Troubleshooting Vision Recall

This guide covers common issues you might encounter when using Vision Recall and how to resolve them.

Connection Issues

Cannot Connect to OpenAI/OpenRouter/Groq

Symptoms:

• "API Error" messages
• Test connection fails
• Screenshots don't process

Solutions:

1. Verify API Key:

   • Check that your API key is entered correctly with no extra spaces
   • Generate a new API key if necessary

2. Check API Base URL:

   • Ensure the URL is correct for your provider
   • OpenAI: https://api.openai.com/v1
   • OpenRouter: https://openrouter.ai/api/v1
   • Groq: https://api.groq.com/openai/v1

3. Check Internet Connection:

   • Ensure your device has a stable internet connection

4. Check Provider Status:

   • Visit your provider's status page to check for outages
   • OpenAI Status
   • OpenRouter Status

Cannot Connect to Ollama

Symptoms:

• "Connection refused" errors
• Test connection fails
• Screenshots don't process

Solutions:

1. Verify Ollama is Running:

   • Open a terminal and run ollama list
   • If Ollama isn't running, start it with ollama serve

2. Check API Base URL:

   • Ensure it's set to http://localhost:11434 (or your custom address)

3. Verify Model Installation:

   • Run ollama list to check if you have a vision-capable model installed
   • If not, install one with ollama pull llava or another vision model

4. Restart Ollama:

   • Close and restart the Ollama application

Processing Issues

Screenshots Not Being Processed

Symptoms:

• Screenshots in the intake folder aren't being processed
• No notifications or progress indicators appear

Solutions:

1. Check Auto-Processing Settings:

   • Ensure "Auto-Process Screenshots" is enabled in settings
   • Check that "Periodic Intake Folder Check" is enabled

2. Verify Folder Paths:

   • Confirm your intake folder path is correct
   • Make sure the folder exists in your vault

3. Check File Types:

   • Ensure your screenshots are in supported formats (PNG, JPG, JPEG, WEBP)

4. Manually Trigger Processing:

   • Use the "Process Intake Folder" command to manually start processing

Poor OCR Results

Symptoms:

• Text from images is missing or inaccurate
• Generated notes don't include text visible in the screenshot

Solutions:

1. Improve Image Quality:

   • Use higher resolution screenshots
   • Ensure text is clear and not blurry

2. Adjust OCR Settings:

   • Try different OCR engines if available
   • Experiment with OCR preprocessing options

3. Supplement with Vision Analysis:

   • The vision LLM can often recognize text that OCR misses
   • Ensure vision analysis is enabled

Poor Vision Analysis Results

Symptoms:

• AI descriptions of images are inaccurate
• Generated notes don't capture the content of the screenshot

Solutions:

1. Try a Different Model:

   • Some models perform better than others for certain types of images
   • GPT-4o or Claude 3 Opus generally provide the best results

2. Check Image Size:

   • Very large images may be downscaled, losing detail
   • Very small images may not contain enough information for analysis

3. Adjust Vision Prompt:

   • Customize the vision prompt in advanced settings for better results

Storage and Performance Issues

Plugin Running Slowly

Symptoms:

• Long processing times for screenshots
• Obsidian becomes unresponsive during processing

Solutions:

1. Reduce Image Size:

   • Lower resolution screenshots process faster
   • Consider enabling automatic image optimization in settings

2. Use a Faster Model:

   • Some models are faster than others
   • For Ollama, smaller models generally process faster

3. Process Fewer Images at Once:

   • Process screenshots in smaller batches
   • Adjust the queue size in advanced settings

High Disk Usage

Symptoms:

• Vault size growing rapidly
• Storage space running low

Solutions:

1. Enable Image Optimization:

   • Turn on automatic image optimization in settings
   • This can significantly reduce storage requirements

2. Adjust Storage Settings:

   • Consider storing screenshots outside your vault
   • Use external links instead of embedding images in notes

3. Clean Up Old Screenshots:

   • Periodically review and delete unnecessary screenshots
   • Use the gallery view to identify and remove duplicates

Plugin Conflicts

Conflicts with Other Plugins

Symptoms:

• Errors when using Vision Recall with certain other plugins
• Unexpected behavior when multiple plugins are enabled

Solutions:

1. Update All Plugins:

   • Ensure all plugins are updated to their latest versions

2. Identify Conflicting Plugins:

   • Temporarily disable other plugins to identify conflicts
   • Re-enable them one by one to pinpoint the issue

3. Check for Known Conflicts:

   • Review the plugin's GitHub issues for reported conflicts
   • Check the community forum for similar reports

Getting Additional Help

If you're still experiencing issues:

1. Check GitHub Issues:

   • Visit the Vision Recall GitHub repository
   • Search for similar issues or create a new one

2. Join the Community:

   • Ask for help in the Vision Recall forum or in the GitHub Discussions

3. Provide Detailed Information:

   • Include your plugin version, Obsidian version, and OS
   • Describe the steps to reproduce the issue
   • Share relevant error messages and logs (enable debug mode in settings)