Localizing the four games that make up Arc the Lad Collection presented a unique set of challenges from that of "normal" game development. The first task to accomplish is to get all of the data files and source code to build so that they match their equivalent files of the Japanese production discs. This was, as it tends to be with most localization projects, the most difficult part of the project. All four games were plagued by various degrees of multiple file versions, missing files, scarce documentation, memory problems, and bugs found in the original Japanese games.
Having different versions of the same file scattered in various directories made it difficult to determine which should be used. There would often be files in one directory that had a later timestamp from files in another directory, yet the older files would be the ones that match what was on the original production disc. Then we would see that the older directory would be missing files, while the new directory had them all, plus extras. And on top of that, we would end up with even another version of these files when we tried to build them ourselves. It was, at times, quite frustrating.
Files that were missing completely was another challenge to overcome. Requests for these files would be sent off as soon as we realized that they were needed, but in many cases the files were lost forever, or difficult to track down. In the end, it was usually quicker just to go to the original Japanese production disc and rip what we needed off of there. Although we had to use this technique on each of the four games, we relied on it most heavily for building Arc the Lad II. Obtaining missing files this way is one of the main reasons we had to get our build to match exactly what was on the game disc.
The fact that there was little documentation, and it was all in Japanese, didn't make things much easier. We would send out any documentation files we found to be translated, but none were that helpful. There was never any file that gave an overview of the build process or project structure. For the comments found in the source code, we would write a script to extract them, translate them using a program called Atlas, and then write another script to put them back. The translations were nowhere near as good as they would have been with a human translation, but they got the job done.
Once each game was up and running, altering them to run in English was a simple process. The next major hurdle was reducing the amount of memory the game currently required, so that we could write full-length sentences and add new features. Many techniques were used to reduce memory, including text compression, source code optimization, and reducing statically defined buffer sizes. After getting back as much as memory as we could, we were able to add a lot of nice changes, such as analog support, DualShock support, and 24-bit (as opposed to 16-bit) full-motion video playback.
By this point in each of the games, there weren't too many problems. Bugs would be introduced by changes that were made, but they were taken care of fairly easily. We did occasionally encounter bugs that also happened on the original Japanese discs, however, and those took a lot of time and effort to track down and correct. The worst bug we had was a lockup that only occurred on a certain model of PlayStation. That took several weeks and a lot of headaches before we got it ironed out, since it wouldn't happen on the development system where we could take a snapshot of what was happening inside the machine. We had to burn a CD every time we wanted to test another theory as to how to fix it.
One situation that was unique to Arc the Lad III involved the Sony PocketStation, which is only supported in Japan. Since Arc the Lad III originally made use of the PocketStation, we also wanted to retain the support for those who owned them, although this became more involved than we expected. To successfully build the newly translated files, we had to borrow the complete PocketStation development system from Japan, then quickly learn how to make it work.
I had worked on a few games prior to Arc the Lad Collection, but this was the first project where I got to start from the beginning of the localization process. It has been an amazing experience, and I have learned so much. I was even able to apply some of that knowledge to make an original creation - the DVD-esque style and interface found on The Making of Arc Collection disc. I hope you like it.