Natural-language image editing
Describe a background, outfit, color, pose, lighting, or object change without manually drawing a mask.
Create and edit images with the original Nano Banana model, Google Gemini 2.5 Flash Image, using text prompts and visual references in a focused online workspace.
Nano Banana is selected by default. You can still switch to Nano Banana 2, Nano Banana Pro, or another available image model.
Generated content will appear here
Facts reviewed:
Nano Banana is the public name for Google’s Gemini 2.5 Flash Image model. It was the first widely known model in the Nano Banana family and combines native text-and-image understanding with image generation and conversational editing.
Google positions Gemini 2.5 Flash Image for high-volume generation, low-latency creative workflows, and natural-language image editing. It can blend images, preserve a person or pet across edits, change selected details, transfer styles, and continue refining a visual over multiple turns.
Nano Banana is now the legacy model in Google’s current family. Google recommends newer models for new API workflows, but the original remains useful when a straightforward, efficient 1024px generation or editing workflow fits the task. This page clearly separates that provider positioning from the settings and point estimate exposed by this platform.
These capabilities come from Google documentation; workspace controls and limits belong to this platform.
Describe a background, outfit, color, pose, lighting, or object change without manually drawing a mask.
Guide recurring people, pets, or illustrated subjects across edits while reviewing faces and identity details.
Ask to alter one part of an image while keeping the camera angle, composition, and unrelated objects stable.
Combine subjects, objects, styles, or scenes from uploaded references into a new composition.
Continue from an earlier result with follow-up instructions instead of rewriting the complete brief each time.
Google documents native 1024px output for this model, suited to drafts, social visuals, and frequent iterations.
Apply an illustrated, photographic, cinematic, product, or material direction while preserving key source content.
Use words and images together so the model can interpret the scene before generating or editing.
These capabilities come from Google documentation; workspace controls and limits belong to this platform.
Try outfit, background, lighting, age, or art-style changes while preserving recognizable traits.
Move a product into a new lifestyle or studio setting and inspect shape, color, label, and proportions.
Describe which object should disappear or be substituted and what surrounding content must remain.
Explore fresh colors, props, crops, and visual styles from one base image.
Restyle rooms, change materials, or test furniture arrangements without treating the result as construction documentation.
Place a recurring subject into new scenes for concepts, storyboards, stickers, or figurine-style artwork.
For editing, state both the requested change and the elements that must remain unchanged.
Change the setting without redesigning the subject.
Keep the person’s face, hairstyle, pose, clothing, and camera angle unchanged. Replace only the background with a softly lit independent bookstore, add natural window light from the left, and preserve realistic skin texture.
Use a product reference and control its geometry.
Place the uploaded ceramic mug on a warm oak breakfast table beside a folded newspaper and one orange. Preserve the mug shape, glaze color, handle position, and printed mark. Morning lifestyle photography, 4:3 composition.
Retain identity while changing visual treatment.
Transform the uploaded pet photo into a hand-painted children’s book illustration. Keep the pet’s markings, eye color, ear shape, and expression recognizable. Add a small garden path and soft watercolor foliage; no text.
Change the mood and furnishings while preserving the room geometry.
Keep the room layout, window positions, camera angle, and wood floor unchanged. Restyle the interior as a calm contemporary reading room with an olive sofa, low oak shelving, warm linen curtains, afternoon sunlight, and a few ceramic objects. Remove clutter; do not alter the architecture.
A playful product-style concept with a clear material brief.
Transform the reference character into a small collectible vinyl figure while preserving the hairstyle, color palette, and recognizable outfit. Place it on a simple cream display plinth, soft studio lighting, subtle packaging box in the background, realistic molded material, square composition.
Start with text for a new visual, or upload one or more useful images for an edit or blend.
Describe the desired result and list identity, composition, product, or color details that must remain fixed.
Choose an available aspect ratio, confirm reference uploads, and check the latest point estimate before submitting.
Review faces, hands, text, object count, product geometry, and factual details before refining or downloading.
Google currently describes Nano Banana as the legacy Gemini 2.5 Flash Image model and recommends Nano Banana 2 Lite for newer API projects. Original Nano Banana generates at 1024px and does not support Search grounding or Thinking. Do not confuse it with the 4K-capable Nano Banana 2 or Nano Banana Pro.
This platform exposes its own aspect ratios, up to five reference uploads, live point estimate, and task availability. Those controls do not imply identical Google API pricing, quotas, latency, storage, or feature access.
Character consistency, one-shot editing, and scene preservation are not guarantees. Generated text, fine product details, identities, anatomy, and facts may drift. Google states that generated or edited images include SynthID; users must still verify rights and suitability before publication.
Model identity, lifecycle, and capabilities were checked against current Google documentation; platform controls come from the live workspace.
Compare Google’s newer general-purpose 4K Flash image model.
02Open the model page for complex professional asset production.
03Start from an existing image and describe a transformation.
04Learn a reusable structure for generation and editing prompts.
Return to the workspace with Nano Banana selected and review each result before publishing.