GPT-Image-1 - Good-vs-bad SOP comparison visual

Side-by-side two-column comparison card: messy SOP on the left, structured SOP on the right - educational explainer.

AU · Australia

**TARGET MODEL: ChatGPT Images 2.0 (gpt-image-2, April 2026 release).** Use 2.0-specific capabilities: - Embed exact text strings — 99% accuracy English, 90%+ Hindi/Punjabi/Bengali/Arabic/CJK. Spell every word verbatim in the prompt; the model will render it as typed. - Let 2.0 plan composition first (the model uses an internal reasoning step). Specify spatial layout in plain language ("primary subject left-third, accent right-third, headline top-center"). - Request 2K (2048x2048 or 2048x1152) when the output is for web hero / OG / LinkedIn — use 1024px square only for thumbnails. - For carousel / slide-deck use cases, ask for a coherent SET of up to 8 images in one call. - If multilingual diaspora copy is needed (Punjabi for Brampton, Hindi for North-India market, Tamil for SG diaspora), specify the EXACT script + transliteration to lock the rendering. --- Generate a comparison card at [ASPECT_RATIO] for "[OUTPUT_USE]". Campaign context: [CLIENT_CONTEXT]. Subject and layout: - Two side-by-side document cards on warm off-white background: - LEFT card label "Before": dense ragged body-copy lines, no headings, uneven margins. Topped with a small draft-orange #F28C28 warning dot. - RIGHT card label "After": clean structured body-copy lines, three visible section headings (abstract pill shapes in indigo #5E54FF), even margins. Topped with a small navy checkmark glyph. - Between the two cards, a single indigo arrow pointing right. Style: - Editorial flat-vector, monoline strokes, no shadows. - Composition: cards mirror each other in size, centered vertically, arrow centered between. Color palette: - Background: warm off-white #F6F4EF. - Card borders: navy #1E3A5F. - Before-card body lines: navy at 50 percent opacity, intentionally ragged. - After-card body lines: navy at 80 percent opacity, neat. - Section heading pills: indigo #5E54FF. - Warning dot: draft orange #F28C28. Typography hints: - Two literal labels: "Before" and "After" - render exactly as written, [BRAND_FONT_PRIMARY] sans small caps, navy. - No legible body copy. - [FIRM_LOGO_INCLUDED]: bottom-right wordmark at 50 percent opacity when yes. GPT-Image-1 directives: - Aspect ratio: [ASPECT_RATIO]. - Lighting: flat editorial neutral. - Mood: instructive, before/after clarity. Negative prompts: - No real-person likenesses. - No trademarked logos. - No legible SOP body copy. - No fabricated official letterheads or government correspondence layouts. - No violent or discriminatory imagery. - No red/gold or blue-dominant palettes. - No AI-art tells (warped card edges, melted labels, distorted arrow). DRAFT - for consultant/firm brand review. Verify generated image does NOT include trademarked logos, real-person likenesses, or any government-issued seal that could be mistaken for an official document. Brand colors must align with firm guidelines.