How visual novel writing is different

Recently, I got into a message board discussion about writing, and the subject of visual novel writing (and how it differs from prose and other more traditional forms of writing) came up. This seems to be a fairly common inquiry from “traditional” writers peering into the world of visual novel development, so with that audience in mind, I thought I’d proffer a writer’s perspective on how visual novel writing is different from other forms of fiction writing.

The Basics

In simplistic terms, visual novel writing is a bit of a hybrid mix between screenwriting and prose writing. You can have characters that just deliver lines (like in a screenplay), so you don’t ever include things like dialog tags, but your dialog can be interspersed with bits of internal narration like you’d see in regular prose.

Another part of visual novel writing is the fact that certain details can be conveyed through the visuals. For example, suppose I have a character who is going to be represented on screen by this sprite:

Vaeril, from Winter Wolves’ upcoming game “Cursed Lands”

It’s not necessary for me to describe him in the text as having long, blonde hair, as that information is already conveyed to the reader through the sprite. Also, most VNs also have multiple expressions/poses for each character, which eliminates the need for a lot of beats like “she smiled” or “she glared at me.” And that tends to go well with the “no dialog tags” thing, because oftentimes I use expression-based beats in place of dialog tags.

Chihiro from Dangan Ronpa can be represented by a variety of sprites with varying moods

Likewise, certain elements of the environment are often included in backgrounds. Because of this, I tend to lean on non-visual senses when writing description. For example, if a character walks into a swamp, I don’t need to describe what the swamp looks like, but I will describe the noxious odor of the swamp, the sound of buzzing flies and mosquitoes, and the feeling of sticky humidity of the air. (I include this kind of sensory detail for regular prose writing, too, but in visual novels it becomes even more important since the visuals usually aren’t part of your textual description.)

The asset budget (and other production realities)

When writing a novel or short story, you have a pretty flexible “budget.” The number of locations and characters in your story is limited only by your ability to describe them effectively. If you decide, three drafts in, that you want to change the location of a fight from a dark, murky cave to a rooftop balcony overlooking some scenic vista, you can do that. If you want to add a brief scene where the characters take the time to walk by some picturesque location that they never return to for the remainder of the story, you can do that.

In visual novels, adding new locations isn’t free, nor is adding new characters. Readers will usually expect every character with spoken dialog to have a sprite, even “nameless NPC” characters who only show up to deliver a few lines. There are a few tricks that you can use to get around this, like having “generic” sprites that can be used more than once in the story (for example, a fantasy story might have a “generic guard” image that shows a man in uniform with his face covered by a helmet), but generally speaking, each character means one new sprite.

This can sometimes be hard for writers to adapt to, since for writers, the “cost” of adding a new character is connected to how much they participate in the script. If you have a “generic” character whose only role in the story is to give the main character directions, you can add that character mostly for “free:” the man with only one line of inconsequential dialog doesn’t need to have backstory or character development; you can just add him to deliver the needed lines and then move on with the story. However, for a visual novel artist, characters don’t necessarily get “easier” to draw just because they play a lesser role. It’s true that more “generic” characters tend to require less intense design work, but it still takes hours of the artist’s time to create those assets.

Depending on the budget and production schedule of your project, sometimes you can modify the script of your story to change the amount of art that is required for it. For example, if the main character needs to exchange several lines of dialog with her mother to advance the plot, and that brief scene is the only part of the story that involves the mother in any way, you might have that conversation take place over the phone, rather than in person, to eliminate the need to illustrate the mother character. Likewise, some teams consider it “wasteful” to have a background that is only used for a short scene, so you might actually change the flow of the story so that the actions take place in fewer physical locations.

The writer needs to accommodate the members of the development team who are contributing artwork, and usually this involves some amount of pre-negotiating things like the number of backgrounds you’re limited to, the number of characters you’re limited to, and the number of CGs. (CG is a term from the old days of Japanese visual novels, literally meaning “computer graphic,” and usually referring to bespoke pieces of artwork that are used to accompany specific scenes in a visual novel narrative.)

