How to Create Free AI Videos Locally with ComfyUI

TL;DR

Install ComfyUI Desktop locally, select a video workflow, download the required LTX 2.3 model files, enter a descriptive prompt, and click Run. The local setup generates text-to-video and image-to-video content without API keys, subscriptions, usage credits, or per-video charges, although the demonstrated model downloads require about 40 gigabytes of storage.

Transcript

Today, I’ll show you how to generate AI  videos for free on your own PC using ComfyUI.   We'll install everything from scratch,  generate videos from simple text prompts,   animate existing images, and by the end of this  video, you'll have everything running on your PC   with no API keys, subscriptions, or usage credits.  I'm Kevin, and let's dive... Read More

Key Insights

• ComfyUI is a free, open-source application that runs AI models directly on a local computer. Once installed, it can generate content without per-video charges and also supports images, image editing, audio, 3D models, and large language models.
• A Windows PC with an NVIDIA RTX graphics card is the recommended configuration for following the tutorial. The desktop version runs locally, while the cloud alternative is suggested for users without compatible hardware and requires a paid subscription.
• Free local video models can be found by opening Video, filtering the Runs On setting to ComfyUI, and selecting a task such as text to video. Available choices shown include LTX 2.3, WAN 2.2, Hunyuan, Kandinsky, and SVD.
• LTX 2.3 is selected because it is fast, produces strong results, and runs across a wide range of RTX graphics cards. Other supported models remain available for later experimentation and comparison based on the needs of a project.
• The LTX 2.3 workflow initially reports five missing models because its files have not been downloaded. Together, these files require about 40 gigabytes of available storage, but they only need to be downloaded once before generating videos.
• A descriptive text prompt typically improves generated results. The workflow also exposes width and height for resolution, duration for video length, and frame rate for motion smoothness, although the tutorial keeps the default settings for its first example.
• Generated videos are automatically saved by the workflow. The Assets section provides a central place to preview previous outputs, save a selected video elsewhere on the computer, or delete content that is no longer needed.
• Image-to-video generation starts with an existing image and a prompt describing the desired motion. ComfyUI also provides workflows for video editing, camera control, inpainting, upscaling, and additional AI tasks, with options to customize and combine workflows later.
• More videos from this channel:
  • How to use AI for work: save time and analyze data
  • How to use Midjourney to create stunning AI art?
  • How to choose the best AI chatbots right now
• More from this channel: Kevin Stratvert

Questions & Answers

Q: How do you install ComfyUI for local AI video generation?

Download the ComfyUI Desktop installer from the referenced website, run it, and launch the application after installation. Select the local option so processing occurs directly on the computer, agree to the terms, and continue. Give the installation a name, such as ComfyUI, leave the remaining settings unchanged for a typical setup, and proceed to the home screen.

Q: What computer setup is recommended for running ComfyUI video models?

A Windows PC equipped with an NVIDIA RTX graphics card is recommended because it is presented as the easiest and most reliable configuration for following the tutorial. ComfyUI Desktop runs the models directly on that computer. If compatible local hardware is unavailable, the cloud version is offered as an alternative, but the transcript states that it requires a paid subscription.

Q: How do you find free local video models in ComfyUI?

Open the Video section from the ComfyUI home screen, then use the Runs On dropdown and select ComfyUI. This filter limits the gallery to models that run on the local computer instead of paid cloud services. Next, open the task filter and select text to video or image to video, depending on the type of content you want to create.

Q: Why does the tutorial use LTX 2.3 for AI video generation?

LTX 2.3 is used because the tutorial describes it as fast, capable of producing excellent results, and compatible with a wide range of RTX graphics cards. It supports both the demonstrated text-to-video and image-to-video workflows. ComfyUI also lists alternatives, including WAN 2.2, Hunyuan, Kandinsky, and SVD, which users can test later to compare their results.

Q: How much storage does the LTX 2.3 workflow require?

The demonstrated LTX 2.3 workflow reports five missing models that together require about 40 gigabytes of storage. Users should confirm that sufficient disk space is available before selecting Download All. ComfyUI then retrieves the files needed by the workflow and provides a download progress view. These model files only need to be downloaded once, according to the tutorial.

Q: How do you generate a video from a text prompt in ComfyUI?

Select the LTX 2.3 text-to-video template and download all missing models. In the main generation node, enter a detailed description of the desired scene. Adjust width, height, duration, and frame rate if needed, or keep their default values. Click Run in the upper-right corner, wait for processing to finish, and then preview the completed video in the workflow.

Q: How do you animate an existing image with ComfyUI?

Return to Templates, clear the text-to-video task filter, and select image to video. Choose LTX 2.3, which can reuse the models already downloaded for the earlier workflow. Drag and drop an existing image into the input area, then enter a prompt describing the motion you want. Keep or adjust the available settings, click Run, and preview the resulting animation.

Q: Where are generated ComfyUI videos saved and managed?

The workflow automatically saves each completed video. Open Assets from the left side of ComfyUI to view and preview generated content. From that area, right-click a video to save it to another location on the computer, or use the available delete option to remove it. This provides one place to review and manage outputs created during local generation.

Summary & Key Takeaways

• ComfyUI Desktop is a free, open-source application that runs AI models directly on a computer. The tutorial recommends a Windows PC with an NVIDIA RTX graphics card, selects the local installation option, and keeps the default settings. Local processing avoids API keys, subscriptions, usage credits, and per-video charges.

• The text-to-video process starts in the Video templates gallery by filtering models that run on ComfyUI and selecting the text-to-video task. LTX 2.3 is chosen for its speed, results, and compatibility with many RTX cards. Its five missing model files require about 40 gigabytes of storage.

• The LTX 2.3 workflow accepts a descriptive prompt and provides controls for video width, height, duration, and frame rate. Generated videos are automatically saved and managed through Assets. A related image-to-video workflow accepts an uploaded image plus a motion prompt, while other templates support editing, camera control, inpainting, and upscaling.