Postmortem: Myst IV - Revelation

Introduction

When I was asked for a postmortem on Myst IV: Revelation, I was faced with a dilemma (not a first for that project), the fact was that with three years in the making, there weren't very many people who had been on Myst IV from conception to finish, and some key players had come, gone and left us with new challenges and many questions. We had changed the creative director, the producer, and the artistic director. Also, opinions on what went well or wrong on the project differed greatly. Therefore I decided to give key players representing different aspects of the game a chance to express themselves on the making of Myst IV: Revelation.

What Went Right

Small multidisciplinary team. Less than a year before the end of the project, things were not going well on the Myst IV: Revelation team: no single zone was in a finished state, communication was difficult between team members and puzzles were taking too long to prototype. We looked at the quantity of work remaining and started brainstorming on how to close this project before the end of September. Instead of having a huge production team divided by specialty, we choose to divide into smaller teams (swat teams) dedicated to specific zones. Each team would have at least one member of each specialty. They were also located together in order to help communication between them. Rapidly, finished and very polished zones started getting out and motivation improved. The number of bugs at the end of the production was very low due to the effort of each swat team to completely finish and debug a zone before switching to another. - Nicolas Beaudette, Lead Programmer

Sound studio. As usual with Ubisoft, the sound got the same attention as the other aspects of the game itself. One of the main goals of the audio pipeline was to give the musician the ability to compose and implement the musical score directly into the game engine from the studio. In order to achieve this, we created a standalone tool outside the game engine because we didn't want to send close to 10 Gig every time something changed in the game. That way the composer was able to modify the musical sequence until the end of the project. We used a dynamic music sequencer similar to the one previously used for Myst III: Exile. It is a way to playback an audio segment in parallel instead of a linear track of a couple of minutes. We stream up to eight different layers simultaneously to recreate a song with more musical variations in it. In a game like Myst with lots of exploration this kind of music playback engine is really a good way to go because the player will not notice any real musical redundancy in their game experience. In general it was a great success but if we work again with a musician dependant on a proprietary tool is telecommuting from outside of the office we will be more aware that it is not easy to support a tool and keep his game version up-to-date 1000 miles away.
- Mathieu Jeanson, Lead Sound Engineer

Experienced team members. Since Myst IV: Revelation was the first prerendered game produced at Ubisoft, we had to recruit experienced resources from outside the studio. Despite their lack of experience in video games, they brought us a lot of experience in area like video editing, compositing, live action, prerendered graphics, animation, 3D and tools programming. - Nicolas Beaudette, Lead Programmer

Game quality. When the zones started being completed, we realized that we could not have dreamt of a better quality for the game: Prerendered backgrounds where astounding, back-story elements and puzzles were fully integrated into the game and the live action high definition video was well mixed with the background. We were also very pleased by the quality of the music produced by Jack Wall . Special effects in both 3D real time and prerendered were working fine and we reached our goal of creating living worlds. Frame rate was high even on slower video cards and memory usage was low. - Nicolas Beaudette, Lead Programmer

Prototyping the puzzle. Because this sort of game was new for all of us we first thought it was possible to start the high resolution model from the beginning of the project. After the prototype we saw that it was too risky and not flexible enough to deal with high resolution material, so we decided to make mock ups of the puzzle in 3D in order to spot technical, design and logical issues before starting everything in production. - Gilles Monteil, Animation Director

Real-time special effects. As a specialist in real-time special effects, I joined the Myst IV team a year and a half into production and was located right in the heart of the programming department; this was a key to the success of my work. Shortly after joining the team I started to realize that a lot of the tools, which I needed to get the job done, were either nonexistent or not implemented yet. Also, the engine didn't have a fully interactive interface. However, a lot of the tools were developed soon after I started, and were built with my needs in mind. For a long while, the Real-Time-Special-effects-art-department was mainly, well, me. Soon however, other artists joined me. With the pressure down, and armed with all the tools we needed, from particle emitters to pixel modifiers we were free to be creative. It was the best environment I have yet to work under. - Ned Mansour, Real-Time-Special-Effects Artist

Hybrid version. The first Myst was a Mac game and the fan base community was really dedicated to both the PC and Mac version. For this reason we decided to keep the same hybrid format that was used in Myst III. The OSX port was planned early in the project, and therefore the engineers developed the win32 engine with cross-platform in mind, which helped a lot for Macintosh version. Overall, the OSX port of Myst IV: Revelation went very well and took about six months to achieve. Once we made in-house tools to optimize the work-flow, the port went very quickly, and after only one month we were able to present an E3 demo. The hardest part was to have a good frame rate and make the effects work fine on the minimum requirement configuration. Through all of this, Apple gave us a tremendous amount of support. We used CodeWarrior as a compiler because we already used this compiler for other projects, but in retrospect, maybe we should have used the Xcode tools provided by Apple so as to be able to fully use their performance-oriented tools. - Eric Thiffeault, Mac Programmer

Creation of the "panelists" team. The tools for animation came late, so we very quickly saw that animators wouldn't be able to produce all the data and integrate every video of the game. We decided to hire "panelists" who were in charge of making the videos of the animations and following their integration with the AI programmers. Good call because at the end of the project there were more panelists than animators.
- Gilles Monteil, Animation Director