A CG from “Black Closet” by Hanako Games

The artists need to know in advance what it is that they’re drawing, and there are limits to the amount of work that the writer can create for them (usually dictated by the realities of production). Once the artists have generated those assets, it’s very difficult to chuck that out and ask them to start over. In a theoretical world where you have an art team with unlimited patience and unlimited time (and an unlimited budget to pay them with), none of these would be an issue, but sadly, we have to live with the production realities that come with working as part of a team creating a multi-media product.

This has two effects: first, it limits the types of visual novels that can be made. If you want to make a grand epic novel that is one million words long with hundreds of unique characters and settings, you can do that, provided you have the patience and time. However, that’s not feasible or practical in the context of most visual novel development teams.

Secondly, you have to “lock in” certain elements of your story much earlier in production. This means that visual novel writing tends to be easier for those who take an “outliner” approach and plan out the full story before writing it, rather than those who take a “pantser” (what’s sometimes described as a “discovery writer”) approach of “writing by the seat of your pants and discovering the story as you go along.” I’m more of an outliner, so it’s usually not a problem for me to come up with a complete asset budget at the start of the project that rigidly defines where the story can take place and who the participants are.

One possible approach for “pantsers” or “discovery writers” who like to begin writing their stories without a clear ending in mind could mitigate this by taking an approach where they write out a script that is more-or-less complete at the start of the project, before any of the visual asset production begins. (That way, any changes that you make to the story as you go along are already locked down by the time work starts on visual assets.) However, this makes the scriptwriter a significant bottleneck in the production pipeline.

Beyond the obvious

That’s the simple stuff that is fairly intuitive to most writers. However, there are other things about visual novel writing that make it different from normal prose writing, most of which relate to the ubiquitous “text box” that seems to be the standard for visual novels. Much in the same way that traditional prose writers often find themselves changing lines and re-arranging sentences out of consideration for things like dialog tags, “white space,” and other things related to how words will appear on a page, as a visual novel writer I’ve had to develop a different set of techniques for making sure that my scripts are tailored specifically to fit the form of visual novels.

There are exceptions, but the text box is usually located at the bottom of the screen, and usually displays only one line of dialog at a time.


Because of the way visual novel text boxes work, most visual novels are read one line at a time, rather than one paragraph or page at a time. By that, I mean that when you look at a page of paper, each line has the line before it to provide visual context; those lines are still in your field of view even if you’re not focused on them. So it’s easier to have exchanges where you have short, one word responses like this:

Speaker 1: “We need to leave now if we want to arrive by nightfall. We don’t want to delay our arrival by another day.”
Speaker 2: “Right.”

The line of dialog from speaker 1 appears on the screen, and the player reads it, then clicks to advance the text. Line 1 disappears from the text box, and is replaced by line 2.

The above dialog, as written, is acceptable, but it’s not exactly ideal for the form of a visual novel. The second line (just the word “Right”) only makes sense in the context of the previous line, and it might throw the reader off if they don’t completely remember exactly what speaker 2 is agreeing with. Unlike in a book, where you can just glance up to “remind” yourself what the previous line was, in a VN, this requires “scrollback,” or a text log, which are technical features that not all VNs support. (Most VNs do support scrollback, but even if scrollback or a text log is supported, my opinion is that effective VN writing shouldn’t rely on these features, in the same way that a well-written book shouldn’t require you to flip back to previous pages for context.) There’s also the issue that each line of dialog is its own “screen,” so any time you have a one line response, like “Yes” or “No,” you have a text box that is just one word accompanied by a LOT of empty space, and that’s generally something that you want to avoid.

So, if I were going to take the above exchange, and “re-write” it to appear in a visual novel, here are some approaches I might use.

In the previous example, the one-word line of “Right” is a bit odd, because it requires the previous line for context, but there’s nothing in the original version of line 1 that suggests that it’s going to serve as context for line 2. But we can “fix” this by making a subtle change to line 1:

Speaker 1: “We need to leave now if we want to arrive by nightfall. And we don’t want to delay our arrival by another day, do we?”
Speaker 2: “Right.”

This version of the dialog exchange modifies the first line to end with a question, and with that question mark the reader is primed to expect, “Okay, a question was just asked, and that means the next thing that appears is going to be the answer to that question.” So when character 2 says “Right” to answer the question, the reader is expecting it and walking in with the proper mental context.

