Verified against Nano Banana / Gemini 3.1 Flash Image · 2026-08-01
Design a poster or flyer with text that actually renders legibly
A poster brief that specifies exact text content and hierarchy word-for-word, built around Nano Banana's native multimodal architecture rendering in-image text more reliably than pure diffusion-based image models.
The prompt
Ready to copy — highlighted parts are example details you can swap.
Act as a graphic designer producing a finished poster or flyer. Every piece of text specified below must render exactly as written, spelled correctly, and legible at the sizes implied by the layout hierarchy — this image's success is judged on the text being readable and accurate, not just the overall visual composition. PURPOSE a flyer for a neighborhood weekend farmers market, to be posted on community boards and shared as a social image. EXACT TEXT CONTENT Render this text, verbatim, with no spelling changes, no rewording, and no omitted lines: Headline: "RIVERSIDE FARMERS MARKET". Subhead: "Every Saturday, 8am–1pm". Body: "Local produce, fresh bread, and live music at Riverside Park." Footer: "riversidemarket.example". If any line is genuinely too long to fit legibly at its intended hierarchy level within the layout, say so explicitly and suggest a shorter alternative rather than silently truncating, misspelling, or rendering it illegibly small just to force it to fit. LAYOUT AND HIERARCHY headline centered near the top, large and bold; subhead directly below it, smaller; body text centered in the lower-middle area; footer small at the very bottom. The most important line — usually the headline — should be the largest and highest-contrast text on the page; supporting details should be clearly secondary in size and weight, following a hierarchy a viewer's eye can parse in the right order at a glance, from three feet away for a flyer or from further for a poster meant to be seen from across a room. VISUAL STYLE a warm, hand-illustrated style with a loose watercolor texture, showing a simple illustration of vegetables and bread in a basket. COLOR AND TYPOGRAPHY warm earthy tones — terracotta, cream, sage green — with a friendly rounded sans-serif for the headline and a simpler serif for body text. Choose a typeface style and color combination that keeps every line of text at a contrast ratio genuinely readable against its background — don't place light-colored text over a busy, similarly light-toned area of the background image, and don't let a decorative typeface choice compromise legibility for the sake of style; the text has to actually be readable, not just look like a real font style. DIMENSIONS portrait 18x24 inch poster proportions. TEXT INTEGRITY CHECK Before finalizing, re-read every word you've rendered in the image against the exact text content specified above, character by character — check for any letter substitution, any dropped word, any spacing that's merged two words together, or any duplicated line. This step matters specifically because getting the visual style right while getting even one word wrong makes the whole poster unusable for its actual purpose. OUTPUT One finished poster or flyer image, with every specified line of text present, correctly spelled, and legible at its intended size within the layout — plus a note on any text you had to adjust for space and why.
Customize
Optional — swap in your own details for the highlighted parts above.
Why this works
Nano Banana's underlying model reads and generates text as part of the same multimodal token space it uses for images, rather than treating text-in-image rendering as a purely visual pattern to approximate the way earlier pure-diffusion image models did — which is the documented, concrete reason Gemini's image models handle in-image text more reliably than models with no native language grounding, and it's exactly why specifying exact text content verbatim, rather than a loose description of what the poster should say, gets meaningfully better spelling and word accuracy here than the same instruction would on a diffusion-only competitor. Second, the explicit permission to flag text that's too long to fit legibly, rather than silently shrinking or truncating it, matters because a poster's core job is communicating specific information — a date, a time, a URL — and a model under pressure to fit everything into a fixed layout will sometimes resolve that tension by rendering the overflow line at an illegibly small size or quietly dropping a word, either of which produces an image that looks complete at a glance but has actually failed at its one real job; asking for the conflict to be surfaced instead of hidden turns a silent failure into an actionable one. Third, the text-integrity check — re-reading every rendered word character by character against the source content — targets the single most damaging failure mode specific to AI-generated text-in-image content: a poster can be beautifully composed, perfectly on-brand, and completely unusable because one word has a dropped letter or two words have merged together, and because that kind of error is easy to miss on a quick visual scan of an otherwise polished-looking image, building a deliberate verification step into the generation instructions catches exactly the class of mistake a purely aesthetic review would sail right past.
Verified against
Nano Banana / Gemini 3.1 Flash Image Gemini 3.1 Flash Image · 2026-08-01
Changelog
- 2026-08-01 — Initial publish, verified against Nano Banana (Gemini 3.1 Flash Image) on a four-line farmers-market flyer with verbatim text accuracy checked.
Need this built into your business?
If a prompt isn't enough — what Scult builds, built and maintained for you — that's Scult's day job.
EXPLORE WHAT SCULT BUILDS
