← AI Library
Visuals · Prompt StructureNano Banana · Image Models

Anatomy of a High-Control Image Prompt

Long prompts work when every section controls a visible decision. Structure the scene so you can revise one variable without rewriting everything.

VisualsPromptsGemini
Control Stack

Write in Visible Layers

Subject

Identity, age range, object, wardrobe, pose, expression, and the features that must remain consistent.

Composition

Shot size, camera position, lens feel, aspect ratio, subject placement, and depth.

Environment

Location, era, weather, background objects, surface details, and spatial relationships.

Light and Color

Source direction, softness, contrast, palette, reflections, shadows, and color treatment.

Material and Finish

Texture, medium, production method, grain, polish, and the physical cues that sell the image.

Exclusions

Unwanted text, logos, extra objects, anatomy errors, style drift, or changes to protected features.

Modular Image Prompt

Create [image type] of [subject and action]. Compose it as [shot, angle, lens feel, aspect ratio]. Place the subject in [environment and key objects]. Light the scene with [direction, quality, contrast, palette]. Render [materials and textures] with [finish or medium]. Preserve [identity or product features]. Exclude [specific unwanted elements]. The image succeeds when [three visible criteria].

Revision rule: change one block at a time. If the composition is wrong, do not also change the palette and wardrobe; you will lose the evidence about what fixed it.

Original Source

Pattern study based on the categories and attributed examples in ZeroLu's Awesome Nano Banana Pro repository. This guide uses original wording and no repository images. Consult the repository for creator credits attached to individual examples.