Skip to content

user-manual.md

Latest commit

 

History

History
130 lines (113 loc) · 5.29 KB

user-manual.md

File metadata and controls

130 lines (113 loc) · 5.29 KB

Web2Comic Bot User Manual

Cross references:

Scope note:

  • This manual documents public user commands and flows only.
  • Admin-only bot commands are intentionally excluded.

Full Public Command Catalog

  • Onboarding/info:
    • /start, /welcome, /help, /about, /version, /user, /config, /explain, /debug <on|off>
  • Story generation:
    • Send text, web link, PDF, image, or voice/audio
    • /invent <story>
    • /random
  • Replay/history:
    • /peek, /peek <n>, /peek<n>
  • Providers/models:
    • /vendors [role], /vendor <role> <name>, /vendor <name>
    • /models [text|image|url|image_extract|pdf|voice] [model]
    • /test
  • Controls:
    • /panels <count>, /objective [name], /objectives
    • /language <code>, /mode <default|media_group|single>
    • /consistency <on|off>, /detail <low|medium|high>
    • /crazyness <0..2>, /concurrency <1..5>, /retries <0..3>
  • Objective shortcuts:
    • /summary, /fun, /learn, /news, /timeline, /facts, /compare, /5yold, /eli5, /study, /meeting, /howto, /debate
  • Style commands:
    • /style <preset-or-your-style>
    • /new_style <name> <text>
    • Built-in style shortcuts: /classic, /noir, /manga, /superhero, /watercolor, /newspaper, /cinematic, /anime, /cyberpunk, /pixel-art, /retro-pop, /minimalist, /storybook, /ink-wash, /line-art, /clay-3d
  • Prompt/options:
    • /prompts
    • /set_prompt story <text>
    • /set_prompt panel <text>
    • /set_prompt objective <name> <text>
    • /list_options, /options <path>
  • Keys/reset:
    • /keys, /setkey <KEY> <VALUE>, /unsetkey <KEY>, /reset_config, /restart
  • Behavior:
    • Unknown /... commands are rejected as unrecognized and not treated as story input.

What The Bot Does

Send plain text, a web link, a PDF link/file, an image link/file, or a voice/audio message, and the bot generates comic panels and sends them back as ordered Telegram images.

Panel captions are prefixed as X(Y):

  • X = current panel number
  • Y = total panel count

Quick Start

  1. Send /start
  2. Send /help
  3. Run /user to get your Telegram id
  4. Add provider key with /setkey GEMINI_API_KEY <YOUR_KEY>
  5. Send a story, link, PDF, image, or voice note

Core Commands

  • /help command list
  • /about creator and project links
  • /config current runtime config
  • /user show your Telegram id
  • /keys key status

Generation Controls

  • /vendors [role]
  • /vendor <role> <name>
  • /vendor <name>
  • Legacy aliases: /text_vendor <name>, /image_vendor <name>
  • /language <code>
  • /consistency <on|off> enable style-consistency flow (supported image models only)
  • /crazyness <0..2> control story-invention temperature (higher = wilder twists)
  • /panels <count>
  • /objective list objectives
  • /objective <name> set objective
  • /style <preset>
  • /new_style <name> <text>
  • /detail <low|medium|high>
  • /concurrency <1..5>
  • /retries <0..3>

Prompt Controls

  • /prompts show active prompt templates
  • /set_prompt story <text>
  • /set_prompt panel <text>
  • /set_prompt objective <name> <text>

Options UI In Chat

  • /list_options list all config paths with predefined options
  • /options <path> show numbered options for one path
  • Apply options via dedicated commands (/objective, /panels, /mode, /vendor, /models, /language, etc.)

If /options is called without required args, the bot explains usage and lists valid paths/options.

Story Modes

  • Normal text: generate directly
  • Very short prompt: bot first expands story with AI, shows invented story, then generates comic
  • URL input: bot prints the exact parsed URL, renders page content, then uses it as source
  • PDF input: bot extracts document text from a PDF link or uploaded PDF file, then uses it as source
  • Image input: bot extracts scene/story context from an image link or uploaded image file, then uses it as source
  • Voice/audio input: bot transcribes the clip, then uses the transcript as source

Image Prompt Context

For each panel image, the bot sends context that includes:

  • Background (short story summary context)
  • Image description (panel.image_prompt)
  • Style
  • strict no-text requirements (English, Hebrew, Russian)

This keeps panel visuals coherent across the full story.

Consistency Mode

  • Default: on
  • Command: /consistency on or /consistency off
  • When enabled, after storyboard generation the bot first creates one summary style-reference image.
  • Then each panel uses prompt text that explicitly asks to follow that reference style and passes the summary image as reference input.
  • This is used only when the active image provider/model supports image-as-input; otherwise the bot falls back to normal panel generation.

Panel Output Details

  • Watermark is configurable and defaults to on
  • Bot messages/photos are forwardable (content protection is disabled)
  • Panel captions are sent in order and prefixed as X(Y)

Reset And Recovery

  • /reset_config clear runtime overrides
  • /restart reset user state and re-onboard
  • /unsetkey <KEY> remove runtime key override