Nano Banana 2 Workflow for Aligning Mixed-Script Text Baselines in Composite Advertisements
Creating composite advertisements that feature mixed-script text, such as combining Latin characters with CJK (Chinese, Japanese, Korean) or Arabic script, often results in visual discord. Different writing systems naturally possess varying baseline heights and x-heights, which can make a design look unprofessional if not corrected. While standard image generation tools might struggle to maintain the integrity of existing text while altering its position, Nano Banana 2 offers a workflow designed for iterative refinement. This approach allows users to address specific alignment issues through a series of targeted edits rather than attempting a perfect result in a single pass.
It is important to clarify that Nano Banana refers to the AI image generation and editing tool available on this platform, distinct from any skincare brand or physical product. The tool supports both text-to-image and image-to-image workflows, making it suitable for refining existing assets. However, users must understand that prompt instructions describe desired outcomes but do not guarantee the preservation of identity, labels, objects, or specific typography. Therefore, this workflow focuses on the visual alignment process, acknowledging that text content may require manual verification after generation.
Selecting the Right Model for Sequential Tasks
The success of a multi-turn editing process relies heavily on selecting the correct model variant. Google documents Nano Banana 2 as Gemini 3.1 Flash Image, Nano Banana Pro as Gemini 3 Pro Image, and Nano Banana 2 Lite as Gemini 3.1 Flash Lite Image. These are distinct models with specific capabilities tailored to different use cases.
For a workflow requiring multiple reference inputs or sequential editing steps, Nano Banana 2 (Gemini 3.1 Flash Image) or Nano Banana Pro (Gemini 3 Pro Image) are the appropriate choices. In contrast, Google describes Nano Banana 2 Lite as focused on speed and cost efficiency. It is explicitly noted that the Lite version is not optimized for multiple reference inputs or multi-turn sequential editing. Consequently, recommending Nano Banana 2 Lite for complex alignment tasks would be inappropriate without explaining this significant limitation. Users aiming to refine text baselines over several iterations should avoid the Lite model to ensure consistent context retention between turns.
Step-by-Step Alignment Workflow
To achieve visually aligned baselines in a composite advertisement, follow this structured input and output process. This workflow assumes you have an initial image containing mixed scripts that require adjustment.
Inputs and Preparation
Begin by uploading your source image containing the mixed-script text to the Nano Banana 2 interface. Ensure the image is clear and the text is legible. You will need to prepare a specific prompt that addresses the baseline discrepancy without demanding the AI rewrite the text entirely. Since the tool does not guarantee typography preservation, the prompt should focus on spatial adjustments.
Usable Prompt Example: "Adjust the vertical positioning of the secondary text line so that its baseline aligns perfectly with the primary English text line. Maintain the original font style and color as much as possible, but prioritize visual alignment of the bottom edges of the characters."
Note: This is an example prompt illustrating the desired outcome. It does not guarantee that the specific text content or font will remain unchanged.
Checkpoints During Iteration
After generating the first edit, perform a visual checkpoint. Zoom in to compare the baseline of the mixed script against the primary script. If the alignment is still off, proceed to the next turn. Do not assume the first attempt was successful. Use the output image from the previous turn as the new input image for the next iteration. This creates a feedback loop where the AI refines the position based on the previous result.
In subsequent turns, refine your prompt to be more specific about the remaining gap. For instance: "Shift the second line down by approximately 5 pixels to match the baseline of the top line exactly." Continue this process until the visual alignment meets your standards. Remember that each turn builds upon the previous one, so consistency in the prompt's intent is key.
Exporting and Finalizing Your Design
Once the baselines appear visually aligned across all scripts, you are ready to export the final image. Navigate to the download options within the Nano Banana 2 interface to save your high-resolution composite. Before using the image in a live advertisement, conduct a final review. Because the tool does not guarantee label or object preservation, verify that no unintended artifacts or text distortions occurred during the alignment process.
If the result requires further fine-tuning that exceeds the current workflow, consider exporting the image and using traditional graphic design software for pixel-perfect adjustments. This hybrid approach leverages the generative power of Nano Banana 2 for rapid layout correction while maintaining professional control over the final typography.
By following this structured approach, users can effectively manage the complexities of mixed-script typography in advertising. The key lies in understanding the model's limitations regarding text preservation and utilizing the multi-turn capability to iteratively refine the visual composition.