ComfyUI AI Video Expert – LTX-2.5 + Consistent Characters + Cloud GPU Automation

I am looking for an experienced ComfyUI AI video expert to help me build a repeatable, automated and cost-efficient workflow for creating long-form, cozy, hand-painted anime-inspired ASMR videos for YouTube. My goal is not simply to create one video. I want to develop a production workflow that can eventually create 30–60+ minute videos regularly from many short AI-generated scenes. The videos will include: • Consistent characters across many scenes • Beautiful hand-painted, cozy anime-inspired visuals • Natural character movements • Gentle cinematic camera movement • Scenes such as cooking, camping, trains, rain, countryside and family life • No dialogue or talking • Synchronized ASMR sounds such as rain, footsteps, cooking, fire, water, wind, birds and trains • Many short AI-generated scenes combined into long videos I am especially interested in someone experienced with: • ComfyUI • LTX-2.5 video + synchronized audio • WAN and other open-source video models • Image-to-video and text-to-video • Character consistency • LoRA • ControlNet / IPAdapter • Batch generation and automation • Cloud GPUs such as RunPod, Vast.ai or Google Cloud • GPU speed and cost optimization Ideally, I want to develop an automated workflow similar to: Prompt/story → Generate consistent images → Animate images → Generate synchronized ASMR audio → Automatically queue scenes → Save finished clips I want the workflow to be easy for me to operate after it has been built. Please send me: 1. Examples of AI videos you have created 2. Your ComfyUI experience 3. Your experience with LTX-2.5 and synchronized video/audio 4. Which model and workflow you recommend for this project 5. How you would maintain consistent characters across many scenes 6. How you would automate batch generation 7. Which cloud GPU/provider you recommend for maximum speed and reasonable cost 8. Your estimated cost for building the workflow I would like to begin with a small 5–10 second test scene before developing the complete workflow. For the test, please report: • GPU used • Resolution • Generation time • VRAM usage • Approximate GPU cost If the test and workflow are successful, this may become ongoing work.