Mastering Eye Contact Angles in Nano Banana 2 Headshots

Creating a compelling headshot with an AI image generation tool requires more than just describing a face; it demands specific directional instructions. When using Nano Banana (the AI image generation and editing tool), users often seek to align the subject's expression with specific communication goals. Whether you need a direct, engaging stare or a thoughtful, off-camera glance, the precision of your prompt determines the outcome. This guide explores how to engineer prompts for specific eye contact angles and head tilts within the Nano Banana 2 workflow.

It is important to clarify that Nano Banana refers strictly to the AI tool itself. It is not a skincare brand, bottle, jar, or physical product. The generated images are digital creations based on text descriptions. While the platform supports text-to-image and image-to-image workflows, users must understand that prompt instructions describe desired outcomes but do not guarantee identity, label, object, or typography preservation. For those looking to experiment with these advanced techniques, Try Nano Banana offers a dedicated environment for testing these concepts.

Understanding Model Capabilities and Limitations

Before diving into specific prompts, it is essential to understand the underlying models powering the tool. 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). There is also Nano Banana 2 Lite, identified as Gemini 3.1 Flash Lite Image (gemini-3.1-flash-lite-image).

These models have distinct capabilities. Google describes Nano Banana 2 Lite as focused on speed and cost efficiency. However, it is not optimized for multiple reference inputs or multi-turn sequential editing. Therefore, if your goal involves complex pose adjustments requiring iterative refinement or multiple reference images, relying solely on the Lite version may yield inconsistent results without explaining this limitation. For high-fidelity headshots requiring precise gaze control, the standard Nano Banana 2 or Pro versions are generally recommended to ensure the model can interpret nuanced spatial instructions effectively.

Five Strategies for Directing Gaze and Posture

To achieve specific eye contact angles, you must move beyond generic descriptors like "looking at camera." Below are five materially different usable prompt examples. These are labeled as examples to demonstrate potential phrasing strategies rather than guaranteed outputs. Each scenario addresses a different communication intent.

1. The Direct Engagement Angle

Use Case: Ideal for LinkedIn profiles, corporate bios, or marketing materials where trust and direct connection are paramount. Prompt Example: "Professional headshot of a person facing forward, eyes locked directly on the camera lens, slight upward tilt of the chin to convey confidence, soft studio lighting, neutral background." Adjustment: If the eyes appear slightly off-center, add "perfectly centered pupils" or "symmetrical gaze" to the instruction. This works best when the model has sufficient context to understand the focal point.

2. The Thoughtful Off-Camera Glance

Use Case: Suitable for creative portfolios, editorial features, or storytelling contexts where the subject appears contemplative. Prompt Example: "Portrait of a subject looking three-quarters away from the camera, gazing toward the upper left corner, head tilted slightly downward, natural outdoor lighting, shallow depth of field." Adjustment: To increase the angle of the turn, specify "head turned forty-five degrees" or "gaze directed past the viewer's right shoulder." Note that extreme angles may require the Pro model for better structural integrity.

3. The Subtle Downward Nod

Use Case: Effective for conveying humility, approachability, or a listening posture in customer service or educational content. Prompt Example: "Close-up headshot, subject looking slightly down and to the side, gentle smile, head tilted gently to the right, warm indoor lighting, blurred office background." Adjustment: If the head tilt feels too exaggerated, refine the prompt to "minimal head tilt" or "slight nod." Ensure the description of the eyes clarifies they are looking down, not closed.

4. The Dynamic Upward Look

Use Case: Perfect for aspirational branding, leadership announcements, or themes of innovation and future-thinking. Prompt Example: "Dynamic portrait, subject looking up towards the sky, head tilted back slightly, bright natural sunlight hitting the face, energetic expression, urban skyline background." Adjustment: Be cautious with the phrase "looking up" as it can sometimes result in the nose being obscured. Adding "chin lifted but eyes visible" helps maintain facial clarity.

5. The Asymmetrical Profile Turn

Use Case: Used in fashion photography or artistic projects where a dramatic, non-standard composition is required. Prompt Example: "Three-quarter profile shot, subject turning head sharply to the left, eyes glancing back over the shoulder, dramatic shadow lighting, high contrast, monochrome aesthetic." Adjustment: This is a complex pose. If the model struggles with the anatomy, try simplifying to "profile view with eyes visible" before attempting the full "glancing back" motion. This example demonstrates the limits of current pose control.

Refining Your Results Through Iteration

Achieving the perfect eye contact angle often requires iteration. Since prompt instructions do not guarantee specific outcomes, users should treat these examples as starting points. If the generated image does not match the intended gaze, modify the directional keywords. Instead of "looking left," try "gaze directed to the far left edge of the frame."

Remember that the Nano Banana library offers example prompts that users can copy or take into the generator. However, these examples are generic and unbranded. They serve as templates rather than fixed rules. By understanding the distinction between the tool (Nano Banana) and the generated content, and by leveraging the specific strengths of the underlying Google models, users can create highly effective, communicative headshots tailored to their unique needs.