New Community Module: Ollama Multimodal Vision Adapter for CodeProject.AI (Images + Video + HA LLM Vision)

[zlatko-lakisic]

Hi CodeProject.AI community 👋

I wanted to share a module I built for the ecosystem:

🚀 OmegaOllama MultiModal LLM (Community Module)

This project integrates Ollama multimodal models with CodeProject.AI for:

• 🖼️ Image description / visual Q&A
• 🎬 Video description (frame sampling + summary)
• 🏠 Home Assistant integration via LLM Vision (OpenAI-compatible route)

Repo:
👉 https://github.com/zlatko-lakisic/CodeProjectAI-OmegaOllamaMLLM

---

Why I built it

I wanted a practical way to run local/open multimodal models through CodeProject.AI and use them in real automations (especially Home Assistant camera workflows).

The module focuses on:

• Easy deployment in CodeProject.AI module structure
• Configurable vision models (default: moondream)
• Synchronous video analysis with controllable frame interval / frame count
• Better compatibility for real-world integrations

---

What I’d love feedback on

• Preferred patterns for multimodal module APIs in CodeProject.AI
• Best practices for OpenAI-compatible route handling in module adapters
• Suggestions for making this easier to ship as a community module
• Interest in additional endpoints / capabilities

---

If anyone wants to test it, share logs, or collaborate, I’d love to iterate with the community 🙌