The Allacrost Project has a long and storied history. If you're interested how we got started and how we ended up where we are today, read on. We share the lessons that we learned along the way in the hopes that other projects can avoid similar mistakes.
Beginnings (June 2004)
Hero of Allacrost started as an accident. One month after graduating from university and waiting for graduate school to start, the project founder Roots found himself exceptionally bored one day, and randomly decided to begin writing a short story about a young knight in a medieval fantasy. After completing it, he decided to share it on the animesuki.com forums to see what others thought. To his surprise, the praise was nearly universal. Encouraged, he wrote a little bit more and it too was highly acclaimed by readers. Roots casually mentioned that perhaps it could become the basis for a role-playing game, which was something he had wanted to create as a child. People took him seriously however, and before he knew it a number of them had expressed interest in working on this game with him. He gave in to the excitement, and together this small group formed and began what would be an incredibly long and difficult journey.
First Era (2004-2005)
The initial team was about five people, three of them programmers. One of them was gorzuate, a friend and fellow graduate of Roots. Everyone on the initial team was very young, inexperienced in game development, and naive. The team decided at the onset that the game would be made available for free and open source, as it still is to this day. The most significant motivation for this decision was Roots' desire to give back to the open source community, whose software he used freely every day. The Battle for Wesnoth, a popular open source game, was used as an inspiration for what technologies to use to build the game. One of the first actions of the team was to decide and debate on what features would go into the game, which led to some intense debate as individuals had their own preferences. What was not controversial was the game's three primary design goals, which have remained the same throughout the lifetime of the project.
Within the first few weeks a very early prototype was up and running that featured a simple boot screen and simple map drawing was up and running. Some early artwork was in the works, and we already had two original pieces of music in a usable state. This early success lead Roots and the rest of the team into a false sense of how difficult game development was and how quickly they could reach a finished state. Roots stated his desire to finish the first playable chapter of the game, based on the narrative he wrote, to be complete by the end of summer 2004. This audacious goal never had any chance of being reached, as the team would soon find out.
Later in 2004, the team was a little more experienced and humbled, but still lacking a solid direction. Realizing the technical limits of what could be accomplished using only the SDL library, the team collectively decided to employ the OpenGL and OpenAL libraries for advanced graphics and audio, respectively. What Roots and the programming team didn't realize was that they had, unknowingly, opted to build their own game engine from scratch rather than use an existing one to expedite the game's development. In fact, they were so ignorant that many had yet to realize what a game engine actually was. As a result of these poor and inexperienced decisions, development on the game itself slowed considerably, as efforts were focused on building the engine to support it. Most of the team were also students at the time, and toward the end of 2004 found much less time to spend on the project.
Development on game content such as art and music continued despite the technical challenges. Rain and Loodwig joined and created some phenemonal musical scores, shaping the game's theme and musical style. Artwork was a little harder to produce. Our art style was continuing to be tweaked and improved throughout the project's first two years. In February 2005 with enough code and artwork in place, the project released its first set of screenshots to the world. While not overly impressive, it did give the team a needed motivational boost to continue to get the game to a playable state. Later in 2005, roos (not to be confused with Roots) joined the team to fill the much needed role of a graphics programmer, and proceeded to quickly build a very capable and feature-full graphics engine and GUI system for the game to use. The team had also come to the realization that they needed a tool to build maps, and gorzuate got to work building a custom map editor in QT4, still used and packaged with the game today. That year we also came to the realization that the game could not be built (well) entirely in C++, and so the Lua scripting language was chosen to complement it. The rest of 2005 focused on development of the engine, tools, and technology we needed to build a working prototype.
One of the curious artifacts from Allacrost's first era was that the development, while open source, was closed for discussion. The team primarily used private forums and a private IRC channel to work on the game, avoiding the public eye. Joining the team also required you to submit an application over e-mail, which would be reviewed by the team with the decision to either accept or reject them from the team. In hindsight, these were both mistakes that did not align with the spirit of Allacrost and years later were abolished for a more open development model. Despite these barriers, the team had little trouble recruiting interested people in applying for the team in this era, and the size of the team became quite large in 2005 (over a dozen individuals). The team also had a more formal development process, such as quarterly real-time meetings on the IRC channel to discuss issues and design topics. These artifacts too were discarded years later as they became less useful as the team evolved.
