigger and more complex each time. I think I spent nearly six weeks writing the Everdawn Basin theme, and that includes rewrites. The risk made it exciting, for Dean and ourselves, albeit an occasionally stressful kind of exciting. But that comes with taking massive leaps like that --for me, anyway.
Click to download an mp3 of composer Chris Geehan comparing old and new musical styles in Dust: An Elysian Tail.
I would sit down at my piano to write a lot of the Dust themes at the start, but any musician will tell you, you write differently on each instrument, so the shower head was also used for pulling out wild melodic runs. The Everdawn Basin theme came about when I had my hair full of suds as I falsetto sang in the shower and quickly rushed out after to record down what I had sung (badly). I have always felt that I've never really been so inspired by a project as I was with Dust -- from the moment we started it, until it was done. Even now, it still feels like the biggest (and is the biggest) project we've ever been a part of, and that was inspiring to work with.
3. Good Publisher Relations
In July 2009 (six months after "real" development started) I decided to lock down what content I had and polish up a submission for Dream.Build.Play, an annual event where the best XNA / Xbox Live Indie Games games are judged. I had low expectations, since this would be the first time anyone outside of a handful of playtesters actually played my experiment, so I was quite surprised when I won the Grand Prize.
A month later, Microsoft got involved. After several months of pitches and contracts, Dust: AET was upgraded to the status of Xbox Live Arcade. I was assigned a producer to work with, Andrew Williams, who not only became instrumental in providing feedback and championing the game up to its release, but a good friend as well.
There's been heated debate these last few years about the relationships between indie developers and mega game publishers. After all, on a philosophical level, it seems almost contrary to the indie spirit -- to be working on this intensely personal project while also working with some of the largest companies on the planet.
But when I say that my partnership with Microsoft went well, that isn't just some publisher platitude or marketing bullet point. Aside from some stressful moments during the certification process (which is less about game design and more about getting the code up to snuff), everything went very -- almost absurdly -- well. They were about as hands-off as a publisher could be, but always provided valuable and objective feedback on the game (which I needed after staring at it from a foot away for nearly four straight years). They never forced me to change a gameplay feature or add in something to appeal to a different market -- they let Dust: AET grow on its own, and I think they know that's the best way to do it.
Signing that deal had charted a new direction in my life. My small three-month project was far behind me and I would spend the next three years, without breaks or weekends off, finishing the game.
4. Story, Dialogue, and Voices
I had written the outline for the story very early on, and had personally done a good chunk of the script for chapters 1 and 2. Soon, though, I realized that to really make it shine, Dust: AET needed a healthy dose of professional wordsmithery. In mid-2011, I brought on a friend, Alex Kain, to help iron out some rough spots and tidy up the script. While he was initially only tasked with editing text, his role grew to the point where I ended up crediting him as co-writer.
Alex Kain: I was originally introduced to Dust: AET through its Dream.Build.Play win. I thought it looked like a real breath of fresh air from the never-ending 3D arms race, and I decided to ping Dean through the e-mail on his website, just saying I loved what he was doing and I'd love to chat about the game.
Surprisingly, he emailed back almost immediately and we started talking about... well, everything. At the time, I was working for a mobile games studio called Venan Entertainment and I'd had a major hand in writing dialogue for an iOS shooter/RPG called Space Miner: Space Ore Bust. Dean, coincidentally, was a fan. It wasn't long before Dean was letting me play builds of the game and I was rewriting huge chunks of dialogue.
One thing that didn't change was the "amnesiac hero" plotline. Since Dust: AET was in many ways paying homage to the games we loved growing up, it only made sense to use one of the medium's most long-standing storytelling tropes. The biggest challenge we had was the "twist" at the end. The existence of a twist doesn't really come as a surprise -- it's inherent to the trope, and everyone expects one by now. A lot of time was spent making the late-game story reveals as interesting as possible.
