> ## Documentation Index
> Fetch the complete documentation index at: https://docs.praison.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Image Processing

> Process images with vision-based AI agents

The `--image` flag enables image processing with vision-capable AI models.

## Quick Start

```bash theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
praisonai "Describe this image" --image path/to/image.png
```

## Usage

### Basic Image Analysis

```bash theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
praisonai "What's in this photo?" --image photo.jpg
```

**Expected Output:**

```
🖼️ Processing image: photo.jpg

╭─ Agent Info ─────────────────────────────────────────────────────────────────╮
│  👤 Agent: ImageAgent                                                        │
│  Role: Vision Assistant                                                      │
╰──────────────────────────────────────────────────────────────────────────────╯

╭────────────────────────────────── Response ──────────────────────────────────╮
│ The image shows a golden retriever dog sitting on a grassy lawn. The dog     │
│ appears to be smiling with its tongue out. In the background, there's a      │
│ wooden fence and some trees. The lighting suggests it was taken during       │
│ late afternoon, creating a warm, golden atmosphere.                          │
╰──────────────────────────────────────────────────────────────────────────────╯
```

### Specify Vision Model

```bash theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
# Use GPT-4o for vision
praisonai "Analyze this chart" --image chart.png --llm openai/gpt-4o

# Use Claude for vision
praisonai "Describe the scene" --image scene.jpg --llm anthropic/claude-3-sonnet-20240229
```

### Combine with Other Features

```bash theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
# Image analysis with metrics
praisonai "Count objects" --image warehouse.jpg --metrics

# Image with guardrail
praisonai "Extract text from image" --image document.png --guardrail "Output as JSON"

# Image with save
praisonai "Describe artwork" --image painting.jpg --save
```

## Supported Image Formats

| Format | Extension       | Support        |
| ------ | --------------- | -------------- |
| JPEG   | `.jpg`, `.jpeg` | ✅ Full         |
| PNG    | `.png`          | ✅ Full         |
| GIF    | `.gif`          | ✅ Static frame |
| WebP   | `.webp`         | ✅ Full         |
| BMP    | `.bmp`          | ✅ Full         |

## Use Cases

### Document Analysis

```bash theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
praisonai "Extract all text from this document" --image invoice.png
```

**Expected Output:**

```
╭────────────────────────────────── Response ──────────────────────────────────╮
│ Invoice #: INV-2024-001                                                      │
│ Date: December 16, 2024                                                      │
│ Customer: Acme Corp                                                          │
│                                                                              │
│ Items:                                                                       │
│ - Widget A x 10 @ $25.00 = $250.00                                          │
│ - Widget B x 5 @ $40.00 = $200.00                                           │
│                                                                              │
│ Subtotal: $450.00                                                            │
│ Tax (10%): $45.00                                                            │
│ Total: $495.00                                                               │
╰──────────────────────────────────────────────────────────────────────────────╯
```

### Chart/Graph Analysis

```bash theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
praisonai "Analyze trends in this chart and provide insights" --image sales_chart.png
```

### Code Screenshot Analysis

```bash theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
praisonai "Review this code and identify bugs" --image code_screenshot.png
```

### UI/UX Review

```bash theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
praisonai "Provide UX feedback for this interface" --image app_screenshot.png
```

### Object Detection

```bash theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
praisonai "List all objects visible in this image with their positions" --image room.jpg
```

**Expected Output:**

```
╭────────────────────────────────── Response ──────────────────────────────────╮
│ Objects detected:                                                            │
│                                                                              │
│ 1. Sofa (center-left) - Gray fabric, 3-seater                               │
│ 2. Coffee table (center) - Wooden, rectangular                              │
│ 3. TV (right wall) - Mounted, approximately 55"                             │
│ 4. Plant (left corner) - Potted fern                                        │
│ 5. Lamp (right of sofa) - Floor lamp, brass finish                          │
│ 6. Rug (floor, center) - Patterned, blue and white                          │
│ 7. Books (on coffee table) - Stack of 3-4 books                             │
│ 8. Window (background) - Large, with curtains                               │
╰──────────────────────────────────────────────────────────────────────────────╯
```

## Image Path Options

```bash theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
# Local file path
praisonai "Describe" --image ./images/photo.jpg

# Absolute path
praisonai "Describe" --image /Users/name/photos/image.png

# Relative path
praisonai "Describe" --image ../screenshots/screen.png
```

## Best Practices

<Tip>
  For best results, use high-resolution images with clear content. Blurry or low-quality images may produce less accurate descriptions.
</Tip>

<Warning>
  Image processing uses more tokens than text-only prompts. Use `--metrics` to monitor costs.
</Warning>

<CardGroup cols={2}>
  <Card title="Image Quality">
    Use clear, well-lit images for best results
  </Card>

  <Card title="Specific Prompts">
    Be specific about what you want to analyze in the image
  </Card>

  <Card title="File Size">
    Large images are automatically resized; originals under 20MB recommended
  </Card>

  <Card title="Model Selection">
    Use GPT-4o or Claude 3 for complex image analysis
  </Card>
</CardGroup>

## Related

* [Image Agent](/agents/image)
* [Multimodal Features](/features/multimodal)
* [Metrics CLI](/docs/cli/metrics)