Kevin demonstrates how to generate AI videos locally on a PC using Pinokio and Wan2GP, which supports models like LTX-2 and Wan. The process is free, private, and requires an NVIDIA GPU with at least 6-8 GB VRAM. He covers installation, configuration, text-to-video, image-to-video, and advanced settings like narration and multiple outputs.
A short editorial from the VEONIB team on why this content matters.
Kevin provides a thorough, beginner-friendly walkthrough for running free, open-source AI video models locally—cutting through technical complexity with tools like Pinokio.
Unlike cloud-based AI video generators (e.g., Sora, Veo), this approach eliminates subscriptions and usage caps, giving creators full control and privacy—a major advantage for frequent or sensitive projects.
Watch this if you want to produce custom AI videos without recurring costs; start with a distilled model and a short, descriptive prompt to test your system.
A one-click installer for AI tools that manages dependencies like Python and CUDA.
A web UI script for running multiple open-source AI video models locally.
A 19-billion parameter open-source video model that also generates audio and narration.
An open-source AI video model available for local use through Wan2GP.
Video RAM on a GPU; at least 6-8 GB recommended for local AI video generation.
A smaller, faster version of a full AI model designed for consumer GPUs.
Generating video clips from a text prompt using AI.
What hardware do I need to run free AI video generation on my PC?
An NVIDIA GPU with at least 6-8 GB VRAM, plus sufficient system RAM and disk space for model weights (35-40 GB for full models, ~20 GB for distilled).
How do I check my GPU VRAM on Windows?
Press Ctrl+Shift+Esc to open Task Manager, go to Performance > GPU, and look at Dedicated GPU Memory.
Can I generate videos with sound and narration?
Yes, models like LTX-2 can produce both audio and dialogue. You can specify narration in your text prompt using quotes.
What is Pinokio and why should I use it?
Pinokio is a one-click installer for AI tools that automatically manages Python, CUDA, and dependencies, simplifying the setup process.
How long does it take to generate a video?
The first generation may take ~2 minutes due to model loading; subsequent clips can render in about 30 seconds depending on resolution and GPU.
Can I use an image as a starting point for video generation?
Yes, Wan2GP supports image-to-video: upload a photo, provide a text description, and the AI animates it.
What are the quality and aspect ratio options?
You can choose up to 1080p resolution and a variety of aspect ratios including 16:9 (horizontal) and 9:16 (vertical for TikTok).
Is my data safe when generating videos locally?
Yes, everything runs on your own PC with no cloud uploads, so your files and prompts remain private.
Can I generate multiple videos at once?
Yes, you can queue multiple prompts and adjust the number of variations per prompt using Advanced Mode.
Where are the generated videos saved?
Inside the Wan2GP project folder under App > Outputs. You can access them via the Wan2GP interface's file tab.