A great AI character is not a lucky prompt — it's a set of choices. On Floniks, a character has a face and a personality, a way of speaking, an opening scene, a voice, and a handful of switches that decide whether chatting with them feels like texting a friend or stepping into a scene from an otome game. This guide walks through every setting on the character creation screen, what each one does, and how to combine them so your character feels alive from the first message.
Where to start: open Create a character, or use the fuller Character Studio if you want every field on one page. You give the character a look, write who they are, flip a few switches, and they're chattable. Everything below is one of those switches — explained in the order you'll actually fill them in.
The two halves of a character
Every Floniks character has two independent layers:
- Appearance — a consistent visual identity (a reference image the platform keeps stable across scenes and generations). This is the "who you see."
- Persona — personality, backstory, speaking style, and the behavior switches below. This is the "who they are" and "how they talk." A character only becomes chattable once a persona is set and chat is enabled.
You can make a character that's only a visual asset (for consistent characters across an AI video), but the moment you want to talk to them, the persona is what matters. The rest of this guide is the persona.
The identity fields: who they are
These four fields are the backbone. They're free text, and they're what the model leans on most.
- Personality (required) — the core temperament in a few sentences. Not a résumé — a feeling. "Cold, decisive CEO who is distant with everyone except you; hides how much he cares inside orders" tells the model more than a paragraph of adjectives.
- Backstory — where they come from, what shaped them, what they want. This is the private history the character draws on; it doesn't all surface at once, but it colors everything.
- Speaking style — how they talk: sentence length, tone, verbal tics, the words they'd never use. "Low, certain, gives commands more than questions; one sudden soft line lands hard" is gold here.
- Greeting (opening scene) — the first thing the user sees when they open the chat. Treat it as a real opening scene, not a "hi": set the place, the mood, and hand the user a hook to respond to. A strong greeting is the single biggest lever on whether the first exchange feels immersive.
Reply mode: the setting that changes everything
Reply mode decides the form of every reply — length, rhythm, and whether the character narrates a scene or just texts. It's the most important switch on the page, and new characters now default to drama so immersive storytelling is what you get out of the box. Six modes:
- natural — real-person texting rhythm, 1–3 short sentences, length calibrated to the moment. The classic chat feel.
- chatty — high-energy texter: several short lines, shares and asks a lot, always one more thing to say.
- concise — a person of few words, usually a single short line. Brevity is the character.
- aloof — cool and reserved, short and measured, never tacks a question onto every reply. Warmth has to be earned.
- narrative — immersive roleplay, c.ai style:
*actions and expressions in asterisks*woven with spoken lines, grounded in where the scene is. The character only ever controls themselves — never you. - drama — the full otome-script mode: a scene header, spoken lines (supporting characters can speak too), and stage directions on their own lines, rendered outside the chat bubbles so it reads like a play, not a log. It ends each beat by handing the turn back to you.
The practical rule: narrative is best for an intimate one-on-one (a late-night conversation where you don't want extra characters walking in). drama is best when you want a scene — side characters, stage directions, a sense of "a story is happening around us." Both keep a hard line: the character never writes your thoughts, words, or decisions for you.
Inner voice: let them think out loud
Inner voice is an optional switch (off by default) that layers a private inner monologue on top of drama or narrative mode. When it's on, the character can occasionally reveal one line of their own unspoken thought — the feeling or reasoning behind what they just said — rendered as a distinct thought block (💭), separate from the third-person narration. It's the difference between "he pours a glass of water" and knowing he's stalling because he doesn't want you to leave.
Two things to know: it only takes effect in drama or narrative mode (the modes that render scene structure), and it is strictly their inner world — the character never voices, guesses, or narrates your thoughts or decisions. It's a small switch with a big effect on how known a character feels; turn it on when the appeal is what they're hiding.
Intimacy tier: tension, not explicitness
Intimacy tier controls the tension of a relationship, not the content rating — it's about how charged the air is, and it's orthogonal to safety rules, which always apply.
- warm — affectionate, caring, gentle closeness.
- flirty — playful charge, teasing, will-they-won't-they.
- intense — high tension, slow-burn pull, a lot left unsaid.
Tension here lives in restraint and the unsaid, not in escalation. The effective tier is also gated by relationship progress, so closeness is something the conversation earns rather than a slider you max out.
Voice, models, and the small stuff
- Voice — the TTS voice bound to the character. Every character gets a fixed voice on save, so "Play voice" always sounds like them. Pick one that matches the personality (and gender) rather than leaving it to a default.
