Turning the pages of a whole document
Books, booklets and multi-page documents survive as a sequence, and they are best understood that way. This experiment presents an item as a book you can leaf through, jump around with a thumbnail strip, search by what is printed on each page, and open any single page full-screen to zoom and pan.
Loading the document…
What this experiment tests
The viewer is driven entirely by item-data.json: the page images, each page’s derived
heading, its full transcribed text (shown in the details panel, expandable per page) and simple
metadata (page position, what kind of content it carries, how many text blocks were captured). The
text filter searches the complete transcription, not just the headings. The same page could be reused
for any item with multiple elements — a letter with several sheets, a photograph album, a
booklet — by pointing it at a different data file.
It also tests reading ergonomics: a two-page spread with a real page-turn on wider screens, a single-page mode, keyboard navigation (← →, f to zoom, Esc to close), a thumbnail strip, and a text filter that dims pages that do not match and steps between the ones that do.