Second Era (2006-2008)
By 2006, the team had acquired the experience and understanding of game development that were absent at the beginning. The team's growing pains had diminished greatly, but not completely. Developers were given a little too much freedom in certain areas that proved detrimental to the game's progress. For example, the audio engine was written and re-written several times as the team kept switching back and forth between using SDL_Mixer and OpenAL as our audio library, and likewise had to switch out the image libraries used in the graphics engine because it only worked well on some systems. When presented with a list of existing binding libraries to use to tie together C++ and Lua code, the developer in charge was allowed to write their own binding code from scratch (which was never completed, and eventually discarded). Another large issue came from the graphics engine, which had been built so quickly and so many advanced features added, that the fundamental components of the engine were sloppy and full of bugs. A major overhaul that took several months had to take place to get the engine to a state where it was stable and predictable in its behavior.
While the technical aspect of the project was struggling, the artistic aspect thrived. By the mid year we had created art and music of outstanding quality, far exceeding what was originally expected for this project. The artistic and musical styles of the game were becoming well-defined and we found that we had the potential to exceed the graphical capabilities of the classic RPGs that inspired this project due to higher screen resolution. Despite having more musical scores to select from than we could use, art assets were perpetually in short supply and it was more difficult to recruit artists to the team than any other role. Another fault came from some of the artwork requirements, where it was desired that we had large character battle sprites that were completely divorced from their corresponding map sprites. This created a huge burden on our artists, who struggled to animate such large sprites. After some deliberation and looking for solutions, we ultimately abandoned these large sprites in favor of using the existing map sprites and increasing their display dimensions in battle. But this change wouldn't take place until near the end of this era.
In October 2006, the team reached a milestone that most game projects never do and published our first release. With the notion that we could produce a full, complete game chapter quickly blown out of the picture months ago, we re-focused our sites and determined it would be best to create a series of "mini-RPGs" to give us more experience and allow the community to play the game and see what they liked and didn't like. Demo 0.1.0 consisted of a single cave map, tile-based movement, and random encounters. While short and basic, it demonstrated all the major functions of an RPG, from inventory management, experience leveling, battles, and dialogues with NPC characters. It was a solid step forward and renewed confidence and excitement on the team.
After getting player feedback from the release, several design changes were made in 2007. Tile-based movement was tossed out in favor of free-based movement. Random encounters were discarded in favor of having enemies visible on the map, and allowing the player to run away from encounters before they started. While certainly improvements over the original design, it did take much effort to make these changes and set the development team back a few steps. But not by much, as we release demo 0.2.0 in June of 2007.
This next demo included some of the core features in Allacrost's design. This demo had two separate maps, a town and a dungeon, and implemented map contexts to allow the player to instantly switch between the interior and exterior of buildings. A simple shopping interface was also supplied to allow the player to purchase equipment. We iterated on this game demo twice, releasing 0.2.1 in December of the same year, and 0.2.2 in October 2008. 0.2.1 introduced many core gameplay features such as treasures and saved game files, while 0.2.2 focused mostly on making the map editor a part of the release package for users to play with. By the end of 2008 and our second era, there was plenty of momentum heading toward our next goal.
Third Era (2008-2010)
With several successful "tech" demos released, the team now sought to produce one final demo that played more like a true role-playing game with defined characters and an engaging plot. This would be the final demo that the team planned on releasing before finally beginning on the full game. However, by the start of the third era many of those on the team had been working on Allacrost for 1-4 years already. The project leader, Roots, was particularly exhausted from spending so much time and energy on Allacrost. Despite the recent successes, several people on the team were feeling burned out.
