Mastering Mouth Expressions: Nano Banana 2 Prompt Engineering for Headshots

Creating a professional headshot often hinges on subtle details that convey confidence and approachability. Among these details, the mouth expression is critical. A generic smile might feel too casual, while a neutral expression can appear stern. For recruiters, actors, and corporate professionals, achieving a specific look—such as a closed-lip smile—is essential for aligning the image with brand standards or personal branding goals. This is where Nano Banana 2 prompt engineering becomes indispensable.

Nano Banana refers to the AI image generation and editing tool available on this platform. It is not a skincare brand, bottle, jar, or physical subject. By leveraging text-to-image and image-to-image workflows, users can guide the model to render specific facial configurations. The goal is to move beyond vague descriptions like "smiling" and instead provide granular instructions that dictate the exact state of the lips and jaw. This level of control ensures the final output matches the intended emotional tone without requiring multiple iterations.

Defining Specific Mouth Configurations

To achieve a closed-lip smile, the prompt must explicitly describe the geometry of the mouth rather than relying on the model's interpretation of the word "smile." In standard usage, models might default to an open-mouthed grin or a wide laugh. To correct this, you must define the contact between the lips. Phrases such as "lips pressed together," "sealed lips," or "mouth closed but corners upturned" provide the necessary constraints.

When constructing your prompt, consider the context of the face. A closed-lip smile often involves slight tension in the cheeks and a subtle lift at the corners of the mouth. You can combine these descriptors with lighting and style keywords to enhance the effect. For instance, specifying "soft studio lighting" alongside "closed-lip smile" helps the model understand the mood without overcomplicating the facial structure. Remember that prompt instructions describe desired outcomes; they do not guarantee identity, label, object, or typography preservation. Therefore, if you are using an input image, the focus should remain on modifying the expression while maintaining the overall likeness as best as possible.

Five Strategies for Expression Control

Below are five materially different usable prompts designed to help you control mouth expressions in Nano Banana 2. These examples illustrate how varying the language changes the outcome. Please note that these are examples of prompt structures and may require adjustment based on your specific input images or desired results.

1. The Direct Constraint Prompt: "Professional headshot, woman looking at camera, closed-lip smile, lips sealed tight, no teeth visible, soft natural lighting, blurred office background." When it helps: Use this when you need a strict adherence to a non-teeth showing smile. It works best for corporate profiles where a wide grin is inappropriate. Adjustment: If the model still shows teeth, add "strictly no teeth" or "lips fully touching."

2. The Emotional Nuance Prompt: "Corporate executive portrait, confident closed-lip smile, subtle upward curve at mouth corners, relaxed jawline, high-resolution, cinematic depth of field." When it helps: Ideal for leadership bios where the expression needs to convey warmth without appearing overly enthusiastic. It focuses on the emotion behind the closed mouth. Adjustment: Increase the weight of "subtle" if the smile appears too forced or exaggerated.

3. The Image-to-Image Refinement Prompt: "Modify input image to show a closed-lip smile only, keep all other facial features identical, lips pressed together, professional attire, neutral expression elsewhere." When it helps: Best used when you have a base photo with the wrong expression (e.g., open mouth) and want to change only the mouth area. This relies on the image-to-image workflow. Adjustment: Ensure the input image has clear visibility of the mouth area for better targeting.

4. The Gender-Specific Approach Prompt: "Male headshot, serious yet approachable, closed-lip smile, thin line of lips, minimal cheek movement, sharp focus, studio lighting." When it helps: Useful for male subjects where a closed-lip smile might otherwise be interpreted as a smirk. This prompt emphasizes minimalism and seriousness. Adjustment: If the result looks too stern, add "warm eyes" or "gentle expression" to balance the rigid lip description.

5. The Lighting and Texture Focus Prompt: "Close-up headshot, closed-lip smile, texture of skin visible, lips slightly parted but not open, glossy finish, dramatic side lighting." When it helps: Effective for fashion or creative portfolios where texture and lighting play a major role in the expression's perception. Note that "slightly parted but not open" is a nuanced way to describe a very tight seal. Adjustment: If the lips appear too dry, add "moisturized lips" or "natural lip texture."

Selecting the Right Model for Precision

While Nano Banana 2 offers robust capabilities, understanding the underlying model differences is crucial for complex tasks like expression control. 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 varying strengths.

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 detailed prompt engineering involving specific mouth expressions, the standard Nano Banana 2 or Nano Banana Pro models are generally more reliable due to their higher fidelity in interpreting nuanced text instructions. Nano Banana 2 Lite may struggle with the fine-grained control required for a perfect closed-lip smile compared to its counterparts.

By carefully crafting your prompts and selecting the appropriate model tier, you can consistently generate headshots that meet professional standards. Whether you are refining a single image or generating new ones from scratch, the key lies in specificity. Avoid generic terms and embrace descriptive language that leaves little room for ambiguity.

Try Nano Banana

For further technical details on image generation capabilities, refer to the official documentation. Always remember that while prompts guide the process, the final result depends on the interplay of your input data and the model's current training parameters.