Integrating External References into Nano Banana 2 Standard Generation

Creating unique visuals often requires a blend of creative text prompts and specific visual guidance. When you need the AI to adhere to a particular style, composition, or object layout found in an existing image, integrating external references becomes essential. This tutorial explains how to effectively use external reference images within the Nano Banana 2 standard generation workflow to achieve precise visual outcomes.

Understanding Reference Integration Capabilities

Nano Banana 2 supports both text-to-image and image-to-image workflows, allowing users to upload external files to serve as visual anchors. When you provide an external reference, the system uses it alongside your text prompt to interpret the desired outcome. It is important to understand that while these references guide the generation process, they do not guarantee the preservation of specific identities, labels, objects, or typography from the source image.

The tool operates on distinct Google models. For instance, Nano Banana 2 utilizes Gemini 3.1 Flash Image, while Nano Banana Pro uses Gemini 3 Pro Image. These are separate entities with different strengths. A critical distinction exists regarding the Lite version. Google describes Nano Banana 2 Lite as focused on speed and cost efficiency. However, this model is not optimized for multiple reference inputs or multi-turn sequential editing. If your workflow relies heavily on complex reference integration, relying on the Lite version without acknowledging its limitations may lead to suboptimal results. Always verify which model you are utilizing before starting a complex project involving multiple visual guides.

Step-by-Step Upload and Instruction Methods

To successfully integrate external references, follow a structured approach to ensure the AI interprets your intent correctly. The process involves selecting the right file, uploading it, and crafting a prompt that bridges the gap between the visual input and your textual description.

1. Select Your Model: Navigate to the generator interface and choose the appropriate model. If you require robust handling of reference data, consider Nano Banana 2 or Nano Banana Pro rather than the Lite version, which has noted limitations with multiple inputs.
2. Upload the Reference Image: Locate the image upload section within the standard generation panel. Select the external reference file you wish to use. Ensure the file format is supported by the platform. Remember that Nano Banana refers to the AI tool itself, not any physical cosmetic brand or product depicted in your uploaded images.
3. Craft the Prompt: Write a clear instruction describing what you want the AI to generate based on the reference. Use the prompt instructions to describe the desired outcome, such as "recreate this composition but change the lighting" or "use this texture for a new background." Do not assume the AI will copy the image exactly; the prompt must explicitly state the transformation or adaptation you seek.
4. Generate and Review: Submit the request and review the output. Compare the result against your original reference to see how well the AI integrated the visual cues with your text instructions.

For those looking to experiment with different scenarios, here is an example prompt structure: "Use the attached image as a composition guide. Generate a new scene with similar framing but apply a cyberpunk color palette and replace the central figure with a robot."

Try Nano Banana

Judging Results and Troubleshooting Common Issues

Evaluating the success of your reference integration requires a critical eye. Since prompt instructions do not guarantee identity or object preservation, you should look for alignment in style, mood, and structural composition rather than exact replication. If the generated image fails to capture the essence of your reference, consider the following adjustments.

First, check your prompt clarity. Vague instructions can lead to the AI ignoring the reference image entirely. Be specific about which elements of the reference you want to retain and which you want to alter. Second, verify your model choice. If you attempted to use multiple references or perform sequential edits, you may have inadvertently used the Lite version, which is not optimized for these tasks. Switching to Nano Banana 2 or Pro might resolve the issue.

If the output looks unrelated to your input, try simplifying the prompt. Sometimes, too many conflicting instructions confuse the model. Focus on one primary attribute from the reference image, such as the color scheme or the angle of view, and build the rest of the prompt around that single anchor point. Additionally, remember that the AI generates new content based on patterns; it does not simply paste pixels from your reference. Expect variations and iterate until the visual target is met.

By understanding the capabilities and limitations of the underlying models and refining your instructional methods, you can effectively leverage external references to enhance your standard generation workflow. Whether you are aiming for stylistic consistency or compositional accuracy, careful planning yields the best results.

For further details on the technical specifications of the models mentioned, you may refer to the official documentation at Google Gemini image generation documentation. This resource provides verified information on the model families and their intended use cases.