Expo image collections

Context

Docs organizes explanations, Blog organizes posts, and Book organizes chapters. Image collections need albums, a compact overview, and a way to inspect each original without making every image into a separate article.

The initial Graph collection contains 7,658 originals across 149 album pages. A separate copy supplied real mixed dimensions, long images, animated GIFs, captions, and existing image anchors. Per-image content files would increase authoring work without improving this collection’s navigation.

Decisions

An album is one Hugo leaf bundle: index.md plus images/. Filenames supply titles until an author adds native resources metadata in the album’s front matter. No separate per-image Markdown, YAML, database, or manifest is required. This keeps the entry cost small while allowing captions, ordering, attribution, explicit IDs, and separate originals when needed. Renaming a file changes its fallback anchor; authors who need lasting citations should provide explicit IDs.

Expo is a layout, independent of Paper, Slate, Ink, Terminal, and light/dark mode. It reuses the shell’s navigation and appearance controls, shared media results, and existing colors, typography, and shape tokens. The original gallery fence remains useful inside ordinary text pages.

Three radio cells select wall, cards, or seamless display, followed by motion control and a visible size slider. Reader layout choices persist across albums; appearance stays in navigation and album pages omit the article action menu. Icons reduce visible UI text; localized accessible names and tooltips remain necessary for keyboard and assistive-technology users.

GIFs play by default, following the maintainer’s October 13 revision. A pause control and the system’s reduced-motion preference provide a static alternative. This increases transfer cost for animated albums; images are lazy-loaded, and authors remain responsible for dividing very large collections into albums. Automatic resource pagination is outside this version’s scope.

The viewer starts at original size, shrinking only to fit the available stage. Expansion uses the more constrained edge and always contains the whole image; manual zoom supports inspecting details and scrolling long images. An original link and an album location link preserve access and citation. Album sequencing belongs to Expo’s conditional runtime, while ordinary Image Zoom keeps its single-image behavior. The viewer uses a native modal dialog rather than adding a lightbox dependency. Without JavaScript, original-image links and captions remain usable.

One item resolver supplies HTML, Print, Markdown, LLMSFULL, RSS, and album covers. Static formats retain image content and metadata without depending on viewer JavaScript. The cost is maintaining these outputs alongside the visual layout.

Validation and ownership

The theme owns expo/items.html, the layout, the conditional runtime, check-expo.py, geometry tests, and generated schema projections. The sibling site owns bilingual examples, tests/browser/expo.spec.mjs, and rendered desktop/mobile, accessibility, appearance, and output checks.

The copied Graph collection is additional large-content evidence. Its acceptance checks verified all 7,658 original hashes, stable album anchors, and the unchanged source tree. Generated output, screenshots, and personal image data remain outside these source repositories. Local acceptance, a public module tag, consumer upgrades, and deployment are separate completion states.