Prompt image edits like an editor, not a caption writer
Image-editing prompts are strongest when they separate what should change from what must remain untouched. If a product photo needs a new background but the logo, proportions, packaging text, and product shape must stay the same, say that explicitly.
State the edit first
Begin with the actual transformation: replace the background, restyle the lighting, remove an object, or generate a new composition. Avoid burying the requested change at the end of a long aesthetic description.
Use preservation instructions carefully
Preservation fields are most useful for identity-sensitive details such as faces, logos, labels, product geometry, clothing, or layout. Do not mark everything as immutable, because the model still needs room to perform the requested edit.
Describe composition and lighting separately
Composition controls where the subject sits and how the frame is organized. Lighting controls the visual treatment. Keeping these separate makes the prompt easier to review and modify.
Nano Banana vs. a general AI image prompt generator
Use this page when edit instructions and preservation constraints are central to the task. For model-agnostic image generation with subject, style, composition, camera, lens, and lighting controls, use the AI Image Prompt Generator.
Frequently asked questions
Does PromptVera upload my image?
No. This tool currently generates prompt text only; it does not upload or process an image.
Can I use this for ecommerce product images?
Yes. Product work benefits from explicit preservation rules for geometry, branding, labels, and proportions.