The team behind the project had never really taken a break in all those years until around the start of the third era. Several members decided to take a step back, and a few of them would never return. Others took off months at a time to pursue other interests or to simply have more free time again, including the project leader, Roots. He had been particularly exhausted by working on Allacrost as his duties had also become increasingly unmanageable, having to balance programming, artwork, team management, publicity, recruitment, and writing. The team attempted to allow Roots to delegate some of his duties, and Roots tried to appoint an interim lead during his absence. Despite this, there was no clear leader when Roots left and the team began to lose motivation and fall apart. Progress on the game came to nearly a dead halt during this absence. Roots and other members did return after several months away and got back to work with a fresh perspective and attitude. But there would be further periods of inactivity during this era where progress was halted, and restarting the project became harder to do each time.
One of the major challenges for the final demo was figuring out how to make maps and battles more customizable by scripting, as maps in past releases had been mostly static. The team also had to deal with a severe lack of art assets and artists to complete them, requiring some of the programmers to step in and create artwork where it was needed. Eventually, it all came together and the team released demo 1.0.0 in August 2010. Two minor releases followed it over the next few weeks, addressing balancing issues and adding translated text, a new feature that was added to the 1.0.0 demo release. There was a great sense of accomplishment, and relief, on the team as this hurdle was finally cleared. The team took a much needed and deserved break after clearing this hurdle
With the last demo behind them, the team was now focused on the goal they had since the very beginning of the project: to release a polished chapter of the main story. By this time, the team had changed its approach to development drastically. The closed development model was discarded. Where design discussions and other happenings were kept hidden from the public eye in the past, they were now available for anyone to read and comment on. The team also more tightly embraced using third party content under free licenses to fill missing art and sounds, as they realized that producing all the content for the game directly was a very tough requirement.
But little did the team realize, they were headed straight into the most troublesome period for the project in its history.
Fourth Era (2011-2014)
The fourth era was nearly the death of the Allacrost project. The team size dwindled, its leader was often missing for months at a time, and no one took up the mantle to lead the project in his absence. The project would sit idle for several months, then some on the team would return and work for a while before disappearing once again. This cycle continued for years, and the project was said to be effectively dead.
In 2011, a new team member by the name of Bertram was very enthusiastic about the project, but disappointed and frustrated by the lack of commitment by Roots and the rest of the team. In the spirit of open source, he forked the Allacrost code into a new project, which he named Valyria Tear. VT was another JRPG that initially looked like and played like Allacrost. The VT project would grow and diverge over the years, and attracted its own team to help complete the release. Some of the other Allacrost team members also worked on VT, as the project had a lot of promising momentum going for it that Allacrost was completely devoid of.
From 2011-2014, Allacrost was in an active state of development for probably less than a year. The great number of technical challenges and frustrations were daunting, particularly for map design, and were one of the reasons that Allacrost failed to gain momentum. In September 2014, Valyria Tear released their first completed chapter to the public. Roots heard the news and played through it, and was inspired by the achievements that the VT project had made. His motivation returned in full-force, and shortly thereafter began working on Allacrost again.
He was, however, working alone. The project having been dead for so long, it felt like it would be an exercise in futility to try and bring the own team back just because one team member had their motivation once again. So he worked solo for several months, starting with a major overhaul of the game's map editor based on the capabilities he witnessed with VT's editor. The purpose being that this would destroy one of the major barriers to the more complex maps required in the full game.
Fifth Era (2014-Present)
The fifth era started with Roots continuing to work alone, finishing the new map editor at the beginning of 2015. The editor was then used to build the largest and most complex map in the history of the project, and progress on the game was well underway again. However, a project of Allacrost's magnitude is certainly not a one-person job, and eventually Roots looked to rebuild a team. By that summer, Allacrost had a full and active team once again for the first time in years. Several updates and improvements were made to the design. And the designers introduced a new primary feature for the game, known as battle fatigue.
In September 2015 after completing some of the maps, a development release was published to showcase the latest features. This was the first release that the team had produced in over four years. It was truly a moment where the project had been renewed and hope in its success restored.
Sadly however, Allacrost returned to an inactive status only a few weeks later. Roots had burned himself out working furiously for the past year and needed a break, and the rest of the team fell apart in his absence. It was a sad but familiar sight for the project, where it continued to go through cycles of great progress and utter inactivity. Roots bounced back in June of 2016 and picked up where he left off, and the project is once again active today.