It took about 1050 days to go from the start to finish for the Narcissu 2 project. But just what were those days used for? I’m sure that many people out there are curious.

Graphs

In my day job, I work with numbers, listen to the stories they tell, then spread their stories to other people. Data plots are my life. Applied to this project’s records, here’s the end result:

A stacked area graph of line commits:

As well as a traditional line graph:

They plot lines added, usually means first translated, and some kinds of editing, and lines edited/changed. Data points were aggregated over 7-day periods. The dates on the chart mark the start of the week the data point represents.

The colored bands represent major points in the translation and each holds its own set of notes.

Data collection

First this whole article wouldn’t even have been possible if a few things hadn’t been in place.

1. Haeleth and I used version control from Day 1. Subversion in this case.
2. I personally made a point to make a commit every night when it was physically possible. Making commits every few weeks in chunks isn’t as useful for this sort of analysis.
3. It helped that we had put put small diary-like entries and conversation into the commit comments, made for a very good reminder of what was going on at the time.

For anyone looking to work on their own translation project, I highly recommend some kind of version control system. It just makes life so much easier, and lets you look back when it’s all done.

The first drafts

One thing that everyone will probably notice was that by early September 2007, both Narci 1 and 2 scripts have been “translated” in some sense of the word. It happened to be summer and I was in grad school, so I had plenty of time to devote to the project. Narci1 also took less than a month to fly through, generating that huge red spike.

“Translated” in my case isn’t really accurate in my case. At the least, it’s more like laying the base sketch, the pencil lines where I’ll be layering ink and paints. That’s why revision #1 comes soon after, chopping away the errors from fatigue and smoothing over rough patches before letting Haeleth even look at them.

Cross-checking

Around when rev2 finishes, I start my first job, and wait for Haeleth to catch up. Around October, we start the process of cross-checking each other’s scripts. This mostly involved me reading Haeleth’s scripts and writing comments when I thought something was wrong, or disagreed with how he rendered a line.

What was really interesting about this process was that we often wound up editing our own scripts during this process. Usually one of us would discover that our scripts don’t agree on a point, and after some work, we’d figure out the other was correct, and we’d go and fix our own version. Sometimes we’d go back and forth and wind up editing both sides.

This is the first time that I worked with another skilled translator, and it’s something that I’d like to do again. It’s very different than working with someone who’s starting out, or worse, has no idea what they’re doing. I highly recommend people try it out. You’ll learn a ton about your own style as well as widen your views on translation.

Almost as an aside here, between working a full day job, working on this project, I had also been finishing up my Master’s thesis at the time. There certainly weren’t enough hours in the day.

Engine debugging and cleaning up

Once cross-checking ended, Haeleth started going off to work on the engine, adding features and other things that I’ll never fully be aware of. Meanwhile, lacking other things to do, I went and did more revisions. In the end it culminated in a huge spike of edits around March.

Softsubs and graphics

About a year into the project, the scripts had settled into something serviceable, or at least, not too embarassing to show another human. Haeleth had been drifting in and out of contact due to personal life things for a while now, so I’d amuse myself by working on all the graphics, and many other things that weren’t counted in this data set.

Incidentally, the softsub timing and translation of the opening movie happened in this time. It was the first time that I tried to make a song translation somewhat singable. Having never wrote lyrics before that was another whole big project. It’s an extremely difficult thing.

Around June 2008, there’s some rumbling about possibly releasing. Or at least, rushing to get scripts and all finalized soon. That led to a number of revision bursts but ultimately it didn’t result in anything.

Trip to Japan

August 2008 hits and I finally go on vacation, to Summer Comike C74. Lots of great things happened then, aside from almost getting sunstroke and severely dehydrated in the Kyoto summertime. One of the highlights of the trip was a meeting and dinner with Tomo Kataoka in Akiba.

We discussed a few things like the future of the Narci projects, backroom production stories, and translation projects in general. Kataoka mentioned that he was very thankful to all the people translating his work into other languages. He had once asked a translator friend at a major game company once how much it would take to translate Narci 2, and the response was in the tens of thousands of dollars.

Considering how much time it’s taken me to translate the game, if it were compressed into a full-time job, it would probably have taken a few months of solid work, and salaries alone would have accounted for that amount. I’m happy to have provided that resource to him, but also realized that the scales of economy that we’re dealing with are huge.

Just as our creating this translation patch saved Kataoka tens of thousands of dollars, something that fan translators could do in the future could also irresponsibly destroy tens of thousands of dollars for someone else in the world.

Economy go ‘splody

Around October 2008, the economy had taken a horrible turn, and the mood around the financial district had turned pitch black. The ripples from that also cost me my job, and so I was left with a period of unemployment. Similarly, Haeleth had been sucked away by life.

The Great Void

This is the ugliest part of the graph. almost 10 months of no progress. In the middle of February 2009, I managed to find a new job. Great for things like food and shelter. However, the job was a tough one, and had me working 10-12 hours consistently in an almost non-stop analytics role.

The end result was that I’d come home exhausted and over-stressed and just couldn’t work up the ability to work on any scripts. There was one period towards the end where I struggled to finalize the scripts in hopes of a release, but that got crushed too.

Revival and Release

In December 2009, I finally managed to find a new job, no small feat in a city with over 10% unemployment. With no transition period, I just dove into the next job the day after leaving the old one. A bit of resting later, I was in shape to revive the project.

By now, Haeleth had totally stopped responding to email, so I had no choice to pull everything together myself. I reached out to all the people in #denpa, and was thankful I had gathered a bunch of great people there over the past year. There had a group of testers, people who could translate to look over scripts, people who could led me servers, people to build installers for me, and Mion was there to fix the many bugs still left in Ponscripter.

With their help, I got beta builds out. Threw myself into making script edits, then aimed for what I hoped would be a late-February release.

Ultimately the QA process took almost a whole month. There were just too many bugs, and people took a while to get to me. But come mid-March everything was back in my hands. With a final insane dash at script revisions in the last week, everything was pulled together and we made out release on 3/31/2010. Woo!

End

Hopefully, this timeline throws some color onto what happened “behind the scenes” during the project. Yes I made 8 full top-to-bottom revisions of the Narcissu 2 script, and 5 of Narcissu 1. Did I need so many? Of course not. I probably could have done with 3-4, but I was also waiting for someone else (who never really did reappear).

There’s one nice thing about the extra long editing process though. I had plenty of time to explore what I wanted to do as a translator. As I went through the edits, I’d be playing with renderings and styles, and I’m sure I emerged from the process different from when I started. Nothing but time spent working could help this growing process, so for that I’m thankful.

Not that it matters to you people waiting on the final product. =)

Random trivia

• Total svn commits: 556
• Exact # of Days from start to release: 1051
• Final number of English lines: 6638
• Total lines added to scripts: 6871
• Total lines edited after adding: 5146