LifeSim: The Deep Roleplay Engine

1. Connection

OpenRouter API Key

Narrative Model

2. Simulation Settings

Game Mode

Prompt Verbosity

Creativity (Temp)

Stream output (shows text as it generates)

Hardcore Mode (no retries — disable for swipe/edit/delete controls)

Main LLM Memory (past actions)

Number of past action/response pairs the narrative model can see. Higher = more context but slower & more expensive.

Character Name Gender Starting Scenario

3. Image Generation (Optional)

Enable NovelAI scene illustrations

NovelAI Persistent API Token NovelAI Image Model Danbooru Tag Converter Model (OpenRouter)

This model reads each scene and converts it into Danbooru-style tags for the image generator. Recommended: a capable instruct model (Claude, GPT-4o, Kimi, etc.)

Tag Converter Memory (past actions)

Number of past action/response pairs (with their generated tags) the tag converter can see for style consistency.

Tag Conversion Mode

Simple: flat comma-separated Danbooru tags. Works with all NAI models. Good for solo/duo scenes.
Complex: pipe-separated character clauses with action interaction tags. Requires NAI Diffusion V4+. Significantly better multi-character composition and control.

Generate Illustration Every

Critical / consequence events always generate an image regardless of interval.

Character Visual Description

Character Preview

Generate a preview to see your character

Generating preview...

LIFESIM ENGINE

1. Connection

2. Simulation Settings

3. Image Generation (Optional)