Maak een taak aan voor het genereren van een model.
Verzoek
curl-X POST 'https://pollo.ai/api/platform/v1/generation/google/nano-banana-2/image'\-H'Content-Type: application/json'\-H'x-api-key: YOUR_API_KEY'\-d'{ "input": { "prompt": "A cinematic shot of a golden retriever running through a field of sunflowers at sunset, warm rim light, shallow depth of field", "aspectRatio": "1:1", "resolution": "0.5K" }}'
Voeg optioneel een webhookUrl toe om een callback te ontvangen wanneer de taak is geslaagd of mislukt.
Optioneel aanvraagveld
{"input":{"prompt":"A cinematic shot of a golden retriever running through a field of sunflowers at sunset, warm rim light, shallow depth of field","aspectRatio":"1:1","resolution":"0.5K"},"webhookUrl":"https://example.com/webhooks/pollo"}
Schema
Invoervelden voor het geselecteerde eindpunt en de modus.
Veld
Type
Vereist
Beschrijving
prompt
string
Ja
Text prompt describing the content to generate.
aspectRatio
string
Nee
Output aspect ratio, as width:height (e.g. 16:9).
resolution
string
Nee
Output resolution of the generated media (e.g. 720p, 1080p, 2K, 4K). Values are case-insensitive.
Nano Banana 2 reasons before it renders. Chain-of-thought planning and real-time world knowledge mean prompts with logic, counting, and factual layout resolve correctly instead of guessing.
With Nano Banana 2 on Pollo API, developers add text-to-image, single-image editing, and multi-image fusion to their apps, generating up to native 4K across ten aspect ratios per request.
Key Features of Nano Banana 2 API
Reasoning-Driven Prompts
A chain-of-thought pass interprets complex instructions, so requests involving order, quantity, spatial relationships, or conditional logic render as described rather than approximated.
Grounded World Knowledge
Real-time knowledge lets the model place accurate details, from correct diagrams and labeled parts to plausible real-world objects, without inventing facts that break credibility.
Native 4K Ultra-HD
Output reaches native 4K, holding fine texture, small text, and dense detail at full resolution, so results stay sharp when zoomed, printed, or cropped.
Dependable Typography
Global typography renders cleanly, including longer strings and non-Latin scripts, making Nano Banana 2 suited to posters, packaging, and UI mockups where text must be legible.
Multi-Subject Consistency
Feeding multiple reference images keeps the same character, product, or object recognizable across variations, so identity stays stable through edits and composited scenes.
Extreme Aspect Ratios
Ten ratios span square through ultrawide 21:9 and tall 9:16, covering banners, story frames, and panoramic layouts without stretching or cropping the composition.
Use Cases of Nano Banana 2 API
Labeled Infographics: Generate charts, cutaway diagrams, and process flows where numbers, labels, and callouts read correctly thanks to reasoning and typography.
Packaging and Label Mockups: Produce product packaging with legible brand text, nutrition panels, or multilingual copy rendered directly into the artwork.
Consistent Character Sets: Keep one illustrated persona identical across poses and scenes for comics, mascots, or app onboarding illustrations.
Reference-Based Product Edits: Composite a supplied product image into new backgrounds while preserving shape, color, and material detail.
Ultrawide Environment Art: Build 21:9 landscape and game-style scenes with coherent perspective for backdrops and hero banners.
Educational Illustrations: Create biology, geography, or history visuals grounded in real-world knowledge so depicted facts stay accurate.
Localized Marketing Variants: Swap headline text across languages in the same layout, generating regional creatives from a single design idea.
Print-Ready Poster Art: Render 4K posters combining detailed imagery and typographic hierarchy suitable for large-format output.
How to Use Nano Banana 2 API
Get an API Key: Create a Pollo API account and generate your API key from the developer dashboard.
Choose the Model: Send requests to the Nano Banana 2 endpoint at /generation/google/nano-banana-2/image.
Add Your Inputs: Pass a prompt and aspectRatio for text-to-image, add imageUrl or images for editing and fusion, then set resolution (1K, 2K, 4K) and thinkingLevel.
