IMA Nano Banana

# IMA Nano Banana (Image Generation)

## Overview

This skill calls the IMA Open API for **image generation** using **only the Nano Banana series**: **Nano Banana**, **Nano Banana Pro**, **Nano Banana 2**. No other models (e.g. SeeDream, Midjourney) are used or documented here.

**Supported task types:** `text_to_image`, `image_to_image`.

---

## 📥 User Input Parsing (Model & Parameter Recognition)

Normalize user wording (case-insensitive), then map to **task_type**, **model_id**, and **parameters**.

### 1. User phrasing → task_type

| User intent / phrasing | task_type | Notes |
|------------------------|-----------|--------|
| Only text, no input image | `text_to_image` | "画一张…" / "生成图片" / "text to image" |
| One input image + edit/transform | `image_to_image` | "把这张图…" / "参考这张图生成" / "图生图" / "风格迁移" |

If the user attaches or links one image and asks to change it or generate something "like this", use `image_to_image` with that image as input.

### 2. Model name / alias → model_id (Nano Banana only)

| User says (examples) | model_id | Notes |
|----------------------|----------|--------|
| Nano Banana / 香蕉 / Banana2 / NB2 / 最便宜 / budget | `gemini-3.1-flash-image` | Nano Banana2, 4–13 pts |
| Nano Banana Pro / Banana Pro / NB Pro / 最好 / premium | `gemini-3-pro-image` | 10–18 pts |
| 默认 / 不指定 | `gemini-3.1-flash-image` | Default: balanced, fast |

### 3. User phrasing → size / aspect_ratio

| User says (examples) | Parameter | Normalized value | Notes |
|----------------------|-----------|------------------|-------|
| 16:9 / 横图 / 9:16 / 竖图 / 4:3 / 3:4 / 1:1 | aspect_ratio | 16:9, 9:16, 4:3, 3:4, 1:1 | Native support |
| 4K / 2K / 1K / 512 | size | 4K, 2K, 1K, 512px | Via attribute_id |

Nano Banana 系列均**原生支持** aspect_ratio。**8K** 不支持，最高 4K；若用户要求 8K，告知并改用 4K。

---

## 💬 User Experience Protocol (IM / Feishu / Discord)

This skill runs inside IM platforms (Feishu, Discord via OpenClaw).  
**Never...