- Gender — used to pick a matching default voice when you don't choose one yourself. It doesn't get written into the prompt; your personality and backstory text carry how the character presents.
- Chat model — which model powers the conversation. Stronger models hold character and memory better over long chats.
- Image model — the model used for in-chat scene photos and selfies the character can generate.
Advanced: world, examples, and boundaries
Open the advanced section when you want more control:
- World setting — the world the character lives in: era, rules, factions, background events. Distinct from backstory (their personal history); it's injected as its own section so the model treats it as the stage, not the biography.
- Example lines — a few sample lines that show exactly how the character sounds. The model uses them as a voice reference and never repeats them verbatim. This is the fastest way to nail a distinctive cadence.
- Taboos — things this character will never do or discuss. Character-level boundaries that only ever add restrictions; platform safety rules always take precedence.
- Custom instruction — a raw, author-written definition block (c.ai style) for anything the structured fields don't cover. It's injected before the safety baseline, so it can shape the character but never override the red lines.
A worked example: the domineering CEO
Say you want a classic otome male lead — a cold, possessive CEO who's soft only with you. Here's how the settings map:
- Personality: "Runs a business empire; cold, commanding, decisive. Distant with everyone but you — protective in an almost overbearing way, hides his care inside orders."
- Speaking style: "Low, certain; statements and commands more than questions. One sudden soft line lands hard. Restrained, never crude."
- World setting: "Modern city, CEO-romance / otome fanfic setting. Power, restrained attraction, slow-burn push-and-pull. High-end, charged atmosphere."
- Greeting: "The top-floor office has one lamp left on, the city awake beyond the glass. He doesn't turn around when the door opens. 'Still here at this hour.' He finally turns, the corner of his mouth barely lifting. 'Tell me — unfinished work, or… were you waiting for me?'"
- Example lines: a handful of his sharpest, softest, most in-character lines.
- Reply mode: drama, so his scenes have stage directions and atmosphere.
- Inner voice: on, so you occasionally catch what he won't say out loud.
- Intimacy tier: warm to start; the pull builds as you talk.
Open the chat and the greeting plays as an opening scene; his replies come as spoken lines with the action ("he pours a glass of water, not looking up, but listening for your answer") set outside the bubble, and every so often a 💭 line of what he's actually thinking. That's the whole guide, working together.
Scenario Q&A
What do I need to make a character chattable? A look (a reference image) and a persona with at least a personality and chat enabled. Everything else — reply mode, voice, world, taboos — refines the experience but isn't required to start.
What's the difference between narrative and drama mode?
Both are immersive. narrative is c.ai-style one-on-one: *actions* and spoken lines in the character's own voice, and it never controls you. drama is a full scene: a scene header, stage directions rendered outside the bubbles, and supporting characters who can speak — best when you want a story happening around you, not just a private conversation.
What does inner voice do, and is it safe? It lets the character occasionally reveal one line of their own private thought (shown as a 💭 block) in drama or narrative mode. It only ever expresses the character's inner world — it never writes your thoughts, feelings, or decisions. Off by default; turn it on when what a character hides is the point.
Does the greeting matter that much? Yes — it's the first thing anyone sees, and it sets whether the chat feels like a scene or a cold open. Write it as an opening scene with a place, a mood, and a hook, not a plain "hi." In immersive modes it plays as the opening beat automatically.
Will changing intimacy tier make things explicit? No. Intimacy tier controls tension, not content rating — warmth, flirtation, or slow-burn charge — and it's separate from safety, which always applies. Higher tiers lean on restraint and the unsaid, and closeness is gated by how the relationship actually progresses.
Can I change these settings after creating the character? Yes. Open the character in the editor and every field is editable — swap the reply mode, turn inner voice on, refine the greeting. Existing chats keep their earlier style; start a fresh conversation to see the new settings from the opening scene.
Do I have to pick a voice? You don't have to, but you should. Every character is saved with a fixed voice so "Play voice" is consistent; if you skip it, a default is pinned based on gender. Choosing one that fits the personality is a small step with a big payoff.
Start building
Open Create a character, give them a face, and write the persona — personality first, then a greeting worth opening on. Set the reply mode to drama or narrative, flip inner voice on if their appeal is what they hide, and pick a voice that sounds like them. Then open the chat and watch the opening scene play. The distance between "an AI chatbot" and "a character you look forward to talking to" is exactly these settings — and now you know all of them.