Nano Banana Pro is an image generation and editing model built by Google DeepMind on Gemini 3 Pro, suited for tasks with high requirements for text readability, brand consistency, and compositional detail. It can transform text descriptions, product photos, and character references into posters, infographics, or scene composites, and adjust lighting, focus, and image elements through natural language.
Submit requests to the public API at api.acedata.cloud using the documented parameters, then use the results in your application.
api.acedata.cloudnano-banana-proInput parameters and result formats vary by service. Use the public API for this model and follow its guide for generation, task retrieval and editing operations.
The number of reference images and characters is within native capability limits; aspect ratio, generation count, and result format are set separately according to the selected endpoint.
The model emphasizes the readability and visual presentation of text within images, making it suitable for integrating headlines, product selling points, and descriptions into posters or packaging mockups. It can also convert multilingual content, allowing text to be adjusted together with the original design. Providing accurate copy, language, and typographic hierarchy is more helpful for controlling results than simply asking to “make an advertisement.”
It can combine references for people, products, and environments to generate scenes with unified lighting and visual relationships. Its strength is not just stitching assets together, but also maintaining consistency in character appearance and brand elements within complex compositions. Prompts should explain the purpose of each image and list the appearance, clothing, or product details that must be retained.
Changing daytime into night, shifting focus to a product, altering the shooting angle, or adjusting the color mood can all serve as clear editing goals. When making local changes, describing separately what “needs to change” and what “must remain unchanged” helps revise an existing visual direction rather than redesigning the entire image each time.
Input brand assets, campaign copy, and design direction to create key visuals with headlines and selling points, then adapt the text for different languages. Suitable for promotional tasks that need to balance image style and information delivery. Before delivery, focus on checking spelling, brand names, and text hierarchy, then choose a landscape or portrait format according to the placement.
Provide product photos or references for people and clothing, describe the target environment, camera, and lighting, and generate product scene changes, styling displays, or lifestyle assets. For product tasks, clearly specify that shape, color, and logos must be preserved; outfit results are suitable as visual concepts, while actual fit, materials, and wearing effects still need to be verified against the physical items.
Transform organized knowledge points, processes, or handwritten notes into infographics, diagrams, and storyboard drafts. Specify the audience, information order, and visual focus in the input so the model can organize the image around the content. Deliverables are suitable for instructional explanations or proposal communication; values and technical relationships should be reviewed item by item.
When a task involves substantial text, multiple reference assets, character consistency, or detailed photography controls, Nano Banana Pro is better suited to professional asset production. The original Nano Banana corresponds to Gemini 2.5 Flash Image and is geared more toward fast, easy editing; simple creative exploration does not need to pursue complex compositions from the start.
Nano Banana 2 corresponds to Gemini 3.1 Flash Image and is positioned as a balanced model that combines speed with general-purpose creation; Pro focuses on complex visual tasks, brand consistency, and fine control. Consider the former for high-frequency routine content, while key visuals with text-heavy layouts, complex asset relationships, or a need for careful adjustment are better suited to Pro.
Choose generate for text-based creation; choose edit to modify existing assets, provide reference images with image_urls, and specify separately what to retain and what to change.
Specify model=nano-banana-pro, action, and prompt for /nano-banana/images; start with aspect_ratio=1:1, resolution=2K, and count=1, setting the aspect ratio and resolution separately.
Retrieve the image from data[].image_url; for asynchronous requests, query or receive callbacks using task_id. When continuing edits, resubmit the selected image and narrow the scope of changes in each round.
Preserve the subject, brand colors, and composition of the reference poster, replace the title with the provided English copy, rearrange the text width, and maintain the lighting and product appearance of the original image.
Verify the localized copy, brand identity, and portrait details word by word; if the text placement is unsuitable, describe only the text area and layout in the next round.
They are different models: Pro corresponds to Gemini 3 Pro Image, while the original corresponds to Gemini 2.5 Flash Image. Pro focuses on complex compositions, text in images, and fine-grained control; when calling it, use nano-banana-pro, and the service name nano-banana does not mean the original model is being used.
When using /nano-banana/images, specify model as nano-banana-pro, action as generate, and select 2K or 4K through resolution. Set the image aspect ratio separately with aspect_ratio; resolution tiers and landscape or portrait composition are two different choices.
The dedicated endpoint uses action=edit, provides image addresses through image_urls, and describes the changes in prompt. You can also use /openai/images/edits, explicitly specifying the model and image. It is recommended to also state what to preserve, for example, preserve the product shape and change only the background and lighting.
Yes, you can continue editing an existing result, for example, first finalize the composition, then adjust the text or lighting. In practice, you can use the previous image as the reference for the next round and clearly specify what to change this time only. Saving satisfactory intermediate results helps prevent subsequent changes from affecting details that have already been confirmed.
The dedicated endpoint provides image_url in data and returns task_id and trace_id for associating tasks. Compatible image endpoints can return either an image URL or a Base64 result; when asynchronous processing is needed, use async and callback_url to integrate completed results into asset management or content workflows.
Choose a development approach or application suited to this model and follow its setup guide.
↗Python SDK Integration Guide
SDK installation, authentication and examples↗TypeScript SDK Integration Guide
SDK installation, authentication and examples
↗
↗
↗This is only a basic call example. See the full docs for more parameters and advanced usage.
View Docs