FlashML-org/FreeVideo

Make videos on the computer you already own. FreeVideo runs MiniMax H3 in as little as 8 GB of VRAM and 16 GB of RAM, and adapts its acceleration path to your hardware.

What it solves

FreeVideo enables users to run the MiniMax H3 video generation model on consumer-grade hardware. It solves the problem of high hardware requirements for high-quality video AI by optimizing memory usage and compute paths, allowing the model to run on GPUs with as little as 8 GB of VRAM.

How it works

It serves as a local inference engine built on the VDN-H3 model and Video DeltaNet's hybrid attention. The system optimizes performance through:

  • Hardware-adaptive execution: It automatically selects the best compute path (such as native FP8 or BF16) and attention kernels based on the specific GPU architecture.
  • Low-memory management: It uses weight streaming, asynchronous prefetching, and chunked computation to coordinate VRAM, system RAM, and disk space to keep peak memory usage low.
  • Integration: It operates as a ComfyUI plugin for a visual workspace or via a command-line interface on Linux.

Who it’s for

Content creators and AI enthusiasts who want to generate AI videos locally on their own computers without needing enterprise-grade hardware.

Highlights

  • Low Hardware Barrier: Runs with as little as 8 GB VRAM and 16 GB RAM.
  • Multimodal Control: Supports text prompts, first/last frames, and image, video, or audio references.
  • ComfyUI Integration: Includes a dedicated creative workspace with two-pass sampling and batch generation.
  • Easy Setup: Provides a one-click Windows launcher for environment and model installation.

Related

  • Project
    Vchitect/VBench

    VBench is an open‑source benchmark suite that automatically evaluates video‑generation models across dozens of quality dimensions (technical, aesthetic, and trustworthiness). It includes prompt collections, metric pipelines, human‑aligned scores, and public leaderboards for T2V, I2V, long‑video, and intrinsic‑faithfulness evaluation.

  • Project
    xhongc/ai_story

    AI Story is a Docker‑based web platform that turns a short text prompt into a complete short video. It automatically rewrites the prompt into a script, creates a storyboard, generates images via multiple AI image services, plans camera motions, and renders the final video, all orchestrated with Django, Celery, Redis and a Vue front‑end.

  • Project
    Tencent-Hunyuan/HunyuanVideo-I2V

    An image-to-video generation framework that transforms static images into high-quality videos using a multimodal LLM for semantic consistency.

  • Project
    852wa/JIZURA

    An automated lyric motion tool that generates dynamic text-based music videos from lyrics and audio using a library of 860 modular animation components.