real-world

Real-world rendering gallery

Four real third-party documents plus one constructed feature showcase, rendered through inkmd as-is, with honest notes about what worked and what did not.

The docs/gallery/ directory one level up contains spec-edge-case verification PDFs: documents specifically constructed to exercise tricky CommonMark or GFM rules. This real-world/ directory is the opposite: documents that real authors wrote for real audiences, rendered through inkmd to show what a visitor would get if they pointed the tool at a document from their own life.

The documents

Document	Source	Pages	Notes
Simon Willison TIL	simonw/til	4	Clean render; longform blog post with code blocks
Rust Book §1.3 "Hello, Cargo!"	rust-lang/book	5	Clean render; multi-page technical book chapter
inkmd's own README	This repo	14	Dogfood; the centred <p align><img> hero renders, image embedded
Ruff README	astral-sh/ruff	11	Mixed; color emoji render, SVG badges fall back to alt-text
Non-Latin scripts	Constructed showcase	2	Non-Latin scripts: Cyrillic, Greek, and Latin-Extended render via the embedded font; unsupported codepoints (e.g. CJK) show a visible [U+XXXX] marker. Table cells still render ? (known limitation)

Each subdirectory contains:

• source.md: the exact markdown that was rendered
• output.pdf: the PDF inkmd produced
• README.md: what the document is, where it came from, and what to look at (including any rendering limitations)

How these were rendered

All were compiled with:

inkmd.compile(
    source_text,
    page_size="letter",
    family="helvetica",
    allow_remote_images=True,
    base_dir=document_directory,
)

allow_remote_images=True was set so that any HTTP image references would be fetched at compile time rather than falling back to alt text. In practice this only affected the Ruff README, where the shields.io badges (all SVG) still fell back to alt text because inkmd supports PNG and JPEG image formats only.

The render is reproducible: see scripts/render_gallery.py to re-run.

Why this gallery exists

This gallery shows what inkmd produces on documents real authors wrote for real audiences, not a curated demo: the README of a known project, a chapter of a book, a blog post in a recognisable style. Those documents cover those shapes; the non-Latin showcase adds a direct look at the v0.4 font-embedding feature on text the base fonts cannot draw. The README for each one is candid about what fell short.