Run ComfyUI + Ollama Together on 8GB VRAM

The impossible made possible. Run a full image generation pipeline AND a local LLM on the same 8GB GPU. Optimized configs, memory management tricks, and step-by-step setup for budget hardware.