Mastering Depth of Field: Nano Banana 2 Image-to-Image Prompt Guide

Understanding Depth Adjustment in Post-Production

When working with images captured by fixed lenses, the depth of field is often locked at the moment of capture. However, modern AI tools allow creators to reinterpret these static scenes after the fact. The goal of adjusting depth of field in post-production is to direct the viewer's attention by manipulating what appears sharp and what remains blurred. A shallow depth of field isolates a subject against a creamy, out-of-focus background, while a deep depth of field keeps the entire scene from foreground to horizon in crisp detail.

Nano Banana 2 supports image-to-image workflows that enable this specific type of creative control. By providing an original image alongside a carefully constructed text prompt, users can instruct the system to alter the optical characteristics of the photo without changing the underlying composition or subject identity. It is important to remember that prompt instructions describe desired outcomes; they do not guarantee identity, label, object, or typography preservation. Therefore, when attempting to blur backgrounds or sharpen foregrounds, users should expect the AI to interpret the visual intent rather than strictly adhering to pixel-perfect replication of every element.

Structuring Your Prompt for Optical Effects

To achieve consistent results when adjusting depth of field, the structure of your prompt is critical. The prompt must clearly distinguish between the subject you want to keep sharp and the areas you wish to soften. A robust prompt structure typically begins with a description of the core subject, followed by specific directives regarding focus planes.

For example, if you are uploading a portrait taken with a wide-angle lens where everything is in focus, you might construct a prompt like: "Portrait of a person standing in a park, shallow depth of field, subject in sharp focus, background heavily blurred and bokeh." Conversely, for a landscape where you want to enhance the sense of scale, the prompt could be: "Wide landscape shot, deep depth of field, sharp focus from the immediate foreground rocks to the distant mountains, high clarity throughout."

These examples illustrate how to frame the request. The AI interprets terms like "shallow depth of field" or "deep depth of field" as instructions to modify the rendering of light and focus. Users can copy these structural patterns into the generator, but they must adapt the descriptive elements to match their specific input image. Remember that Nano Banana refers to the AI image generation/editing tool in these articles. It is not a skincare brand, bottle, jar, or physical subject. The tool processes the digital file provided, applying the requested optical changes based on the textual guidance.

Selecting the Right Model for Depth Tasks

Not all versions of the tool are equally suited for complex editing tasks like depth adjustment. Google documents Nano Banana 2 as Gemini 3.1 Flash Image (gemini-3.1-flash-image), Nano Banana Pro as Gemini 3 Pro Image (gemini-3-pro-image), and Nano Banana 2 Lite as Gemini 3.1 Flash Lite Image (gemini-3.1-flash-lite-image). These are distinct Google image models with different capabilities.

Google describes Nano Banana 2 Lite as focused on speed and cost. It is not optimized for multiple reference inputs or multi-turn sequential editing. Do not recommend it for those workflows without explaining this limitation. For precise depth of field adjustments, which often require nuanced understanding of spatial relationships, the standard Nano Banana 2 or Nano Banana Pro models are generally more reliable. The Lite version may struggle to maintain the integrity of the subject while blurring the background, potentially introducing artifacts or failing to apply the blur uniformly. Always verify the model capabilities before starting a project that requires fine-tuned optical simulation.

Judging Results and Fixing Common Issues

After generating an image, evaluating the success of the depth adjustment is straightforward. Look for a clear separation between the focal plane and the rest of the image. In a successful shallow depth of field edit, the subject should appear crisp while the background transitions smoothly into softness. If the transition looks jagged or if parts of the subject have become blurry unintentionally, the prompt may need refinement.

Common issues include over-blurring, where the background becomes indistinct to the point of losing context, or under-blurring, where the effect is barely noticeable. To fix over-blurring, try adding modifiers like "subtle background blur" or "moderate bokeh" to the prompt. If the subject loses definition, reinforce the instruction by explicitly stating "keep subject eyes and face in sharp focus." Another frequent issue is the introduction of strange textures in the blurred areas. This can sometimes be resolved by rephrasing the prompt to emphasize natural optical properties, such as "natural lens blur" or "photographic depth of field."

If the initial result is unsatisfactory, consider iterating with slight variations in the prompt wording. Since prompt instructions do not guarantee identity preservation, minor shifts in the generated output are normal. You may also need to adjust the strength of the image-to-image influence if the tool provides such a slider, ensuring the new depth effect blends seamlessly with the original lighting and shadows.

For those ready to experiment with these techniques, Try Nano Banana offers the necessary interface to upload images and test various prompt structures. Whether you are aiming for a cinematic look or a documentary style, mastering the prompt structure allows you to transform fixed-lens photography into dynamic visual stories.

By following these steps and understanding the limitations of each model, users can effectively simulate professional camera optics within the digital workspace. The key lies in clear communication with the AI through precise language and an iterative approach to refining the visual outcome.