[In this Gamasutra-exclusive postmortem, the Atlus team behind Shin Megami Tensei: Persona 4 -- most of which continued on from Persona 3 -- discusses its experience creating the acclaimed RPG, including a development overview as well as specific "What Went Right" and "What Went Wrong" highlights.]
Our goal with Persona 4 was to create a title for young adults in a modern day school setting, but with appeal for a wide audience. For players who became fans of the series with the previous game, Persona 3, we retained gameplay basics that had proved successful, while adding an element of suspense. We hoped that a murder mystery plot in which a group of high school students pursued the culprit would connect with the players.
Many of the core members of the Persona 4 project were from the internal development team led by the director, Katsura Hashino. Their previous titles included Shin Megami Tensei: Nocturne, the Digital Devil Saga series, and Persona 3. The art director, Shigenori Soejima, joined the team during Persona 3. However, the rest of the staff, including composer Shoji Meguro, had been working together for about ten years.
Except for the animation, the game was done entirely in-house by the team. Our primary asset creation tools included 3ds Max 8, Adobe Photoshop, PaintTool SAI, and Adobe After Effects 6.5J, while design and effects were handled with our own internal tools.
(As a side note, the surgical action game Trauma Center: Under the Knife was also designed and produced by Katsura Hashino. A few members of that game's staff were assigned to form a new team, and since then, they have been handling the development of the Trauma Center titles.)
What Went Right
1. Critical acclaim: Persona 4 received high praise from both the media and the consumers. Our primary consideration when deciding what to change and what aspects of development to examine was the commitment to making this game better than the last. The players were generally satisfied with the game, and we feel confident as we move on to our next project.
2. Budget: The development cost was set at about the same level as that of Persona 3. To satisfy the consumers, we spent the majority of our resources on increasing the volume of features that the players liked, improving the game systems, and working on the story and characters (the two key components of an RPG). In the end, we were able to keep the number of omitted features to a minimum, while incorporating many changes and additions to reflect the opinions gathered toward the end of development.
3. Player feedback: We were able to carefully select many of the new features by internally evaluating the previous title and examining player feedback. By doing so, we could concentrate on making adjustments to the areas that were most crucial to the game's quality.
4. Team stability: We didn't make many changes to the roles of the development team members who continued on from the previous title. This gave each person a clear understanding of what issues needed to be addressed, allowing us to operate smoothly during development. It also allowed the various staff members to communicate effectively with the director.
5. New hires: Some of the new staff members who joined the Persona 4 development team were fans of Persona 3. They did a great job gathering feedback on the previous title and evaluating the content of the new game.
What Went Wrong
1. Storytelling challenges: It took a tremendous amount of time and effort to finish the intense, suspenseful story with as many twists as it has, as well as integrating the game's theme of "how one accepts information from the media." While it was fun creating the mystery novel-like scenario, we had no previous experience in working on such a plot, so we were making adjustments to the storyline until the very end.
Also, one of the villain characters changed in the middle of development; since the character design was done before the story change, the design did not reflect the fact that he was a villain.
2. Real-time weather design: Unlike in the previous title, Persona 4's time limit for each dungeon was affected by the in-game weather. We did this with the belief that such a system would create the feeling of urgency, since the player didn't know when damage-causing fog would appear.
However, when we implemented it in the game, players were inclined to make dungeon investigation their first priority. Their mentality was, "If I don't know when the fog appears, I should finish the dungeon as soon as possible." As a result, dungeon crawling and working on the inter-character Social Links, which are equally important, became completely separate and imbalanced.
We tried to compensate by adjusting the weather, in-game messages and story progression, but that created an unexpected workload. A huge amount of data could not be finalized until the weather was set, but the weather kept changing due to our design adjustments.
3. Urban misconceptions: When we decided that the story took place in a rural town, we found out that each staff member had their own image of a rural town that was completely different from the others'. So we immediately held a meeting to discuss what the most typical rural town was like, and the entire team went out to various places for location hunting. This was the first time the development team conducted such a large-scale location hunt.
4. The QA time sink: We end up having this issue every development cycle, but doing QA for an RPG takes a significant amount of time. For example, it takes more than a month for the director to go through the entire game once, checking the content and giving feedback to the team.
We kept playing the game over and over again, as many times as possible, until the game went gold. The more time we spent on debugging, the harder it became for us to put ourselves in the mindset of how gamers would feel when playing the game for the first time. In the end, we all wish for our next project to be an action game.
5. Horizontal feedback process: About two months before finalizing the code, we had the entire team post comments and criticisms about the game on our internal development website. For Persona 4, the new staff members (most of whom were Persona 3 fans) made the biggest contributions to this process, and we ended up with close to 2,000 posts -- anything from fundamental problems to personal tastes.
We addressed more than 1,500 of those concerns in some way or another, but the staff kept making comments like "This should be changed like this," or "This part is no good." The design director who decided how to fix the issues and the staff members who implemented the fixes were on the verge of a nervous breakdown, begging, "Please... No more..."
Nevertheless, the game's quality increased and the consumer satisfaction was high, partially as a result of that internal feedback process, so we're glad that we did it -- but the thought of having to do it again for another project gives us chills.