Free AI Video Generator on Your PC (No Subscriptions, No Limits)
How to Generate AI Videos for Free on Your PC
Introduction to AI Video Generation
- Kevin introduces the topic of generating AI videos for free directly on a PC, emphasizing no usage limits or subscriptions.
- He mentions that the process also includes generating audio and narration alongside video content.
System Requirements
- The importance of checking system requirements is highlighted; generating AI video is resource-intensive but manageable on lower-end machines.
- A dedicated graphics card, preferably an NVIDIA GPU with at least 6 to 8 GB of VRAM, is recommended for optimal performance.
- Instructions are provided on how to check VRAM on a Windows PC using Task Manager.
Using Pinokio for Installation
- Kevin introduces Pinokio as a tool that simplifies the installation of various AI models by managing dependencies automatically.
- Users can download Pinokio from its website, with compatibility across Mac, Linux, and Windows platforms.
Installing and Setting Up Models
- After installing Pinokio, users will see a welcome screen where they can discover different verified scripts available for installation.
- Kevin selects the Wan2GP script for generating AI videos and explains its interface which allows running multiple video models in one place.
Choosing Video Models
- The installation process for Wan2GP is straightforward due to Pinokio's one-click install feature.
- Kevin discusses selecting between popular models like Wan2.2 and LTX-2; he opts for LTX-2 due to its ability to generate sound alongside video output.
Model Configuration Options
- The LTX-2 model has 19 billion parameters requiring significant disk space (35 to 40 GB), highlighting its complexity.
- Users have options between default and distilled versions of the model; the distilled version is more practical for consumer GPUs while maintaining quality.
Performance Settings Adjustments
- Before running the model, adjustments in configuration settings are necessary based on individual PC specifications.
AI Video Generation Process
Selecting Model Profiles and Input Methods
- The speaker discusses the selection of profile two for video generation, highlighting the importance of system specifications like RAM and VRAM.
- Various input methods for generating videos are introduced: text prompts only, image plus text, end images, and continuing existing videos.
- A sample prompt is provided to illustrate how users can describe their desired AI video. Users can also specify narration within the prompt.
Configuring Video Settings
- Emphasis on the significance of detailed prompts; tools like ChatGPT or Gemini can assist in refining these prompts for better results.
- Users can select video quality up to 1080p and choose aspect ratios (e.g., 9:16 for vertical videos or 16:9 for horizontal).
- Duration settings allow users to adjust video length from under a frame to about 30 seconds, with a recommendation to start at lower resolutions for faster rendering.
Generating Videos
- After configuring settings, users click "Generate" to create their video. Additional prompts can be queued while waiting for the first video to finish.
- Progress tracking is available in the top right corner; initial generation may take longer due to model loading but speeds up significantly afterward.
Reviewing and Refining Outputs
- A preview of generated videos is shown, demonstrating dialogue inclusion as part of the output. Users have options to refine prompts post-generation.
- Advanced Mode allows users to generate multiple variations from a single prompt by adjusting settings accordingly.
Starting with Images
- The process includes an option to start video creation using images. Users can upload media files they wish to animate.
- New prompts must be entered after selecting images; example given involves animating a scene with specific actions described in detail.
Accessing Generated Files
- To view all generated outputs, users navigate through File Explorer within the project folder structure where files are stored.