Skip to content

Visual Analysis

Visual analysis enables the AI to examine and describe images in your Zettelkasten.

Setup

Requirements

  1. Visual model configured - A model capable of image analysis
  2. Images in your vault - PNG, JPEG, GIF, or other common formats

Configuring a Visual Model

During setup, when asked: 

Do you want to enable visual analysis? (y/n):

Type y and select a vision-capable model:

For Ollama (2025 Models):

High Quality (36-48GB+ RAM): - gemma3:27b - Recommended - Excellent vision, efficient resource use - qwen3-vl:32b - Most powerful Qwen vision model - llama3.2-vision:90b - Alternative (requires 64GB+ RAM)

Balanced (16-32GB RAM): - gemma3:9b - Recommended - Good quality, reasonable resources - qwen3-vl:8b - Fast and capable - llama3.2-vision:11b - Alternative balanced option

Lightweight (8-16GB RAM): - gemma3:2b - Fast inference with vision - llava:7b - Fallback for very limited resources - llava-phi3 - Minimal resource requirements

For OpenAI: - gpt-4o - Latest multimodal GPT-4 - gpt-5 - State-of-the-art multimodal (if available)

Using Visual Analysis

Analyzing Images

Simply ask about an image:

You: What's in the image at diagrams/architecture.png?

AI: Using tool: analyze_image
Path: diagrams/architecture.png

The image shows a system architecture diagram depicting three main layers:
1. Frontend - React-based web application
2. Backend API - Node.js services
3. Database - PostgreSQL with Redis cache
...

Common Use Cases

Diagram Analysis 

You: Describe the flowchart in process-diagram.png

Photo Description 

You: What does the whiteboard photo show?

Chart Interpretation 

You: Analyze the graph in sales-data-2024.jpg

Text Extraction 

You: What text is in the screenshot?

Supported Formats

  • PNG (.png)
  • JPEG (.jpg, .jpeg)
  • GIF (.gif)
  • WebP (.webp)
  • BMP (.bmp)

Image Paths

Images can be referenced:

Relative to vault: 

assets/diagram.png
images/photo.jpg
attachments/screenshot.png

In subdirectories: 

projects/project-a/wireframes/design.png
notes/embedded-image.jpg

Best Practices

  1. Be specific - Ask clear questions about what you want to know
  2. Good organization - Keep images in logical folders
  3. Descriptive names - Use meaningful filenames
  4. Ask follow-ups - Request more details if needed

Example Workflow

You: What diagrams do I have in my vault?

AI: [Lists image files]
Found 5 diagrams:
- architecture.png
- dataflow.png
- ui-mockup.png
...

You: Describe the architecture diagram

AI: Using tool: analyze_image
[Provides detailed description]

You: How does this relate to my implementation notes?

AI: Using tool: find_documents
Query: architecture implementation
[Finds related documents and explains connections]

Limitations

  • Model dependent - Quality varies by model
  • Complex images - May struggle with very detailed diagrams
  • Text readability - Small text may not be readable
  • File size - Large images may take longer to process

Troubleshooting

No Visual Model Available

Error: "Visual analysis not configured"

Solution: 

# Reconfigure and select a visual model
zk-chat interactive
# Follow prompts to select visual model

Image Not Found

Error: "Image not found at path"

Check: - File exists in vault - Path is correct (relative to vault root) - File extension is supported

Poor Results

Try: - Use a different visual model - Ensure image quality is good - Ask more specific questions - Describe what you're looking for

See Also