A.L.I.V.E. Technology. Since the beginning of the project, our main goal was to remove the static feeling of previous Myst games. We focused a lot of research efforts in order to reach our goal of "worlds that come to life." To achieve this level of quality, we put all of our effort into prototyping five nodes (a node is one spot of 360 degree view in a world). This small prototyping allowed us to upgrade the quality of the tools, the engine and the pipeline. - Nicolas Beaudette, Lead Programmer

QA. The most important thing during the entire test process for Myst IV: Revelation was that we had a very high level of communication with the team. Because we had to know exactly when each step of the integration for every little feature was done and implemented in the game in order to test them simultaneously at the right time. On every version we went though the complete tests - walkthrough tests, configurations tests, gameplay tests, script events tests - based on the storyline desired by the game designer. The test process was not easy and it was the first time that we put a game on dual-layered DVD format. Finally, the game was released at the time that we had originally planned and at the quality level that we wanted to achieve.
- Yanick Beaudet, Quality Control Lead

What Went Wrong

Big pre-production team. Myst IV: Revelation was the first pre-rendered adventure game ever produced by the Montreal Ubisoft studio and our lack of experience in pre-rendered games soon hit us very hard. To accelerate pre-production and train our employees for the production, we hired more than 50 new employees. The A.L.I.V.E. engine was barely working, tools did not initially exist, the story was not final, artistic direction was not well established, and no real level design was ready when all these people joined up the project. Extra pressure was put on game designers and programmers to feed the modeling team and a lot of the artists started to feel frustrated about not being busy enough, not receiving clear goals, and having embryonic tools. A bad mood was established between the game designers, programmers and modelers that only decreased slightly in the last few months of the project when the swat teams where created and everybody started to work together. Less than eight employees from the original pre-production team finished the project. With this high turnover rate we lost precious experience. We learned the hard way that in pre-production you must have a very small dedicated team and you should never start game production before the story, the design, artistic direction and technologies are ready and fully tested. A creative director and a production manager with strong power and vision at the beginning of the project would have helped clarify the focus of the game and the processes as well. - Nicolas Beaudette, Lead Programmer

Animations integration. Myst IV: Revelation is a game where the main effort is put on creating a beautiful landscape. We could only do final animation on final and approved sets. The lighting was done at the end but was prepared in order to light the sets but not the animations. This caused the animations to have to adapt themselves to the sets and not the other way around like in movies. - Gilles Monteil, Animation Director

Tracking changes. No unified system or format was established to track and communicate the changes to assets required between all members of the team. It really became a nightmare to follow the completion of critical tasks and to have an accurate up-to-date perspective of overall planning. A lot of work was trashed because of the inefficiency in tracking production assets. This was only resolved in the last ten months of production, by the introduction of a database tracking the assets and validations.
- Nicolas Beaudette, Lead Programmer

Lack of time to integrate the secondary animations. Integration of the secondary animations (created just to give life to the environments) was late and we had to close the game, so at the end of production we had to cut some of what was previously planned. However, it was impossible to cut into the story, or into the bulk of the gameplay, (except for some stand-alone puzzles) and cutting sets would have cause more trouble than just leaving them like they were. So it was animation (and especially panels: animation rendering) that were cut out, replaced most of the time by real-time effects in order not to lose our goal, which was to have Myst IV: Revelation “come to life.” - Gilles Monteil, Animation Director

Defining a new production pipeline. Myst IV: Revelation's A.L.I.V.E. engine uses 360-degree bubble, mixing a lot of videos with alpha and real-time 3D effects. Myst III was already using this kind of technology but we chose to expand the concept much deeper. From the engine point of view, it was not so difficult to integrate the technologies together but the tools and production pipeline were a lot harder. We spent a lot of time finding good tool combinations and procedures, whereas before the effect simply looked slapped on top of things. - Nicolas Beaudette, Lead Programmer

Dual layer DVD. Myst IV: Revelation was one of the first games to use the DVD9 format. It quickly appeared that a lot of the DVD readers were not able to read the dual layers copies despite assurance from manufacturing that we would experience no problems. Worst of all, some problematic readers didn't have any available firmware upgrades ... Customers were complaining about our game not working on their systems but we could do nothing except putting some pressure on DVD manufacturers to make the firmware available. We also bought a wider array of DVD drives to upgrade our chances of finding non-compatible drives in the future and improve customer support.
- Nicolas Beaudette, Lead Programmer

Conclusion

When I arrived on the project, there was a year left of production, which totaled three years all together. The first thing I did was to pass through the concept as it was. As a Myst fan, I was very impressed about the fact that the team had succeeded in keeping all that makes a true Myst game, while improving all that could be improved.

However, it was soon clear to me that we were not going to meet our deadline. Ideally, I would have liked to stop production for three months in order to complete the tools that were still missing and reorganize the pipeline and validation process but of course we had no such time with E3 coming fast. With the help of a very thorough audit by our technology group, the changes described above were implemented (Implementation of a database, redesign of the pipeline, implementation of task groups, etc.).

However, one aspect that has not been covered much in the sections above, even though I believe it to be crucial, was the team spirit between key team members. To address this, the following personnel modifications were implemented:

These are all personal-related issues, which might seem unimportant on the surface but deeply affected the decision and communication process of the team.

All the changes described in this post mortem helped but were not enough in themselves, and it has to be said that if Myst IV: Revelation came out in time and with zero critical bugs, it is due to the relentless work and ingenuity of most members of the team who, knowing they were working on a legendary brand, compensated for all the difficulties of the project.