Dean: I was extremely happy with how the story wrapped up, and feel we did a good job taking the narrative into a fresh direction from what was a seemingly familiar beginning. In fact, after launch, we had quite a few gamers expressing to us that the narrative really resonated with them.
Alex: When Dean brought me on, he was very clear that he didn't just want "cartoon fluff" -- the goal was to do a story that dealt with deeper themes. The story, from its earliest stages, had always been a dark one, but we always looked at it as being more in line with where animated shows and films had been in the early-to-mid '80s, before the Disney Renaissance hit.
The animated films that Dean and I grew up with were generally pretty dark, with stuff like An American Tail, Secret of NIMH, The Black Cauldron, and even Jim Henson's Dark Crystal (which wasn't animated, but you get the idea) all dealing with very heavy themes ranging from genocide to animal experimentation, but they also had an undercurrent of absurdity to keep things fresh for kids. That was the vibe we wanted to present with Dust, and I think it helped us craft a story with real gravitas that appealed to both children and adults while also allowing us to maintain the family-friendly E10+ rating.
Of course, even though we knew where we wanted the story to go, Dean and I labored extensively over pretty much every line of dialogue. Sometimes the changes would be insignificant, like altering the wording, while other changes had tremendous ripple effects. It got much more strenuous the closer we got to the localization lock, which is when all text in the game has to be 100 percent final so it can be sent off for translation.
Dean: When Dust: AET initially graduated to XBLA, Microsoft had asked about the possibility of adding voices to the game. I thought it would be too difficult to manage, being my first project, as I had enough challenges already. But after a couple years of development, it seemed a shame that such an epic story had to remain silent. Chris agreed, and set off to find a way to bring the characters to life.
Chris: In early 2011, I brought on Deven Mack of Toon Platoon Casting, an extremely talented voice actor I found on Newgrounds. I had really wanted to get his voice in Dust: AET somewhere, perhaps narrating, but once we started talking, we realized there was potential to bring a full cast on board. Mack was a blessing in disguise, and I speak for Dean and myself when I say we're grateful to have found him. Clear vision and plenty of directorial skills made casting a lot easier than we imagined; so much choice and good communication made the stressful parts of the voice acting project more manageable.
Deven Mack: When contacted by Chris in April of 2011, I quickly discovered that the game was going to be something very special, and I wanted to ensure that its quality of voice work reflected the great care being put into its development. I have a very extensive background in voicing TV cartoons, and had just come off directing actors for a Newgrounds Flash game called Hunters: Relic of Stars -- experiences which proved invaluable in preparing me for the monumental task that lay ahead.
Taking the indie-level budget into consideration, for months on end I focused on scouring several corners of the internet for aspiring and freelance vocal talents. The hundreds of actors I reached out to combined into roughly one thousand audition files for me to wade through. I would then narrow each character's auditions down to a top five or so actors, which I would submit to Dean, Chris, and Alex for further consideration.
Although I gave occasional tasks to my assistant and trusted second opinion, Edward Bosco, I personally handled the vast majority of the audition coordinating, recording session scheduling, directing and post editing process for the game's voice work. Having all my actors record remotely from their home studios and directing them over Skype provided some challenges in and of itself, but made the end result that much more unique and special to me. Six different countries are represented throughout Dust: AET's final cast of 40 very talented performers, many of whom were getting their first big break.
Dean: One of my greatest joys was hearing each character come to life. Working with Deven to choose from each fantastic audition, and then preparing the final lines for the game, are memories I will always cherish. It was quite surreal to hear such incredible voice talent contributing to my little indie game. Of course, once the voices started coming in, the static character portraits that I'd been using up to that point seemed a little stark. Microsoft asked if they could be livened up a bit.
That solution -- crazy as it was -- was to have animated portraits for every character in the game (over 80 unique animations). I would separate the portrait art into segments and then map/rig them onto orthographic geometry in my 3D suite. Some artwork would be mapped onto deformable geometry so that ears, cloth, and belts could sway as the character idled.
