Nano Banana 2 AI Image Generator and Editor on Buzzy

Nano Banana 2 Image Creation on Buzzy

Create a new visual from a prompt or use a reference image to guide the subject, composition, text, and style. Refine the result in Buzzy, then carry a selected image into a video workflow when the concept needs motion.

More Buzzy Image and Video Tools

Compare this workflow with GPT Image 2, GPT Image 2.5, and Midjourney, or browse the current Buzzy models.

Nano Banana 2: Detailed Guide to Prompts, References, and Image Control

The fastest way to improve an AI image brief is to decide what the image must do before deciding how it should look. A product hero image needs a recognizable product, a controlled environment, and room for layout. A storyboard frame needs readable action and continuity with the next scene. A poster needs visual hierarchy and enough contrast for its text. When the task is clear, the prompt becomes a practical set of decisions instead of a pile of adjectives.

Nano Banana 2 can begin from text, a reference image, or a combination of both in Buzzy. Text is useful for describing the objective and the change you want. A reference can anchor identity, composition, color, product details, or an existing design language. The strongest brief explains the role of each input rather than assuming the model will infer it.

Choose the Right Starting Input

Use a text-only prompt when you are exploring a new visual direction and do not need to preserve an existing subject. State the subject first, then the action or pose, environment, composition, lighting, and intended use. This order helps separate the essential request from optional styling.

Use a reference image when visual continuity matters. A reference can show a person, product, sketch, packaging concept, room, or composition. Tell the model what to preserve and what to change. For example, keep the product shape and label placement, replace the background with a bright kitchen, and leave clean space on the right for campaign copy. That instruction provides a clearer hierarchy than “make this better.”

When several references are available, include only the ones that serve the same objective. Conflicting subjects, lighting, perspectives, or visual styles can make the request harder to interpret. A smaller, coherent reference set is usually easier to review and refine.

Write a Prompt That Can Be Evaluated

A useful image prompt gives you something specific to inspect. Begin with the primary subject and the visual task. Add the setting and composition. Then add lighting, palette, material, style, or camera language only where those details affect the result.

For a product visual, name the product details that must remain accurate. For a character sheet, repeat the defining appearance, clothing, and proportions. For a poster, provide the exact text and describe its hierarchy. For a diagram or infographic, define the information order and confirm that labels remain legible.

Avoid combining several incompatible requests in one pass. A photorealistic catalog image, hand-drawn storyboard, and dense infographic each require different visual priorities. Create the version that serves the immediate task, then develop related assets in separate prompts while preserving the important references.

Refine One Visual Decision at a Time

Prompt-based editing works best when each follow-up identifies the part that needs attention. If the subject is correct but the background is distracting, revise the background. If the composition is useful but the lighting feels flat, change the lighting. If the poster layout works but a line of text is wrong, correct the text and keep the rest stable.

This controlled approach makes comparison easier. You can see which instruction improved the result and keep the decisions that already work. Replacing the entire prompt after every generation makes it harder to preserve a successful subject, composition, or design direction.

Plan Text as Part of the Composition

Text inside an image needs more than correct spelling. Decide what the viewer should read first, how much copy the design can support, and where the text must sit relative to the subject. Short headlines are easier to inspect than dense paragraphs. Labels, prices, or calls to action should have enough contrast and clear spacing.

Enter critical copy exactly as it should appear and review it character by character. Multilingual text, product names, dates, measurements, and legal copy deserve an additional check. Even when the visual looks polished, generated text should not bypass normal proofreading and brand review.

Maintain Consistency Across an Image Series

Consistency comes from stable inputs and repeatable descriptions. Keep the same reference image when the same character or product should return. Reuse defining traits rather than rewriting them differently in every prompt. Preserve core lighting, palette, lens language, and composition rules when the assets belong to one campaign.

Not every image needs to be identical. A series can include a wide establishing image, a product close-up, a detail shot, and a social crop. The connection should come from shared identity and visual rules, while each image has a distinct job.

Review the Image for Its Final Placement

Judge a generated image in the format where it will appear. A vertical social asset needs a mobile-safe subject position. A landscape hero needs room for navigation and copy. A square product post needs a clear focal point at a smaller size. Choose the aspect ratio before finalizing the composition rather than relying on a later crop that removes important details.

Check subject accuracy, edges, hands, repeated details, product markings, text, lighting direction, and background logic. For commercial or public use, confirm that you have permission to use uploaded images, likenesses, logos, and protected materials. Buzzy provides the creative workflow; final factual, legal, and brand approval remains part of the publishing process.

Continue from Still Image to Motion

An approved image can become the key visual for a video concept. Before moving it into video, decide what should move and what should remain stable. Describe subject movement separately from camera movement. A slow push-in, a turn toward the camera, fabric moving in wind, or a product reveal each creates a different shot.

Use the still image to preserve the character, product, environment, or composition, then add only the motion needed for the scene. This gives the video workflow a clear visual starting point and keeps the next creative decision connected to the image you already approved.