Nano Banana 2.1
Nano Banana 2.1 for clearer detail and design
Nano Banana 2.1 brings clearer typography, more natural image detail and steadier subjects to AI image generation and editing. Build a visual from a brief, then refine its composition, materials and lettering.
1K–4K
Model output resolutions
Text + image
Generation and editing inputs
2.1
Updated Nano Banana model
Generate professional-quality AI images at Flash speed with Nano Banana 2.
Web Grounding
Use Google Search for real-time web grounding
Aspect Ratio
Output Format
Resolution
Sign in to view your generated images
Your generation history will appear here once you start creating images
Nano Banana 2.1 is an AI image generation and conversational editing model that updates Nano Banana 2. Its strengths center on visual quality, instruction following, text rendering and character consistency through successive edits.
A useful image often needs several decisions to work together: the right subject, a clear focal point, convincing light and space for the message. This model combines a written brief with reference images, so a product shape, a portrait or a layout can guide the result. It supports 1K, 2K and 4K output, while its wide-format improvements help with panoramic compositions. Think in terms of an image you can develop: establish the scene first, then refine a specific element without rewriting the whole creative direction.
A portrait, a product and a setting need a shared visual direction.
For a campaign image, describe how the person and product belong in the same frame. A warm window scene can connect skin, fabric and glass through one lighting choice. Give the reference portrait responsibility for identity, and the product image responsibility for packaging. Reserve a quieter part of the frame for a headline. These creative choices make the brief easier to interpret and help you judge an edit by what changed, rather than by whether the entire image simply looks different.
Capabilities for images that need to communicate a specific idea.
The update improves image realism and overall visual quality. In product photography, that makes lighting, surfaces and small details useful parts of the brief. Name the finish you want: frosted glass, woven cotton or matte ceramic. Keep the light direction consistent so the material reads clearly. A tightly framed image can emphasize a texture; a wider view can show how the object sits in its environment.
Improved text rendering and infographic composition support images where words have a clear role. Supply the exact headline, then describe its position and relationship to the subject. A poster benefits from a readable title, a short supporting line and enough space between them. Treat small labels separately from the main message, and inspect the final spelling before using the graphic in a campaign.
Multi-turn editing places more emphasis on keeping a subject recognizable as the scene develops. For a portrait series, retain the same reference and describe the requested change narrowly: a new background, a different jacket or softer light. State which facial and hairstyle details should remain. This gives a series a coherent identity while allowing variation in pose, crop and atmosphere.
Multiple reference images can contribute different visual information to one composition. A product photograph may define the bottle, a room image may define the setting, and a style board may establish color. Explain those roles in ordinary language instead of assuming every reference has the same purpose. Choose clear views of important details and remove images that contradict the direction you want.
The 2.1 update addresses tiling artifacts in extreme portrait and panoramic formats at 2K and 4K. This is relevant to wide compositions such as display banners, long landscape scenes and tall editorial artwork. Place the main subject deliberately, then describe what occupies the surrounding space. A consistent horizon and restrained background can make a panoramic image feel like a single scene.
Use a source image and a written instruction to develop a visual through successive changes. Describe both the edit and the elements that should stay: replace the wall color while keeping the camera angle, or simplify the props while retaining the product. Working on one decision at a time makes it easier to compare versions and choose the direction that best serves the original brief.
Use the EIMG AI workspace with the model shown in its selector.
Open the workspace and select text-to-image for a new scene or image editing for an existing visual. Check the visible model selection before you begin. Decide whether your image is mainly a portrait, a product photograph or a graphic layout.
Describe the subject, setting, lighting and composition. For editing, upload a clear reference and name the change you want. Choose the available aspect ratio and resolution to suit the final placement, with room for any text or crop.
Inspect the result at the size where people will see it. Check important product details, faces and lettering, then use a focused follow-up brief for the next version. Download the selected image through the workspace when it meets your needs.
| Capability | Nano Banana 2 | Nano Banana 2.1 |
|---|---|---|
| Main role | Efficient generation and editing | Updated generation and editing |
| Image detail | Existing visual-quality baseline | Improved realism and visual quality |
| Typography | Text in generated images | Improved text and infographic layouts |
| Sequential edits | Conversational refinement | Improved subject consistency |
| Output sizes | 1K, 2K and 4K | 1K, 2K and 4K |
| Extreme formats | Wide-format generation | Tiling correction at 2K and 4K |
Start with the subject and the purpose of the image. Add a setting, a camera position and one coherent lighting direction. If lettering matters, quote the exact text and describe its size and placement. For reference images, assign each a role and identify the details to retain. A focused edit should say what changes and what remains; a long list of competing styles makes the desired result harder to judge.
Create a wide product campaign image using the uploaded bottle as the packaging reference. Keep its ivory color, proportions and black label. Place it on a pale stone shelf with warm light from the left. Leave open space on the right for the headline “Quiet mornings”, with a small line below reading “Daily care”. Use a restrained editorial composition and realistic material detail.
Keep the bottle, headline and camera angle. Replace the stone shelf with brushed metal and make the background a warmer cream. Retain the same light direction.
Develop packaging scenes with a consistent product shape, a considered material palette and a clear message. Use the same visual direction across a hero image, a square social crop and a vertical campaign concept.
Combine a short headline with an image or a small set of explanatory elements. Establish reading order before adding decorative detail, so the graphic remains useful at the size where it will be shared.
Build a sequence around a recognizable person or character. Keep a stable reference, then vary the environment and styling through specific edits. Compare facial details and silhouette across the chosen images.
The update concentrates on image quality, instruction following, text rendering and consistency through multiple edits. It also addresses tiling artifacts in extreme aspect ratios at 2K and 4K.
Yes. Text-to-image generation lets a brief describe the subject, setting and visual style. Include the composition and lighting when those choices are important to the image.
Yes. Image-and-text editing can change elements, style or color in an existing image. State which parts should stay intact so the edit has a clear scope.
The model supports 1K, 2K and 4K outputs. Choose an output size based on the image's intended use; the available options in EIMG AI follow the model currently selected in the workspace.
Use clear images that show the details you need to preserve. Explain whether each reference defines a person, an object, a setting or a visual style, and avoid contradictory examples.
Improved typography is one of the update's main capabilities. Provide short, exact text and a clear layout. Check spelling and legibility in the finished image before publishing it.
Keep the same identity reference and describe each change specifically. Mention stable features such as hairstyle and face shape; review the selected results together to judge continuity.
The visible model selector identifies the model used for generation. Choose an available model there before submitting your prompt, and check its displayed options and cost.
Prepare your brief, choose a model in EIMG AI and develop a visual through focused changes. Keep the subject and the message at the center of each version.
Open the workspace
