Skip to content
TheChatArchiveThe Chat Archive
DeepSeekdeepseek-v3-0324· DeepSeekHugging Face LMArena Human Preference 140K

I have a working renpy script with badly written dialogue, I want a LLM to fix the…

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with Amazon Novaamazon.nova-pro-v1:0’s answer.
  1. User

    I have a working renpy script with badly written dialogue, I want a LLM to fix the dialogue in the renpy script, what prompt should I use ?

  2. DeepSeek Assistant

    To have an LLM (like ChatGPT) improve the dialogue in your Ren'Py script while preserving the structure, variables, and labels, you can use a prompt like this:


    Prompt:
    "I have a Ren'Py script with poorly written dialogue. I want you to rewrite the dialogue to be more natural, engaging, and character-consistent while keeping all Ren'Py-specific code (labels, jumps, variables, etc.) unchanged. Here’s the script:

    [PASTE YOUR SCRIPT HERE]

    Guidelines:

    1. Preserve Code: Do not modify labels, jumps, $ expressions, or Python blocks. Only edit dialogue lines (text in quotes after say statements).
    2. Improve Dialogue:
      • Make conversations flow naturally.
      • Ensure character voices are distinct and consistent.
      • Fix awkward phrasing, exposition dumps, or unnatural tone.
    3. Keep Intent: Don’t alter plot-critical meaning or choices.
    4. Formatting: Maintain line breaks (\n) and special Ren'Py tags (e.g., {i}, {b}).

    Return the full script with only dialogue improved, and no other changes."


    Example Input/Output:

    Original Script Snippet:

    label start:
        e "I am very mad at you because you did a bad thing yesterday."
        mc "I am sorry. I won’t do it again."
    

    Improved Output:

    label start:
        e "How could you do that yesterday? I’m furious with you."
        mc "I... I didn’t mean to hurt you. It won’t happen again."
    

    Additional Tips:

    • If characters have defined traits (e.g., shy, sarcastic), mention them in the prompt.
    • For long scripts, process them in chunks to avoid truncation.
    • Specify genre/tone (e.g., "Make the dialogue darker/more humorous").

    This ensures the LLM respects Ren'Py’s syntax while upgrading the dialogue.

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.