Nano Banana 2 Workflow for Creating a Before-and-After Comparison Postcard Layout
Creating a compelling before-and-after comparison postcard requires visualizing how a location has changed over time. This process involves generating two distinct images: one representing the historical state and another depicting the current appearance. Once these assets are ready, they must be combined into a single split-screen layout using an external design tool. This workflow leverages the text-to-image capabilities of Nano Banana 2 to produce high-quality source material without needing complex multi-turn editing within the interface itself.
Step 1: Define Inputs and Craft Your Prompts
The foundation of this workflow lies in precise input definition. You must clearly articulate the specific location, the era you wish to depict, and the desired visual style for both the past and present versions. Since Nano Banana names the AI image generation tool and not a physical product or brand, your prompts should focus entirely on the architectural and environmental details of the scene.
Start by gathering reference data about the location. Identify key landmarks that have remained constant versus those that have changed. For the "before" image, describe the historical context, such as "early 1900s street scene," "vintage architecture," or "sepia-toned atmosphere." For the "after" image, specify modern elements like "contemporary glass buildings," "modern traffic," or "daylight photography."
You can utilize the prompt library available on the Nano Banana 2 page at /nanobanana2 to find example structures. However, remember that prompt instructions describe desired outcomes; they do not guarantee identity, label, object, or typography preservation. Therefore, treat any generated text or specific signage as illustrative rather than factual.
Example Prompt Structure (Untested):
- Past Image: "A bustling city square in 1920, vintage horse-drawn carriages, cobblestone streets, sepia tone, historical architecture, wide angle shot."
- Current Image: "The same city square today, modern skyscrapers, electric buses, paved roads, bright daylight, photorealistic, wide angle shot."
These examples demonstrate the contrast needed but are untested for specific location accuracy. Users should replace generic terms with their specific location details to achieve the best results.
Step 2: Generate Images Using the Appropriate Model
Once your prompts are refined, proceed to the generation phase. The website supports text-to-image workflows through Nano Banana 2. It is crucial to select the correct model based on your needs. Google documents Nano Banana 2 as Gemini 3.1 Flash Image (gemini-3.1-flash-image), while Nano Banana Pro corresponds to Gemini 3 Pro Image (gemini-3-pro-image).
For this workflow, which requires generating two separate images sequentially, standard Nano Banana 2 or Nano Banana Pro models are recommended. Do not recommend Nano Banana 2 Lite for this specific task without noting its limitations. Google describes Nano Banana 2 Lite as focused on speed and cost, and it is not optimized for multiple reference inputs or multi-turn sequential editing. Using the Lite version might result in inconsistencies between the two generated images if you attempt to maintain strict spatial alignment across generations.
Generate the "past" image first, ensuring the composition matches the intended final layout. Then, generate the "current" image using the same camera angle and perspective description to facilitate easier alignment later. While the tool does not guarantee perfect alignment, maintaining consistent descriptive keywords for the background and horizon line will help.
Step 3: Checkpoints and External Assembly
Before moving to the final assembly, perform a quality check on both generated images. Verify that the lighting, perspective, and aspect ratios match. If the images differ significantly in orientation or scale, the split-screen effect will look disjointed. At this stage, you may need to regenerate images until the visual continuity is acceptable.
Since Nano Banana 2 generates individual images, the actual combination into a postcard layout happens outside the application. Export both images from the generator. Open your preferred external design editor (such as Canva, Photoshop, or GIMP) to create the postcard canvas.
Create a vertical or horizontal split-screen layout. Place the historical image on one side and the modern image on the other. Use a dividing line or a subtle gradient overlay to distinguish the two eras. Add text overlays if necessary to label the dates or locations, keeping in mind that the AI cannot reliably preserve specific text in the generated images themselves.
Finalizing and Exporting the Postcard
The final step involves refining the composite image. Ensure the resolution is sufficient for printing or digital sharing. Adjust brightness and contrast to make the transition between the two halves feel intentional rather than accidental. Save the final file in a high-quality format suitable for your distribution channel.
This workflow demonstrates a practical approach to creating historical comparisons using AI tools. By separating the generation phase from the layout phase, users maintain full control over the final aesthetic. Remember that while the tool offers powerful generative capabilities, the creative direction and technical assembly rely on the user's input and external software skills. This method ensures flexibility and high-quality output for your comparison postcards.
For more information on the models used in this process, refer to the official documentation on Google Gemini image generation.