I would then traditionally animate the mouths with a general mouth pattern, and then assemble a large sprite sheet with the character animation above, mouths below.
In code, as each line of dialogue began, I would queue up the next sprite sheet (which was a multi-threading nightmare, as it had to load and run seamlessly). A pre-baked "voice analysis" would be read, and the animated mouth would display if there was voice playing. It wasn't a perfectly synched effect, but since it ended up looking about as good as most anime, I was happy with the result. I created a dialogue editor capable of handling portraits and branching options, and also handled voice analysis for the pre-baked mouth movement (a feature implemented over a nightmarish matter of hours just before submitting to cert).
The final element necessary to tell my story would involve animated cutscenes. When an accelerated deadline was imposed near the end of production, I was left with three options: do the cutscenes as motion comics (a technique I'm not fond of), remove them completely, or use a limited-animation technique similar to anime.
Despite having only animated in the full-blown "Disney" style (at 12/24 fps), I had no choice but to learn how to create limited animation in a very short time frame. During the most severely brutal period of crunch, I created over 6 minutes of hand-animated cutscenes for Dust: AET in a matter of weeks. This was in fact the very last content to be put into the game before cert.
5. PAX East 2012
Aside from the Dream.Build.Play win in 2009, and an "XBLA Announce" Trailer for E3 2010, Dust: AET remained dark as I toiled away in my dungeon. I entered the fourth year of development, and while I hadn't burned out yet -- and Microsoft remained generously patient -- I had high hopes that this would be 'the year' the game saw release.
However, in early 2012, I took a hard look at the project, and while I could finish the game by the end of the year, I would have to wait until 2013 to avoid the holiday onslaught. I had already sacrificed enough of the last three years from my family (again, with hardly any days off), and I didn't want to put them through a fourth.
Microsoft internally had high hopes for the game, and my producer arranged to have a demo shown at PAX East in Boston. This was incredibly generous of them, as they handled the booth, leaving me only to cover my own accommodations. Considering I wasn't swimming in money, I was thankful for the opportunity.
PAX arrived, and everything changed in the span of about 48 hours. I'd expected a small showing, a few interested fans, and a lot of people passing by my booth to play the other XBLA titles that were being prepped for release in the coming weeks. We started with a moderate showing -- but then Giant Bomb swung by and mentioned the game at their PAX panel. Then Mike Krahulik from Penny Arcade stopped by to check out the game. Then Cliff Bleszinski. It wasn't long before we had lines over an hour long to play the game.
Needless to say, after spending so long on the game, I'd lost nearly all objectivity. But seeing so many people playing and enjoying the game was one of the most validating moments of the game's development for me. Also, after working with him for over a year via Steam Chat by that point, I finally got to meet my co-writer, Alex Kain, in person. Between that and hanging out with my partners at Microsoft, it was a fantastic show.
Less than a week after returning from PAX, Microsoft suggested I pursue a Summer of Arcade release. How could I say no?
What Went Wrong
1. The Most Difficult Three Months of My Life
After a lot of back and forth with Microsoft, my family, and a whole lot of prayer, I said yes to Summer of Arcade. I would have to take an already massively difficult eight to 10 months of development and cram it into three. I knew I would see very little of my family during that period (despite working from home), including my newborn daughter, due just two months later.
Much of what happened after that is a blur. I was already used to working seven days a week, but now those turned into nearly 20 hour workdays. I couldn't go to bed saying "I'll finish that tomorrow" because tomorrow would bring a new set of challenges. I'm a pretty "got it together" kind of guy, but I broke down a number of times. If it wasn't for my family literally placing food between me and my monitors, I wouldn't have eaten. The days where I realized I hadn't stepped out of the house and looked at the sky for over two weeks made me realize how crazy this all was.
I can hardly remember cranking out all the cutscenes, wrapping up the remaining enemies, and populating the rest of the world during those hectic weeks. I have no idea how I was able to rewrite several massive systems -- including one to resolve a bug that threatened to cut all the voice over from the game -- just days before certification. I consider myself a man of faith, and it was certainly tested during those dark hours.