Generate and Retrieve: Submit the task, poll the returned task status, and download the finished image once it succeeds.
Prompting Best Practices for Nano Banana 2 API
Write prompts that give the model something to reason about. State the exact text, counts, and spatial relationships you want, and let its world knowledge fill accurate detail around them.
A Simple Prompt Formula
Subject and scene + exact text or labels in quotes + layout and spatial relationships + style and lighting + aspect ratio and detail level
What You Should Notice
Quote Your Text: Put on-image words in quotation marks so typography renders exactly, including headings, labels, and multilingual copy.
Specify Counts And Order: Name how many items and their arrangement; the reasoning pass respects quantity and position.
Anchor Real Facts: For diagrams and educational art, describe the real subject so world knowledge grounds the details.
Supply Clean References: Use clear HTTPS reference images to lock character or product identity across edits.
Raise Detail Level For 4K: Choose higher resolution and a high thinking level when text density or fine detail matters.
Example Prompts
Labeled Science Diagram
"A cross-section diagram of a lithium-ion battery cell, four layers labeled 'anode', 'cathode', 'separator', 'electrolyte' with thin callout lines, clean textbook illustration style, white background, 4:3, sharp legible typography."
Multilingual Packaging
"A tea box mockup on a light studio surface, front panel reading 'Jasmine Green' in English and Japanese, small brewing instructions below, soft daylight, realistic product photography, square 1:1, crisp readable labels."
Consistent Character Across Scenes
"Using the reference image, place the same freckled red-haired girl reading under an oak tree, then waving at a train window, matching face and outfit, warm storybook illustration, gentle afternoon light, 3:2."
Ultrawide Game Environment
"A floating archipelago at dawn, stone bridges linking three islands, waterfalls spilling into cloud, distant airships, painterly concept-art style, cool blue and gold palette, ultrawide 21:9, high detail."
Nano Banana 2 vs GPT Image 2 vs Seedream 4.0
Capability
Nano Banana 2
GPT Image 2
Seedream 4.0
Reasoning-based prompt logic
✅ Chain-of-thought planning
✅ Strong instruction following
✅ Good adherence
Native max resolution
Up to native 4K
Up to 4K
Up to 4K
Typography accuracy
✅ Global scripts, dense text
✅ Reliable text
✅ Solid text
Multi-image fusion
✅ Multi-reference consistency
✅
✅
Aspect ratio range
10 ratios incl. 21:9
Common ratios
Wide range
Recommended For
Logic-heavy, text-rich, 4K images
General editing and rendering
Fast stylized generation
Why Choose Nano Banana 2 API?
Nano Banana 2 suits products where images must be correct, not just attractive, handling labeled diagrams, multilingual text, consistent characters, and 4K detail from a single model.
Through Pollo API, you reach Nano Banana 2 with one API key alongside 300+ leading image and video models, with clean docs, task status polling, and generation that costs less than Fal AI.
Integrate the Nano Banana 2 endpoint on Pollo API to ship reasoning-grounded, text-accurate 4K image features, then switch or compare models without touching your integration.
Nano Banana 2 API FAQs
What is Nano Banana 2?
Nano Banana 2 is Google's image model that uses chain-of-thought reasoning and real-time world knowledge to generate and edit accurate, text-rich images at native 4K resolution.
Does Nano Banana 2 API support image editing and fusion?
Yes. You can edit a single reference image or fuse multiple images, keeping characters and products consistent while composing new scenes and variations.
What resolutions and aspect ratios are available?
Nano Banana 2 outputs 1K, 2K, or native 4K across ten aspect ratios, from square and portrait to ultrawide 21:9 and tall 9:16 formats.
How accurate is its text rendering?
Typography is a core strength, including long strings and non-Latin scripts, which makes it dependable for posters, packaging, UI, and other text-heavy layouts.
What does the thinking level do?
It controls how much reasoning the model applies before rendering, helping complex prompts with counts, spatial logic, or dense detail resolve more reliably.
Why run Nano Banana 2 through Pollo API?
Pollo API gives one integration for Nano Banana 2 and 300+ other models, with clean documentation, task tracking, and lower-cost generation than comparable providers.
