Narrative-driven Presentation Design
Organize material around a single, overarching question. On each slide, let the audience first see a discovery, a scene, or an original excerpt, then provide the necessary evidence. The visual language should be drawn from objects, scenes, contrasts, or data baselines present in the topic.
Choose a Narrative Structure
- Surveys & Research. Start with a counter-intuitive statistic, state the sample and methodology, show distributions and differences, then link recommendations back to the corresponding evidence. Reference Fruit Survey.
- Institutional, Technical & Social Case Studies. Use a familiar moment for the audience to introduce how a system works, demonstrate its impact through cases and data, then close by revisiting the same character, interface, or question. Reference Algorithm Chat.
- Literary, Character & Quotation Close Reading. Enter with a single verifiable quote, flesh out the scene, magnify key words and objects, then bring the interpretation back to the audience's experience. Reference English Close Reading.
Reserve one or two recognition points throughout, such as chapter identifiers or recurring objects. Plan the visual rhythm first, then detail each slide. Between the opening, main sections, and transitions, choose from documentary full-bleed images, light-background evidence, dark-themed scenarios, keyword close-ups, or colored section pages based on the storytelling task. Reuse layouts when comparing data from the same group. See Page Type Variations & Template Study.
Before handing off to another AI or rendering tool, specify the user-defined references, composition, alignment guides, font roles, text color scales, graphic styles, and whitespace. Deliver these requirements alongside the page, and state clearly which layout decisions the tool is responsible for executing. Components should use their own geometry and colors. Re-read the rendered output and correct any themes, decorations, or font size modifications automatically added by the tool. See Tool Handoff & Style Verification.
When user references are available, start by extracting the visual focal point, reading order, and whitespace usage from specific pages. For a new set of slides, you can first create representative pages for different tasks, establish shared visual rules, then expand. When encountering branches, loops, or handoffs in responsibility, determine the relationship first, then choose the graphic. See Reference Study, Relationship Diagrams & Design Review.
Plan Slides First, Then Search for Missing Material
- Write a slide-by-slide checklist. For each slide, note: "What should the audience take away?", "What original text, data, or fact supports it?", "What subject does the image need?", "Image placement and cropping?", "Source and speaker notes supplement." First identify slides that need to reveal, transition, and close, then arrange the regular evidence slides.
- Prioritize raw statistics, surveys, or authoritative primary sources provided by the user. Record the year, region, unit, sample size, and comparison metrics. Verify policies and figures using official documents or original reports. Verify quotations using the original text, a reliable edition, or an archive. Reuse information already present in user materials directly. When external verification is needed, search using specific terms like object, location, date, or scene description.
- Assign a role to each image. For authentic cases, find on-site photos that can identify scenes, facilities, or actions, and prepare the original publication page and caption. For topic introductions, find environment or object photos directly connected to the argument. Source from Wikimedia Commons, archives, open-license stock libraries, or original publishers related to the topic as needed. Check usage conditions for each item, and record the author, original link, license, and any cropping or color adjustments.
- First decide what the image is saying, how much space it occupies, and where text goes; then combine "subject or action + location or time + aspect ratio / whitespace / perspective" to search. Prioritize test-cropping one image before collecting batches of similar material. After placing, retain the original aspect ratio and check if the actual crop preserves key people, objects, and text. See Image Search & Test Cropping.
- Place traceable information into the corresponding slide's caption, footer, or speaker notes. Captions explain identifiable facts in the image; speaker notes handle calculation processes, methodology, and verbal cues. Indicate the scope of simulations, scenario illustrations, estimates, and grouped conclusions as needed.
Organize Layouts & Graphics
- First, define the visual focal point, shared alignment guides, and reserved whitespace area. Standard pages can use column kickers, main titles, and an evidence area. Use kickers only when they provide orientation. Kickers and main titles align along the same left text edge; numbers, separators, and text align visually. Adjust column width, shorten text, or split slides based on content during execution. See Column Kickers.
- Choose composition based on the image's task. Scenes can use full-bleed stills or a horizontal band at the top of the page; details can use a large image with a close-up; multiple pieces of evidence can use a main image with smaller ones or an equal-weight gallery; key objects can be centered with whitespace. When arranging adjacent image slides, compare image proportion, position, and viewing distance. Reuse the same image frame only when side-by-side comparison or cross-slide continuity is needed. See Final Product Breakdown & Image Composition.
