Mastering Material Swaps: Ceramic to Glass Coffee Cup with Nano Banana 2

Creating photorealistic images of everyday objects often requires precise control over surface properties. When working with a coffee cup, the difference between a matte ceramic finish and a transparent glass vessel is defined by how light interacts with the surface and how heat manifests as steam. Nano Banana 2, identified in Google documentation as Gemini 3.1 Flash Image, provides a robust environment for these specific text-to-image and image-to-image workflows. This tutorial focuses on mastering the technique of swapping the coffee cup material within a single prompt session, ensuring that the steam and lighting remain consistent while the underlying texture changes.

Understanding Surface Reflectivity and Heat Interaction

The core challenge in material swapping lies in accurately describing the physical properties of the new material without losing the context of the scene. When you request a change from ceramic to glass, you must explicitly define how the surface handles light. Ceramic typically scatters light, creating a soft, diffuse reflection, whereas glass allows light to pass through, refract, and reflect sharply off its edges. Similarly, metal introduces high-contrast specular highlights and mirror-like reflections.

To achieve a convincing result, your prompt must address the interaction between the hot liquid and the container. Steam rising from a coffee cup behaves differently depending on the material. With a ceramic cup, the steam rises naturally from the rim. However, when the cup is made of glass, the viewer might see condensation forming on the exterior if the glass is cold, or a clear view of the dark liquid inside if the glass is thin. For metal cups, the focus shifts to the sleek, cool surface that contrasts with the warm vapor. Describing these nuances ensures the AI understands that the object has changed identity, not just its color.

Step-by-Step Guide to Executing the Swap

Executing a successful material swap requires a structured approach to your input. Follow these numbered steps to refine your results using the Nano Banana 2 interface:

1. Prepare Your Base Input: Start with an existing image of a coffee cup or generate a base image using a simple description like "a steaming coffee cup on a wooden table." Ensure the composition is stable before attempting modifications.
2. Define the Target Material: In your prompt, clearly state the desired material. Use phrases such as "transparent borosilicate glass" or "brushed stainless steel" rather than generic terms. This specificity helps the model adjust the rendering engine's expectations for refraction and reflection.
3. Describe Light and Steam Dynamics: Add instructions regarding the visual effects of the new material. For example, specify "show internal refraction of the coffee liquid" for glass or "sharp metallic highlights reflecting the window light" for metal. Include details about the steam, such as "soft white vapor curling above the rim" to maintain continuity.
4. Iterate and Refine: If the first result looks too glossy or lacks transparency, adjust the prompt to emphasize opacity levels or surface roughness. You may need to rephrase the instruction to "matte black ceramic" versus "glossy white ceramic" to test the boundaries of the tool's understanding.
5. Review Output Consistency: Check that the steam does not disappear or become disconnected from the cup. The physics of the vapor should remain logical relative to the new material's thermal properties.

Crafting Effective Prompts for Texture Changes

A usable prompt serves as the blueprint for the generation process. Below is an example prompt designed to swap a ceramic cup for a glass one. Please note that this is an example; prompt instructions describe desired outcomes but do not guarantee identity, label, object, or typography preservation.

Example Prompt: "Change the coffee cup material from matte ceramic to clear thick glass. The cup should show refraction of the dark coffee inside and sharp edge reflections. Keep the steam rising gently from the top, appearing slightly denser against the transparent background. Maintain the same wooden table lighting and shadow direction."

Another example for a metal swap would be: "Replace the ceramic cup with a polished silver metal mug. Add strong specular highlights on the rim and handle. The steam should look crisp and distinct against the shiny surface. Preserve the original angle and background blur."

These examples illustrate how to layer material definitions with environmental cues. By combining specific texture keywords with atmospheric descriptions, you guide the model toward a more coherent transformation.

Judging Results and Troubleshooting Common Issues

Evaluating the success of your material swap involves checking for visual consistency and physical plausibility. A good result will show the correct optical behavior for the chosen material. If the glass cup appears opaque, the prompt likely lacked sufficient emphasis on transparency or refraction. Conversely, if the metal cup looks dull, it may have missed the specular highlight instructions.

Common issues include the steam disappearing entirely or the cup blending into the background. If the steam vanishes, try reinforcing the instruction to "maintain visible steam volume" or "keep vapor density constant." If the cup loses its shape, ensure you are using the image-to-image workflow correctly rather than generating a completely new scene. Remember that Nano Banana 2 Lite is focused on speed and cost and is not optimized for multiple reference inputs or multi-turn sequential editing; for complex swaps, standard Nano Banana 2 is recommended.

For those looking to experiment further with these advanced texture techniques, Try Nano Banana to access the full capabilities of the platform. By carefully crafting your descriptions of reflectivity and heat interaction, you can achieve professional-grade material swaps that enhance the realism of your generated imagery.

Note: This guide relies on verified product facts and does not invent statistics or download functionality. All model names referenced correspond to Google's official documentation.