About

Hi, I'm Beanut.

I'm a middle-aged millennial software developer by day, and a dedicated gamer by night.

I've been playing games almost as long as I've been alive. As the youngest in my family by a lot, I was born a small little Beanut with an NES controller in my hands. I grew up with Mario Bros. and Duck Hunt in my blood, and some of my fondest memories were picking up an SNES with Super Mario World and Mario Paint for Christmas, getting my PlayStation with Final Fantasy 7, Bushido Blade, and PaRappa the Rapper, and getting my N64 the year Ocarina of Time was released. I would visit Blockbuster every other weekend to try out some new game, just browsing the selection until my mom grew impatient and urged me to get something so we could leave... usually with something pretty bad, honestly. The last console she ever bought me was a PS2, when I got Gran Turismo 3 and Final Fantasy X without a memory card. I managed to get all the way to Seymour Flux on Mt. Gagazet without a single save before I got my first game over.

Eventually, I was old enough to have a job and make some real money.

I wasn't an Xbox or a GameCube kid, but the lure of the Xbox 360 drew me in. It was the first seventh-generation console, and I was beyond giddy with excitement to pick up the first console I could get for myself. I put so many hours into that, and it's by far my largest collection of physical games. My attempts to get a PS3 at launch were fruitless, as people far more dedicated than I was had camped out for what little stock existed. Eventually, a friend who worked at Best Buy secretly hid one for me when they got some new stock in. Luckily, from my Xbox 360 and the release of Dead Rising, I also had a banging 42" 720p DLP (thanks to Dead Rising being unreadable on a standard CRT) and a surround sound system, so my first experiences with the PS3 were much better than my first with the 360. I took the next day off work just to play Resistance: Fall of Man all the way through. It was a glorious experience. I camped out for the Wii and snagged that at a midnight launch with Ocarina of Time, Red Steel, and Excite Truck, and, of course, one of the greatest pack-in games: Wii Sports.

Eventually, I enrolled in a computer science program at my local university, and gaming had to be put on the back burner for quite a while.

Mounds of student debt and severe burnout after graduating and earning my BSc. left me in a fairly bad spot in life for a very long time, and eventually depression destroyed most of my joys in life, from music to movies and pop culture to video games.

It's taken me over a decade to at least partially recover. I have my ups and I have my downs, like most people, but I am recovering.

But I never lost my love for video games.

I've dabbled with romhacks and translations in my own time, but never really had the motivation to get all that far and release anything... depression does suck.

Methodology

With the advent of AI, however, and the progress it's made, I can utilize my knowledge and skills in ways many vibe coders just can't. I understand (dis)assembly and the romhacking process. I am used to the tools and debugging methods. I know how to describe, in technical terms, what I want and direct AI on how I want to proceed in a very systematic way.

My goals with my translations are to create something good, not just shotgun out a translation in a weekend or let bots run roughshod all over it with little to no supervision.

I do not use AI to just say "Translate this kthnx" and see what it fumbles around with. I provide it with everything it needs, and let it explore and analyze systems, functions, memory, and data as much as we can before we even begin anything. We create a proper project plan, and we work incrementally towards each goal in order to make the best presentation I can.

I usually take a hands-on approach to graphics editing, though some things have proven to essentially require AI, lest I destroy the original art in my feeble attempts to remove text.

For text editing, I take the approach of giving the AI as much context and official material as absolutely possible outside of the game itself before we even begin. Whether it's the instruction manual, strategy guides (which are REALLY hard to come by sometimes), official websites, articles, or staff interviews about the game, I feed it everything I can.

It proceeds through the game text, and I ask it to, in conjunction with the previously analyzed materials, create a translation bible. The guide contains all the facts and spellings we adhere to. It establishes character profiles for every relevant character, describing their personality and the standards their prose and word choice would generally follow.

We analyze the dialogue and ensure we translate it in the context of the conversation, never in isolation, and in the voice of the character, while trying to adhere to the spirit, substance, and meaning of the original Japanese text and ensuring proper idiomatic English.

Philosophy

AI is NOT perfect. I will be the first to admit that. I do many, many, many editing passes over my translations with AI, as it does tend to mischaracterize or misinterpret what the Japanese is trying to communicate in many instances. When possible, if there aren't 400,000 lines of text to go through, I do read and edit what I can to ensure it at least makes sense in English.

While AI is not perfect, frankly very few translations are. Translation is an art. It takes a tremendous amount of skill and talent to do properly, to understand and dissect the nuance, and to choose the best words possible for a given piece of text. With that, even human translations can fall short. Just see most games from the 80s and 90s released out west, and even a number of fan translations to this day entirely done by humans.

AI is a powerful tool and can do things that would otherwise be unthinkable undertakings for most romhackers. An AI translator should absolutely be utilizing it to its fullest extent to ensure the translation can be the best it can while avoiding as many of the compromises a traditional team may have to deal with as possible, such as modifying the UI to fit the English text better, increasing string lengths (which can be a very difficult or insurmountable task sometimes), or adding subtitles for voice-overs, FMVs, and gameplay. These are things that can take an AI agent literally minutes to accomplish, but which would otherwise take a team weeks, months, or even years of poring through assembly.

I choose my projects from games I'd play and games I think are cool. I also choose games that are too niche and not worth anyone's time to properly translate, have frankly far too much text for any translation group to actually tackle, are far too technically challenging for a romhacking group that refuses to use AI tools to tackle or do justice to, or any combination of the above.

My stance on AI translations and how they affect other groups tackling these projects themselves is that if they are deterred from a project because of an AI translation, they either think the AI translation is good enough and not worth replacing, or they weren't willing to commit to it anyway to do it the justice they think it deserves. AI translations can coexist with human translations, and if they're a deterrent, then no one was serious about doing it anyway. So I appreciate the effort I hope the person behind the AI put into the version you got... and I hope you can appreciate that some people are not just trying to turn out AI slop and actually do put effort into these.