- Font size adjusts based on the page's task, whitespace, and viewing distance. First reserve page margins and spacing between groups, then assign font size, weight, and lightness for titles, key figures, explanations, auxiliary labels, and sources. Adjust gray text size by its information role; if supplementary explanations take up too much space, shorten them first before reducing the font size. Cover pages, quotes, and data pages can have a wider size span; metrics within the same group should maintain order. Choose Chinese title fonts based on the expression task and check must-read text, font substitution, and line breaks at the actual playback size. See Chinese Typography & Playback Layout.
- Use moderate corner rounding to soften bounded content. Apply primarily to independent photos, quote backgrounds, and areas that need grouping. Full-bleed and edge-bleeding images follow the canvas edge; precise bar charts retain their true endpoints. See Corner Rounding.
- Pick colors based on the topic and imagery. On each page, first establish the hierarchy of titles, keywords, summaries, and paraphrases, then differentiate them using lightness, font weight, whitespace, and localized colored blocks. For chapters, topic overviews, and colored-background transitions, use obvious contrasting colors and three to four meaningful theme colors. Maintain consistent colors with consistent meanings for data, and accompany them with text labels. See Color & Contrast and Hand-drawn Lines & Flat Vector Illustrations.
- Prioritize graphics suitable for the reading task. Use large numbers with units for key values, place actual values and standard ranges on the same scale, use horizontal bars with direct value labels for categorical sorting, specify the number of people and denominator for real subsets, use small-segment donut charts or stacked bars for composition, and use tables for item-by-item comparison. For timelines, flows, and routes, draw the main line first, attach nodes in sequence, and place labels close to their nodes. Use clear connections or start/end indicators when lines cross. See Continuous Paths & Timelines.
- Schematic illustrations default to code-drawn graphics. First draw recognizable outlines, then use a single set of continuous curves to express the main action, and use a small number of flat colored blocks to highlight key points. The hand-drawn feel comes from slight asymmetry, length variation, and natural stroke endings. Functional props retain their key structures; text and state changes are independently editable. Can use native PPT shapes or independent SVGs. Use traceable original images for real events and places. See General Illustration Workflow and Functional Prop Drawing.
- Presentation tables default to open borders. First organize reading via column alignment, font weight, and row spacing. Retain table header underlines, a few horizontal lines, or a light background when grouping is needed. Add lines as necessary for dense lookup tables, merged headers, or cross-referencing matrices. Retain native table format. See Table Lines & Whitespace.
- Default delivery includes native slide transitions and moderate entrance animations. For standard page turns, prioritize short fades. For key pages, choose fade-in, wipe, synchronous entrance, or morph based on whether the audience is seeing new content, reading along a line, comparing items, or observing a state change. Give at least one opening or key paragraph an object entrance triggered by the content. Set the sequence according to the script, action, and pauses. Follow user requirements for static or reduced animation requests. See Animation Selection.
- When delivering a PPTX file, ensure that titles, body text, numbers, and citations retain text editability. Prioritize native objects for data tables and standard charts. Calculate dimensions for custom data shapes from the same set of values. Place text and charts separately from photos. Choose the delivery format and tool based on the user's current requirements.
Review the Presentation Before Delivery
Scan thumbnails to verify the tone and composition rhythm. Check if adjacent image slides intentionally vary or maintain a fixed image position based on the narrative need. Then, on the exported file, check cropping, actual font sizes, truncation, font substitution, captions, contrast, and numerical metrics. Use slide show mode to test at least one standard transition, one key entrance animation, and all pages with media or special triggers. When the tool only supports static rendering, check the PPTX file for transitions and timelines, and state the scope of live playback verification. The final page should provide a clear response to the opening question, object, character, or figure.
When reviewing the exported file, first check from the audience's first impression and reading order, then check projection readability, actual glyph spacing, alignment, and line endpoints. Check large numbers with units, the last line and footer, and text in adjacent columns one by one. Ensure visible gaps between glyph edges. Record issues with the page number, visual evidence, and the fix method; after making changes, re-check the affected pages and their neighbors.