🎭 RimTalk StyleExpand - Style Module
[i]Make RimTalk dialogues reflect any writing style you want![/i]
[hr]
📢 Community
Join our QQ group for feedback and discussion: [b]1076958977[/b]
[hr]
✨ Introduction
Love the writing style of a certain novel? Want your colonists to speak with an ancient Chinese flavor or medieval fantasy tone?
This MOD helps you [b]replicate any writing style[/b] from text samples, making RimTalk dialogues match your favorite style!
Simply provide a style sample (novel excerpts, etc.), and the MOD will automatically analyze and inject style features into dialogue generation.
[hr]
🔧 Requirements
[b]Required Dependencies[/b]
[list]
[*] RimTalk Mod
[/list]
[b]Embedding API Configuration (Required)[/b]
The MOD requires an Embedding API for text vectorization. We recommend [b]Ollama local deployment[/b]:
[b]📥 Ollama Setup[/b]
1. [b]Download and install Ollama[/b]: [url=https%3A%2F%2Follama.ai%2Fdownload]https://ollama.ai/download[/url]
2. [b]Run Ollama application[/b] (ensure background service is running, check taskbar)
3. [b]Open terminal and run to download model[/b]:
ollama pull nomic-embed-text
4. [b]Configure in MOD settings[/b]:
[list]
[*] API URL: [url=http%3A%2F%2Flocalhost%3A11434%2Fapi%2Fembed]http://localhost:11434/api/embed[/url]
[*] Model: nomic-embed-text
[/list]
[i]💡 You can also use OpenAI, DeepSeek or other compatible APIs[/i]
[hr]
🎮 How to Use
[list]
[*] 1️⃣ Place style samples (.txt files) in the Styles folder
[*] 2️⃣ Open MOD settings in-game → Scan Styles
[*] 3️⃣ Select a style → Chunk Style
[*] 4️⃣ Done! RimTalk dialogues will now reflect the style
[/list]
[hr]
💡 Tips
[b]Recommended: Use "Generate Style Description" feature[/b]
The MOD includes LLM analysis to automatically analyze style samples and generate style descriptions. This produces better results than using retrieved chunks alone!
Configure your LLM API in settings (can reuse RimTalk's config), select a style, then click "Generate Description".
[hr]
🧠 Features
[list]
[*] [b]Style Management[/b]: One txt file = one style
[*] [b]Semantic Chunking[/b]: Smart chunking with semantic coherence
[*] [b]Vector Retrieval[/b]: Auto-retrieve relevant style chunks
[*] [b]LLM Analysis[/b]: Optional auto-generated style descriptions
[*] [b]RimTalk Integration[/b]: Auto-register variables, works out of the box
[/list]
[hr]
📝 Version History
[b]v1.0.2[/b]
[list]
[*] Fixed Chinese encoding issues
[*] Fixed Scriban variable registration issues
[*] Optimized default prompt templates
[/list]
[b]v1.0.1[/b]
[list]
[*] Added "Get Models" feature to fetch available models
[*] Fixed Ollama API compatibility issues
[*] Fixed config file sharing violation
[*] Improved new player experience
[/list]
[b]v1.0[/b] - Initial release
[hr]
📌 Notes
[list]
[*] Recommended sample size: 5,000 - 20,000 characters (Chinese)
[*] First-time chunking requires API calls, may take time for large files
[*] Progress is auto-saved, can resume after interruption
[/list]
[hr]
🙏 Credits
[list]
[*] RimTalk author
[*] All playtesters
[/list]
[hr]
🔗 Links
[list]
[*] GitHub: [url=https%3A%2F%2Fgithub.com%2FNUOQIP%2FRimtalkStyleExpand]https://github.com/NUOQIP/RimtalkStyleExpand[/url]
[*] Issues and PRs welcome!
[/list]
[hr]
[i]If you like this MOD, please subscribe and favorite ⭐[/i]
Questions? Join our QQ group: [b]1076958977[/b]
