README_API.md

# Nano Banana Image Edit API

A FastAPI-based REST API for the Qwen Image Edit model that allows you to upload images, edit them with text prompts, and retrieve results.

## Features

- **Upload Images**: Upload image files via POST request
- **Edit Images**: Submit image ID and text prompt to edit images
- **Get Results**: Retrieve edited images by task ID
- **Multiple Formats**: Get results as files or base64-encoded strings

## Installation

1. Install dependencies:
```bash
pip install -r requirements_api.txt
```

2. Run the API server:
```bash
python api.py
```

Or using uvicorn directly:
```bash
uvicorn api:app --reload --host 0.0.0.0 --port 8000
```

## API Endpoints

### 1. Upload Image
**POST** `/upload`

Upload an image file.

**Request:**
- Content-Type: `multipart/form-data`
- Body: `file` (image file)

**Response:**
```json
{
  "image_id": "uuid-string",
  "message": "Image uploaded successfully",
  "timestamp": "2024-01-01T12:00:00"
}
```

**Example using curl:**
```bash
curl -X POST "http://localhost:8000/upload" \
  -F "file=@/path/to/image.jpg"
```

### 2. Edit Image
**POST** `/edit`

Edit an uploaded image using a text prompt.

**Request:**
- Content-Type: `multipart/form-data`
- Body:
  - `image_id`: ID of uploaded image
  - `prompt`: Text description of desired edit

**Response:**
```json
{
  "task_id": "uuid-string",
  "status": "completed",
  "message": "Image edited successfully",
  "timestamp": "2024-01-01T12:00:00"
}
```

**Example using curl:**
```bash
curl -X POST "http://localhost:8000/edit" \
  -F "image_id=your-image-id" \
  -F "prompt=make the sky more blue"
```

### 3. Get Result
**GET** `/result/{task_id}`

Get the status and result information for an editing task.

**Response:**
```json
{
  "task_id": "uuid-string",
  "status": "completed",
  "result_image_id": "uuid-string",
  "result_image_url": "/result/image/uuid-string",
  "timestamp": "2024-01-01T12:00:00"
}
```

### 4. Get Result Image (File)
**GET** `/result/image/{result_image_id}`

Download the edited image as a file.

**Response:** Image file (PNG)

### 5. Get Result Image (Base64)
**GET** `/result/image/{result_image_id}/base64`

Get the edited image as a base64-encoded string.

**Response:**
```json
{
  "result_image_id": "uuid-string",
  "image_base64": "base64-encoded-string",
  "format": "png"
}
```

### 6. Health Check
**GET** `/health`

Check API health and model status.

**Response:**
```json
{
  "status": "healthy",
  "model_loaded": true,
  "model_available": true
}
```

## Usage Example (Python)

```python
import requests

# Upload image
with open("image.jpg", "rb") as f:
    response = requests.post(
        "http://localhost:8000/upload",
        files={"file": f}
    )
upload_data = response.json()
image_id = upload_data["image_id"]

# Edit image
response = requests.post(
    "http://localhost:8000/edit",
    data={
        "image_id": image_id,
        "prompt": "make the background sunset orange"
    }
)
edit_data = response.json()
task_id = edit_data["task_id"]

# Get result
response = requests.get(f"http://localhost:8000/result/{task_id}")
result_data = response.json()
result_image_id = result_data["result_image_id"]

# Download edited image
response = requests.get(f"http://localhost:8000/result/image/{result_image_id}")
with open("edited_image.png", "wb") as f:
    f.write(response.content)
```

## Notes

- The current implementation uses in-memory storage for tasks. For production, consider using:
  - Redis or database for task storage
  - S3, Azure Blob Storage, or similar for image storage
  - Proper authentication and rate limiting
  - Async task processing with Celery or similar

- The model loading code is a placeholder. You'll need to adjust it based on the actual Qwen-Image-Edit model interface and requirements.

## API Documentation

Once the server is running, visit:
- Swagger UI: `http://localhost:8000/docs`
- ReDoc: `http://localhost:8000/redoc`