Skip to content

manual.md

Latest commit

 

History

History
127 lines (106 loc) · 3.6 KB

manual.md

File metadata and controls

127 lines (106 loc) · 3.6 KB

Web2Comics Engine CLI Manual

Purpose

engine/ provides a standalone command-line comic generation engine that does not require Chrome.

It accepts:

  • an input file (.html or .txt)
  • a YAML configuration file (providers, models, generation and output settings)

It produces:

  • a single PNG comic sheet (similar to extension download flow)

Command

node engine/cli/comic-engine-cli.js --input <path> --config <path> --output <path> [options]
node engine/cli/comic-engine-cli.js --url <https://...> --config <path> --output <path> [options]

Equivalent npm command:

npm run engine:comicify -- --input <path> --config <path> --output <path>

CLI Options

  • -i, --input <path>: input file path (.html, .htm, .txt)
  • -c, --config <path>: YAML config file
  • -o, --output <path>: output PNG path
  • --url <https://...>: fetch URL via Playwright and save HTML snapshot before generation
  • --snapshot-path <path>: explicit snapshot HTML path (otherwise auto-generated near output PNG)
  • --fetch-timeout-ms <ms>: Playwright page load timeout (default 45000)
  • --title <value>: override generated storyboard title
  • --debug-dir <dir>: save raw storyboard/debug artifacts (storyboard.raw.txt, storyboard.json, result.json)
  • -h, --help: print usage
  • -v, --version: print CLI version

Environment Variables

The engine loads .env.e2e.local and .env.local from repo root (if present).

Supported keys:

  • GEMINI_API_KEY
  • OPENAI_API_KEY
  • OPENROUTER_API_KEY
  • HUGGINGFACE_INFERENCE_API_TOKEN (or HUGGINGFACE_API_KEY)
  • CLOUDFLARE_ACCOUNT_ID
  • CLOUDFLARE_API_TOKEN

YAML Format

Example structure:

input:
  format: auto           # auto|html|text
  max_chars: 14000
  strip_selectors:
    - script
    - style
    - nav
    - footer

generation:
  panel_count: 3
  objective: summarize
  output_language: en
  detail_level: low
  style_prompt: "clean comic panel art, consistent characters"

providers:
  text:
    provider: gemini     # gemini|openai|openrouter|cloudflare|huggingface
    model: gemini-2.5-flash
    api_key_env: GEMINI_API_KEY
  image:
    provider: gemini
    model: gemini-2.0-flash-exp-image-generation
    api_key_env: GEMINI_API_KEY

runtime:
  timeout_ms: 120000
  image_concurrency: 3
  retries: 1

output:
  width: 1400
  panel_height: 700
  caption_height: 120
  header_height: 120
  footer_height: 34
  padding: 24
  gap: 16
  background: "#f8fafc"
  brand: "Made with Web2Comics Engine"

Included Example Configs

  • engine/examples/config.gemini.yml
  • engine/examples/config.openai.yml

Example Runs

Gemini:

npm run engine:comicify -- --input engine/examples/sample-story.txt --config engine/examples/config.gemini.yml --output engine/out/gemini-comic.png --debug-dir engine/out/gemini-debug

OpenAI:

npm run engine:comicify -- --input engine/examples/sample-story.txt --config engine/examples/config.openai.yml --output engine/out/openai-comic.png --debug-dir engine/out/openai-debug

Saved HTML:

npm run engine:comicify -- --input path/to/page.html --config engine/examples/config.gemini.yml --output engine/out/page-comic.png

Live URL (snapshot + original flow):

npm run engine:comicify -- --url https://en.wikipedia.org/wiki/Comics --config engine/examples/config.gemini.yml --output engine/out/wiki-comic.png --debug-dir engine/out/wiki-debug

Performance Notes

  • Increase runtime.image_concurrency for faster multi-panel image generation.
  • Use lower panel_count for lower latency.
  • Keep input.max_chars focused to reduce storyboard latency.

Test Command

npm run test:engine