visual-composition

原文涵盖的主题

• Visual Composition (with annotation)
• Purpose
• When to invoke
• Inputs
• Required intermediate representation
• Procedure
• Medium adjustments
• Hard constraints

Given honest encodings, make the artifact read correctly at a glance: the intended element is seen first, the layout supports the intended comparison, and every mark is explained. Composition converts a correct encoding into a comprehensible communication. After visual-encoding; before visual-render-qa. Also when a figure is "busy", has a weak focal point, relies on a detached legend, or lacks a self-contained caption. The encoding table from visual-encoding and the intent record's takeaway, audience, and medium. A composition plan: 1. Assign layers: exactly one primary, zero-to-three secondary, optional context. Verify the primary layer has the strongest combination of position, size, contrast, and isolation (HIER-01). If context competes at first glance, reduce its weight (HIER-04, MAP-02 for maps). 2. Set up the intended comparison with grouping cues — proximity, enclosure, shared color (HIER-02). Start neutral (grays) and add emphasis only to the message (HIER-03). 3. Lay out for figure-ground and balance: subject reads as foreground (LAYOUT-03); remove non-informative ink unless it serves the reader (LAYOUT-01); prefer small multiples over one overloaded chart (LAYOUT-02); ali

这里只展示节选，请以完整原文和当前要求为准。