Nano Banana Pro prompts — detail, text and scenes worth the price

At 124 credits / image, about $1.24 a take at the entry pack, Nano Banana Pro is the priciest image model on eroq, positioned as the highest-fidelity model of Google's Nano Banana line, with its sharpest detail, its most reliable lettering and its strongest grasp of long prompts. A prompt that justifies that price does at least one of three things. It carries detail someone will zoom into, it carries words that must read, or it stacks constraints that all have to hold at once. These 14 prompts are built on exactly those, and each comes with a line on why it needs Pro rather than something cheaper.

The three-question test

Before you spend, ask:

1. Will anyone look closely? Print, a crop for a banner, a hero image on a big screen.
2. Are there words that must read? A sign, a label, a title.
3. Does the prompt hold more than a few must-be-true clauses? Who stands where, holding what, wearing what.

Three no's mean the image belongs on Nano Banana 2.1 at 31 credits / image, or on another cheaper model. One yes is enough to make Pro worth it. The cost math between the two lives in Nano Banana Pro vs Nano Banana 2.1.

The layered prompt

Pro rewards specificity, and the layers go in this order: subject, then wardrobe or material, setting, light, lens, mood, and finally any text, in double quotes. Every layer you skip is a decision the model makes for you, which is a waste on the model you are paying the most for.

All five formats cost the same; pick the one the deliverable needs in the Format control.

Detail

1. Embroidery in progress (4:3). Every stitch is a detail a viewer will hunt for.

A macro photograph of hand embroidery on an indigo linen hoop, a half-finished golden sunflower in satin stitch, the needle mid-pull trailing silk thread, individual linen fibres visible, soft north window light, shallow depth of field at f/4, 100mm macro, quiet craft mood.

2. A luthier's bench (16:9). Spruce grain and maple shavings reward close inspection.

A close-up of a luthier's workbench: an unvarnished violin top with fine spruce grain, curls of maple shavings, a small brass finger plane, a pot of hide glue with a brush resting in it, late afternoon window light raking across the wood, 60mm macro, warm and patient.

3. Sourdough crumb (1:1). An open crumb and a crackled crust are texture-level detail.

An overhead photograph of a just-cut sourdough loaf on a flour-dusted walnut board, glossy irregular holes in the crumb, a blistered crackled crust, a butter knife carrying a curl of cultured butter, scattered crumbs, soft side light, 50mm, rustic and appetising.

4. Frozen waterfall with a climber (9:16). Ice texture up close and a tiny figure far away, both sharp.

A frozen waterfall in a narrow canyon at dawn, blue ice columns with trapped air bubbles, a lone ice climber in a red jacket halfway up for scale, frost feathering the canyon walls, a pale pink sky above, 35mm, crisp from foreground to background.

5. Beekeeper portrait (3:4). Skin, wax cells and moving bees all have to hold in one frame.

A close editorial portrait of a beekeeper in her sixties lifting a honeycomb frame, bees crawling over the wax cells, honey catching the light, fine lines around her eyes, a mesh veil pushed back on her hat, overcast soft light, 85mm at f/2.8.

Text

6. Bookshop sign and chalkboard (3:4). Two pieces of text in two lettering styles, in one scene.

A narrow bookshop on a rainy evening, a hand-painted wooden sign above the door reading "THE LAST CHAPTER", a pavement chalkboard reading "Poetry night", stacks of books glowing behind the window, rain beading on the glass, warm interior light against blue dusk, 35mm.

7. Cereal box lineup (16:9). One brand name, repeated identically across three surfaces.

Three cereal boxes standing in a row on a pale blue kitchen counter, front-facing, one strawberry red, one banana yellow, one cocoa brown, the same brand name "MORNING CRUNCH" across the top of each box, bright morning light, clean commercial photograph.

8. Botanical chart (3:4). A stress test with seven text elements; expect to run two takes and proofread.

A vintage botanical chart titled "EDIBLE MUSHROOMS", six hand-drawn mushrooms in two rows on aged cream paper, each with a small caption: "CHANTERELLE", "PORCINI", "MOREL", "OYSTER", "ENOKI", "SHIITAKE", fine ink linework with soft watercolour fills.

9. Garden welcome sign (3:4). Calligraphy, two lines and names that must be spelled right.

A welcome sign on a wooden easel in a garden at golden hour, white calligraphy on a dark green chalkboard reading "WELCOME" with "Ana & Teo" beneath it, eucalyptus garlands draped over the frame, guests blurred in the background, 50mm.

Complex scenes

10. Kitchen during service (16:9). Four people, four actions, one coherent space.

A busy restaurant kitchen during service, seen from the pass: a chef in the foreground plating three scallops on black slate, a cook behind her flambéing a pan in a burst of orange flame, a dishwasher at the far left in a cloud of steam, order tickets curling from the rail, hard overhead light, 24mm.

11. Chess endgame (4:3). Exact positions are the hardest kind of constraint; Pro is the model to try them on.

A chess endgame on a marble board photographed from a low angle: a white king on e1, a white rook on a7, a black king on e8, no other pieces on the board, a hand hovering over the rook, dramatic side light, 85mm.

12. Orchestra from the podium (16:9). Placement by section, with one named player in each.

A community orchestra rehearsal seen from the conductor's podium: violins on the left, a cellist in a green sweater front right, a timpani player raising mallets at the back, sheet music on every stand, afternoon light slanting through tall windows, 24mm.

13. Café diorama (1:1). Three floors, each with its own contents.

An isometric miniature diorama of a three-storey corner café: a baker sliding bread into an oven on the ground floor, a reading room with two armchairs on the first floor, a rooftop garden with tomato plants and a parked bicycle, warm lights in every window, tilt-shift look.

14. A study at night (4:3). Six named objects, every one of which has to appear.

A Victorian study at night: a green banker's lamp lit on a mahogany desk, an open atlas showing the coast of Norway, a brass telescope by the window, a tabby cat asleep on a stack of books, rain on the glass, embers glowing in the fireplace, 35mm, warm and still.

Getting your money's worth

• Run one or two takes, not four. A full batch on Pro is 496 credits. Once a prompt is proven, two takes (248 credits) usually give you a choice.
• Prove the composition first. Run the prompt once on Nano Banana 2.1, check the layout, then send the keeper to Pro (the numbers across the lineup).
• Proofread text takes. One wrong letter undoes the price.
• Failed renders refund themselves, and a declined prompt (content_blocked) is never billed.

For text-heavy work beyond these, how to get legible text in AI images goes deeper, and the Nano Banana Pro guide covers the model end to end. Paste a prompt into the image studio and pick Nano Banana Pro.

FAQ

What makes a good Nano Banana Pro prompt?

Layers and a reason. Subject, material or wardrobe, setting, light, lens, mood and any quoted text, plus at least one thing that needs Pro: fine detail, words that must read, or several constraints that must all hold.

How many takes should I run on Nano Banana Pro?

One or two once the composition is proven. Four takes cost 496 credits, so do your exploring on Nano Banana 2.1 at 31 credits / image.

Will these prompts work on Nano Banana 2.1?

Yes: 2.1 understands the same plain-sentence prompts. The detail, the lettering and the many-clause staging are what Pro adds, so judge 2.1 renders on layout rather than on those.

Can Nano Banana Pro use my product photo or a face reference?

No. On eroq it renders from text alone. When a product or face has to match real photos, Eroq One is the engine: it takes those photos as references, ten at most.