Manga panel composition follows a learned visual grammar distinct from that of American or European comics. The page is the unit of design — not the panel — and individual panels relate to the page through size, position, and the directional reading flow. This page collects the most-cited conventions and their conventional functions, drawn from Scott McCloud's Understanding Comics, Natsume Fusanosuke's writing, and the Japanese manga theory of the 1990s.
Reading order on a page
A Japanese manga page is read from the top-right corner. The eye moves leftward across the row, then down to the next row, and so on. Within a row, panels are arranged so the reading order is unambiguous; ambiguity is corrected with gutter spacing (larger gaps between row boundaries than between column boundaries) and occasionally with explicit arrows in tight layouts.
Panel-to-panel transitions (after McCloud)
Scott McCloud's Understanding Comics (1993) identifies six transition types between adjacent panels. McCloud's later analysis (Making Comics, 2006) found that Japanese manga uses the 'aspect-to-aspect' transition — moving across distinct aspects of a single scene without temporal progression — substantially more than American and European comics do. The aspect-to-aspect transition is unusual in Western comics but routine in slice-of-life manga, where pages may dwell on the texture of a moment for half a page or more.
Page-level composition
A manga page is composed as a single image, with panels arranged for visual rhythm as well as story sequence. Large panels carry weight; small panels indicate quick action or rapid passage of time. Two-page spreads — mihiraki (見開き) — are reserved for moments of major impact: the first appearance of a new character, the reveal of a setting, the climax of a fight.
Speech balloons and tails
Japanese speech balloons are vertical to accommodate vertical kanji text and are typically thin ovals or rectangles. Thought balloons use cloud-edge forms; off-screen or distant speech is rendered without a tail, often in different lettering. Internal monologue is rendered in unboxed text floating beside the speaker — a convention that does not survive English translation, since horizontal English text in unboxed positions is difficult to typeset cleanly.