And it wasn't exactly a walk in the park for my collaborators. On top of working on multiple projects themselves, we were all caught off guard by the accelerated schedule. HyperDuck produced music and sound at an insane pace, Toon Platoon was recording and editing hundreds of lines across 40 actors seemingly overnight, and Alex and I rushed on last-minute changes to the script to accommodate the actors and to lock it down for localization.
Looking back, I honestly can't explain how I got it done. Everyone on the inside knows how deathly close we made it, from nearly cutting all dialogue and cutscenes, to issues in certification and localization -- the list goes on.
2. Biting the Bullet and Axing Features
On the subject of trimming content, Dust: AET was originally envisioned as a much larger game than it ended up being. Once I had the core mechanics down, I made the tough decision to prune my initial concept. There were actually a lot more non-combative plans, such as owning property, farming, fishing, and so on. I'd spent a lot of time planning these features and, perhaps even more problematically, envisioning them naturally as parts of the game in my head. For the longest time, Dust: AET needed these disparate components to be Dust: AET.
Most of these grander concepts were pulled simply out of necessity. It's immensely hard parting with a concept -- any concept -- that you see adding value to your dream game, but the reality of being one man and really wanting to get the game done before the universe imploded weighed heavily on my mind.
One idea, however, stuck around longer than any other: the concept of building up a small town using your amassed wealth, similar to the town-building mechanics seen in games like the latter Assassin's Creed titles. This town was where your rescued "friends" would live, and new building types would provide passive bonuses to you on your adventure. Remnants of these ideas persist in some of the materials you acquire.
Alex: This was still on Dean's plate when I came onto the project in 2011, and at first it sounded like an awesome idea. First of all, it served as a money sink, which is always helpful when you can grind enemies and wealth. Secondly, it gave players a kind of third "progression pillar", in addition to advancing the storyline and their own stats, giving players something else to build up. Unfortunately, as Dean mentioned above, he's just one guy, and Dust: AET was still far from complete.
I suggested that Dean reconsider the "build your own village" concept, but I could tell he had already crossed it out of his own internal design documentation. Once in a while he would wistfully refer to it and ponder whether he could squeeze it in, even in a less-robust state, but the realities of game development hit hard. There would be sacrifices and cuts, and the early feature creep that represented the fun and experimental beginnings of the project were long behind us.
3. Overcomplicating the Finale
Alex: Towards the end of 2011, Dust: AET was beginning to come together. It was impressive seeing the levels going up screen-by-screen, but a rather large problem was looming in the background this whole time: the story was nowhere near finished.
In actuality, the outline Dean had laid out called for an entirely new sixth chapter -- a massive third act that took the player to brand new locations fighting against a new main antagonist and changing the entire focus of the storyline. Length-wise, this finale would have tacked on another couple hours of gameplay time, which equated to another several months of development time.
In the interest of getting the game done, Dean and I resolved to cut this final section of the game and tighten up what we had, promoting General Gaius up to main antagonist and centering the game around his campaign. While this sounded great on paper, it meant a lot of late-night, multi-hour conversations with Dean where we hashed out numerous alternate story possibilities. How would we tie the caverns, meadows, and mountains into the new Gaius plotline? How would we be able to establish Gaius as a villain when you don't actually meet him until the end of the game?
For Dean, the problem was more about cutting content that he had already worked on. When you're on a team, cutting content is a fact of life. When you're one guy, every single asset you create is important. The area had already been laid onto the world map, the music had already been written, and Dean had spent years envisioning the ending in his head. Then I came in and politely (sort of) threw that out the window.
In the end, Dean ended up agreeing with my decision to cut the third act and consolidate the story we had. It was hard to part with the mysterious final villain, but the game ended up a lot tighter, and the story felt much more satisfying ending where it did.
4. Nearly Throwing Out All the Finished Voice Over
Immediately after the final voice files came in, an unforeseen issue popped up just da