# Prompting for Image Generation

## Generate vs Edit Decision

Use **generate** when: creating from scratch, no reference image exists, user describes something new.

Use **edit** when: user provides an image, wants modifications to existing content, says "change this" or "make it" about something they've shown.

## Prompt Structure

Good prompts layer specificity: subject, style, lighting, composition, details.

Weak: "a house"

Strong: "Victorian mansion at dusk, warm light in windows, autumn leaves in foreground, cinematic composition, photorealistic"

The model handles aspect ratio and resolution automatically based on the prompt content.

## Edit Prompt Patterns

Preserve context by describing only what should change:

- "Add sunglasses to the person"
- "Change the background to a beach"
- "Make it look like a watercolor painting"
- "Remove the text from the image"

Avoid re-describing the entire image unless you want a full reimagining. The model understands context from the source image.

## Refining User Requests

When a user gives a vague prompt like "a sunset", help them add specificity:

- What kind of sunset? (beach, mountain, city skyline)
- What style? (photorealistic, oil painting, anime)
- What mood? (peaceful, dramatic, romantic)
- Any specific elements? (silhouettes, reflections, clouds)

Ask clarifying questions or offer options rather than generating with the vague prompt.

## Output Location

Generated and edited images save to `nanobanana/` directory. Create uses a slug from the prompt (e.g., `victorian-mansion-at-dusk.png`). Edit uses the original filename with `-edited` suffix.