However, I’d probably be more likely to try and edit things to try and make the lines more “symmetrical” in terms of length, like this:

Speaker 1: “We need to leave now if we want to arrive by nightfall.”
Speaker 2: “Right. We don’t want to delay our arrival by another day.”

This makes it so that for line 1, the text box is less crowded, and for line 2, the text box is less sparse. It’s more aesthetically uniform this way. Now, this does have its downsides. For example, there are situations in which you intentionally want to write dialog where one character’s speech is terse while another character’s manner of speech is more verbose, in which case I’d try to structure it so that one character has shorter lines and the other has longer lines. The meaning is also changed in this version, because some of the reasoning and observation that was done by character 1 in the first version of this little example is now done by character 2 in the second version.

Obviously, there are lots of other considerations and so on that become more important when you get into characterization (rather than just thinking about the most effective way to convey exposition), but this is just an example to show how the technical limitations of how visual novels work can affect the way that you’d write a script. Also, in VNs, you can sometimes get away with characters having a “similar voice” due to the fact that each character is represented by an on-screen sprite, so readers have a visual cue that helps distinguish characters apart in the text. (This kind of “lazy” writing is still to be avoided where possible, but when this kind of lazy writing does happen in a visual novel, it tends to be less distracting or problematic than it might be in a prose novel.)

Breaking ideas apart

In addition to the above considerations (which are based on the fact that the reader doesn’t physically see dialog from two different characters at once), the text box also affects the maximum length of each line. The text box generally has a fixed size, and while most visual novel engines allow for exceptions for certain moments when you want to display more text than would fit in your “default” text box, arbitrarily increasing the size of the text box for certain lines requires extra work from the programmer (and the designer too, if your UI uses a custom text box). If you use them at all, you generally want to reserve those “exceptions” for specific moments when you want to use them for effect, like this moment from Katawa Shoujo, depicting a scene in which a character suddenly “blurts out” a long run-on sentence.

I assume the reader isn’t expected to actually read all of that, much in the same way that the character that Rin is talking to in this scene probably didn’t actually hear and mentally process all of that. (Screenshot from Katawa Shoujo, Four Leaf Studios)

The particulars will of course depend on your project, but for the majority of VN projects, the upper limit on line length will likely be somewhere around 300-400 characters. That’s usually enough to fit more than one sentence, but also too short to contain what we would consider an entire paragraph in most cases. Because of this, some “paragraphs” (or ideas) need to be split into multiple text boxes, and where you draw the line makes difference in how the story is told.

I personally like to use the “split” a transition that can serve as a soft separation between two related but contrasting thoughts. For example, if I have a paragraph of description for a new setting and needed to split it into multiple lines, I might have each text box focus on a different sense, or have one text box describe the ambiance and the following text box describe the viewpoint character’s reaction to it.

You can also “pace” a scene by varying the amount of text in each text box, much in the same way that you might “pace” a prose paragraph by varying the length of each sentence. This is something that I often apply when writing dialog: text box transitions often mark “pauses” in speech, or places where the characters themselves are transitioning between contrasting thoughts.

Another feature which most visual novel engines support is having multiple “lines” within a single textbox that the player must advance through. In terms of player experience, usually when the player clicks, the text box empties and is then repopulated with the text of the next line. However, it’s also possible to “partially fill” the text box, and then when the player clicks again, keep the old text and append more text to it. In practical terms, it might play out something like this: you start with a line that begins like this:

I answered the door, and it swung open with a creak. I tensed up.

The player clicks again, and the text box is “updated” now to display:

I answered the door, and it swung open with a creak. I tensed up. It was Violet.

You can use this function to put a functional “pause” in the middle of a line. It can be a great way to build tension when you want to “withhold” the second half of a line from a reader until they’ve read all of the text leading up to that point. This is the kind of feature that I tend to use sparingly, because I generally feel that the less I use it, the more effective it will be when I do use it.

