Plan scene action before dialogue
game-world-design establishes scene purpose, agendas, knowledge, player intervention and exit state before lines. A visual novel can prioritize reading while consequential interactions still change responses, information or opportunities.
Original gallery scene: a borrowed photograph is missing after closing. The keeper protects confidence in the loan process; its owner wants the object traced. The player sees an empty frame but does not know who removed it.
The common scene obtains grounds for inquiry. Adapt interior prose through the chosen viewpoint; narration should not declare concealment before investigation establishes it.
Expand that setup into a readable original script proposal. The player is in the closed gallery beside an empty frame. The keeper’s “The register says last night” remains a claim. These new local exchanges do not predetermine who removed the photograph:
The keeper was collecting keys into a box. “The register says last night.”
The player chooses “Inspect the register first.” The keeper slides it over. The player sees a crossed-out date and a new date beside it, without knowing who made the change.
On returning to questions, “Why was this date changed?” becomes available.
The keeper stops collecting keys. “Someone told me to change it. I haven’t checked the receipt page. I need another look.”
The player chooses “I’ll check that page with you.” The keeper agrees and indicates the handoff cabinet. They begin auditing the process without establishing the photograph’s location.
On another path the player does not inspect the register, so the alteration question is unavailable. “Who signed for the photo?” receives “I haven’t checked the receipt page. Will you look with me?” In this example the player selects “Yes, let’s check,” and the keeper indicates the cabinet. Declining would not establish cooperation, and the first path’s observation does not enter this history automatically.
Selected observation, questions and cooperation occur, while “someone told me” remains an explanation candidate. Neither path recovers the photograph yet. At convergence by the cabinet, an inspector may follow up on the date; an uninspected path cannot claim “I saw it earlier.” This passage demonstrates the common scene and investigation entry, not a completed character route or implemented game.
Write conditions and actual results into the scene table
| Segment | Presentation, line and state |
|---|---|
| Common opening | Empty frame and rain. Keeper says the register gives last night; this establishes a claim, not verified truth. |
| Inspect register | Date close-up. Record that the player saw alteration, without identifying the editor. |
| Return after inspection | Keeper pauses while collecting keys. “Why was this date changed?” appears only after inspection; the reply is an explanation candidate. |
| Uninspected question | In the same gallery, “Who signed for the photo?” does not assume unseen evidence. |
Record entry and exit states; vary expression and sound when useful. Line IDs can connect text, captions and voices. Adapt the format to the production tool: this worksheet is not universal executable engine syntax.
Give each character route its own question
| Route | Aim, action and payoff |
|---|---|
| Audit with keeper | Find the handoff failure; review signatures; the keeper needs to acknowledge a procedural lapse. The route aims for traceable loan records. |
| Trace with owner | Recover the object and its personal meaning; follow recipient evidence while the owner chooses whether to disclose private context. The route aims to recover the photo and agree custody. |
These are new paper-design routes for the example. Open them through established investigation and explicit cooperation, not an unrelated hidden chat choice.
Routes may share a gallery, but swapping portraits cannot replace distinct conflicts. A core route needs sustained relationship, leverage and payoff. Keep a brief encounter as a supporting line until broader depth is justified.
Check convergence, reading and production handoff
Walk each route from the common scene and trace entry requirements backward. Characters must not refer to unseen evidence, unmet cooperation or another route’s facts. Give failure and return paths clear destinations.
Inspect choices at the actual font and viewport. Place necessary evidence near the decision and long records in accessible archives. Keep picture, voice and text on current line records; synchronize affected entries after edits.
Ask AI to extract source facts before proposing adaptation changes. Confirm those changes before expansion. Test input, outcome and return to interaction in the target runtime; smooth dialogue alone does not prove playability.
FAQ
Should visual novel routes be equally long?
No. Give each route enough space for its conflict and payoff, and set expectations through the game’s structure. Unequal length is fine when the shorter route still feels complete.
Do it with the skill
Use $game-world-design to define viewpoint, agendas and entry/exit states before a visual-novel line and choice table. Give routes distinct aims, label adaptation additions and check knowledge at convergence.
Read the method: Scenes, knowledge and branches · Character agendas and choice dialogue · Route depth and scope