Developer Blog: The 4x4 Design of Tower Song

(This post is a text version of the above video.)

I have a problem: I really like coming up with RPG character designs.

Tower Song features four heroes for players to choose from as their point of view character. Each character comes with their own unique class design, dialogue, and quest rewards.

Tali is a spellcaster, so while she uses mana points, she doesn’t really have a spell list - instead she pulls Tarot cards, each of which grants her a different spell. Of course, she also fights with Jiji, who is himself a party member. In the field, Jiji can sneak into tiny spaces, finding treasure or unlocking doors from the other side.

Neat is the ultimate tempo fighter; each skill he has adjusts his Combo Stage, meanwhile; depending on the stage, he can access different moves. So he has an entirely different moveset each turn, depending on what you are doing. In the field, he can walk Song-addled victims through meditation techniques, allowing the party to pass into areas without violence.

Porphry is fury, a mix between a berserker and a fighter. Her skills generate Trance, which, like Neat, can change what skills she has access to. But she can also lower her Trance; why do this? Well once her Trance gets high enough, she goes into a berserker rage, and players lose control of her. In the field, she can break down weakened walls or obstacles, opening up new areas to explore.

Ezoan is a fun hybrid spellcaster and fighter; he casts spells like others, but he also drives an exosuit; the exosuit is powered by gravity, which he generates by casting spells. In the field, he can hack exosuits, moving them or just exposing them to the player for loot.

All of this is driven by a compulsion for me to design RPG characters; as many as I have ideas for, usually, and for whatever reason, I am never short on those.

You might think this is a boon, and it is certainly a more helpful problem than having no ideas at all. Let me give you a little context though.

The original RPG Maker 2000 game had 8 members in the main cast. It was also quite a bit longer than Tower Song, with a main quest running around 20 hours and a bunch of side quest content bringing it to about 30 or 40 hours to 100%. If 8 sounds maybe reasonable, however, you should also know I included an additional 9 hirable mercenaries as well. They didn’t have any speaking lines and some of them couldn’t hit the level cap, but, you know, that’s just a lot of characters. Too many, some might say. (Meaning myself.)

Even the original Tower Song prototype had this problem, featuring 7 characters with a main quest clocking in only around 2 hours.

I knew Tower Song would need to rein it in and be more focused. The harder option - increasing the game’s length - was out of the question. Not only did I not want to ask players to commit to another 40-hour long RPG, I also wasn’t going to have the art budget to create a game of that scope.

The solution seemed genius: narrow down the prototype’s cast to 3 main characters and allow the player to choose their own main character from the remaining four. We could then commit our resources to a single main quest line (and the handful of biomes that went with it), while still being able to play around with all of the class designs we wanted to.

One thing I did not want, however, was a silent protagonist. Nor should their lines simply be inserted into longer cut-scenes as “fill in the blank” asides. No, I felt strongly that the player’s character (let’s call them the “hero”) should have their own goals, motivations, and personalities - and be vocal about it.

All of this, of course, created a single game of reasonable scope that was then multiplied into four games, with a perhaps less reasonable scope! Everything we did, had to be done four times.

Bug testing: many bugs only appeared for one or two heroes, necessitating campaign playthroughs with all four characters in order to certify a build.

Combat balance: getting encounters, particularly the “story bosses”, to feel just right in combat meant they had to feel right for each potential party. A trivial fight for one party composition could be near impossible with another.

Side quests: if we’re adding a side quest with a unique reward for one hero, at the very least, four rewards had to be designed - one for each. To make side content feel truly special, though, we also wanted areas and encounters unique to a hero - which then meant we had to design four times as many.

Storyline: While some cut-scenes do not really involve the hero and the dialogue is the same for all four, allowing each hero to bring their own personality to bear often meant cut-scenes were so driven by character that there was no option except to write the scene four different times.

Class design: This was the whole point, so you can believe we had four different designs.

So did it work out?

Well, yes, I think so!

The game certainly took longer to developer than I anticipated, but the result is something special. Although the outcome of the main story does not change, each hero in Tower Song is otherwise a unique experience. Whether you play it once, or four times, your choice is rewarded. You even learn about the world just a little differently depending on your hero and the perspective they bring; party builds and battles also play out differently, due to the totally different class designs.

So will we do it again? Not sure! It’s a lot of work. But I did learn a valuable lesson about scope, all the same. Maybe that’s progress for me. A sign of progress.

That’s good, right?

See you next time.

(All of this is a lie, of course, because we still found cause to add another two playable characters into the game’s final chapter. Once an addict, always an addict, I suppose.)