Realistic AI Video Generator

I need a custom generative-AI model that can turn a simple text prompt or idea into an entertaining, realistic-looking video every single day, ready to post on YouTube and Instagram rails. The footage must be good enough to pass as genuine live-action so it stays clear of current AI-detection tools.

Here is what I expect:

• A trained model or workflow (Diffusion, GAN, or any hybrid you trust) that outputs 60–120-second clips in vertical and horizontal formats.
• A lightweight inference script or API so I can batch-create videos on my own machine or in the cloud without you each time.
• One-click presets for YouTube Shorts and Instagram Reels (1080×1920 & 1920×1080, 30 fps, H.264).
• A short guide showing how to feed prompts, add background audio, and render final files.
• Two sample videos produced by the system to prove realism and non-detectability.

I welcome whatever tech stack you prefer—Python, PyTorch, Stable Diffusion, DreamBooth, ControlNet, or something entirely custom—so long as it hits the quality bar and remains scalable for daily output. Please include your price, timeline, and any upsells (extra fine-tuning, UI wrapper, cloud hosting, etc.) when you reply.