GPT Image 2 generation playbook: Infographics through slides

GPT Image 2 is not only photoreal—infographics, packaging translation, ads, logos, UI mockups, science art, charts, and slides all land here.

Series：(I) Model selection & parameters · (II) Prompt basics · (III) Advanced prompt techniques · (V) Editing use cases · (VI) Character consistency

4.1 Infographics

Fit: distill structured knowledge—learners through exec decks.

Prompting: declare audience, grid, headings, arrows, hierarchy; raise quality when type is microscopic.

4.2 In-image translation

Localize lettering on packaging, screenshots, signage.

Rule: rewrite glyphs only—lock alignment, strokes, spacing, icons, subjects. Transliterate faithfully; forbid extra slogans.

4.3 Natural photorealism

Describe a decisive photographic instant: lens cues, choreography, sculpted light—plus believable imperfections (grain, fraying, moisture) minus plastic skin.

4.4 World knowledge

Hint “Bethel, NY—August ’69” and the model may infer Woodstock. When facts matter, reinforce them—hallucinations still happen.

4.5 Logo generation

Articulate temperament, substrates, scalability, trapping. Mention bold silhouettes & negative-space balance. Sweep variants via parameter n.

4.6 Ad stills

Write a creative brief: audience tension, insight, palette, casting, headline copy inside quotes if it belongs in frame.

4.7 Story → comic storyboard

Segment into visual beats—one panel = one escalation. Narrate blocking, emotion, transitions across rows.

4.8 UI mockup

Talk like a shipped product: nav, states, rhythm, trustworthy components—not decorative noise.

4.9 Science & education visuals

Audience + learning outcome + schematic grammar + regulated labels.

Default: flat icon system, readable arrows, breathing room. Use quality=high when annotations pile up.

4.10 Slides / charts / productivity art

Treat as spec: deck aspect, layer stack, honest numbers/strings, footnotes embedded in prompt text itself.