The consequence of viewing lines “one at a time”

To put it more broadly, each individual line in a visual novel needs to be able to “stand on its own” to a much greater extent than a line from a prose novel might. This leads to things like using fewer pronouns, and characters often repeating or restating part of the question they’re answering. (This aspect of visual novel writing is most noticeable when it’s done clumsily: “How do I feel about the situation? Well, let me tell you…”) One frequent complaint about VN writing is that it tends to be unnecessarily verbose, and I think that a lot of this is an artifact from aspect of the way Japanese visual novel production historically worked, where writers were paid by the word count of their script (or to be more precise, they were paid based on the file size of the script file). But as I’ve spent more time writing VNs, one thing that has become apparent to me is that the technical framework of VNs sometimes makes it sensible to use more words in some situations (like restating and recontextualizing important ideas or concepts), while using fewer words to convey other information (like visual description and “beats” related to characters’ poses and facial expressions).

Concessions to form are inevitable (and no form is exempt)

There are compromises that go into VN writing, as described in the above example. However, VNs aren’t exactly unique in that regard. For example, there are a lot of times when I’ve rewritten lines in a prose story so that the dialog tags would be more convenient, like the classic:

“Right,” said Daniel. “We don’t want to delay our arrival by another day.”

In a lot of my prose fiction, characters tend to conveniently start a lot of ideas with a one-word thought like “Yes,” or “Yeah,” or “Right,” or “Y’know,” just so I can put the dialog closer to the beginning of the sentence, rather than writing:

“We don’t want to delay our arrival by another day,” said Daniel.

That’s the kind of sentence that I would shy away from putting into a book because the reader needs to read to the last word of the sentence before you find out that Daniel is the speaker. So saying “the form of the writing has a lot of concessions to practicality” is hardly something that is exclusive to VN writing; most published books have “cheap tricks” and other concessions to practicality that they do just because they need to use language to convey ideas in a way that makes sense for the medium.

Part of being a good writer is understanding both the strengths and limitations of the medium that you’re working with, so hopefully it isn’t too controversial to claim that writing a visual novel requires certain techniques and considerations.

How Necrobarista is different

With all that being said, Necrobarista (due to the ways in which its presentation differs from that of “traditional” visual novels) gets to avoid some of the pitfalls of traditional VN writing. For example, take this screenshot:

It’s not exactly a joke, but it is a “funny” line that is a bit of a one-two punch in the form of setup into punchline, and because of that, comedic “timing” is important. If you did laugh at the line “No, but it is very cool,” the line’s brevity is probably a big part of what made you laugh. However, that same brevity would make it harder to do in a traditional visual novel, where you just have one text box that can only show one character’s dialog at a time.

If I were writing that in the form of a “traditional” VN, I would probably leave the line like that just because I find it more amusing as it is, but there would be a real cost to doing it like that, and it would be very tempting to edit the short line to include the context from the first line. e.g. something like, “It’s not really practical, but it is very cool.”

The fact that we’re working with 3D assets in Unity3D means we dodge some of the production issues that traditional VNs can face. Once we have a fully-modeled setting, having a different “background” for a scene can sometimes be as simple as changing the camera perspective, or shifting the character models to a different space within the environment. Likewise, creating new poses for a character model doesn’t require you to completely “rebuild” that model in the same way that creating new poses for a 2D drawing often requires you to simply redraw the character.

On one hand, this requires significantly more work up front (as any 3D artist can tell you), but once we have our asset, it gets much easier and more practical to “tinker” with them than it would be if we were doing everything with 2D art. These things aren’t “free” (we still have to do a lot of work to make sure the scene “works” even after it’s been rigged properly), but it’s allowed us to iterate more rapidly than we might otherwise.

How we’re doing it

Because of the unique technical framework that it uses, Necrobarista is different from other VNs that I’m working on. Ultimately, our director (Kevin Chen) is the one who decides how my script is presented to the reader, and because Necrobarista is taking such an unorthodox approach in its presentation, there’s been a lot of back-and-forth between me and Kevin during the script editing process. It’s significantly more work than writing to a static text box like I’ve done for previous visual novel projects, but I’ve been quite pleased with the results so far, and I think that Kevin’s unique directorial touch is a big part of what made our early demos so effective.

Leave a Reply

Your email address will not be published. Required fields are marked *