Nano Banana Pro: typography, reference composition and precise editing
Use Google's current Cookbook to organize complex image briefs without confusing official capabilities with an integration's controls.
Model identity and availability
Google's Python Cookbook currently identifies Nano Banana Pro as gemini-3-pro-image-preview and describes thinking, 4K output, complex typography and high-fidelity composition. HUUUIAI uses the Nano Pro label; its actual input limits and settings come from the current workflow.
Official examples can differ by update date and SDK. Confirm the API version instead of combining parameters from unrelated samples.
Structure the information before styling it
Write the exact headline, supporting copy and reading order before requesting a poster. Verify the information yourself. Assign each reference a role: subject identity, composition or lighting. More references are useful only when they contribute distinct information.
Reference composition example
Create a product poster using the glass perfume bottle in image 1. Preserve its cap, label and liquid color. Use image 2 only as a soft side-light reference. Place the bottle in the lower center, with clean empty space above and no generated text.
Check the finished asset
Inspect transparent materials, reflections, label text and product structure at source resolution. Search grounding can help find information but does not replace checking dates, numbers and names. Keep an approved intermediate result before making another change.
Sources and verification
Checked on the dates below. Prompts are editorial examples, not measured performance results. Model availability and parameters in HUUUIAI depend on the current service.
