Open Source
Open-Source AI Video & Avatar Projects
Plain-English guides to the open-source AI models and tools behind talking avatars, video generation, and voice — what they do, how to run them, and when a hosted tool is faster.
OpenMontage: The Open-Source AI Video Agent, Explained
A practical guide to OpenMontage, the open-source AI video agent that turns your AI coding assistant into a full video studio — what it does, how to set it up, what it costs, and when a hosted tool like VisionStory is the faster choice.
LongCat-Video-Avatar 1.5: We Ran the Open-Source Talking-Avatar Video Model
A hands-on teardown of LongCat-Video-Avatar 1.5, Meituan's open-source, MIT-licensed model that turns a photo plus audio into a talking-avatar video. We ran all three tasks on an A800 GPU, measured the real speed and VRAM, hit five setup traps, and lay out honestly when self-hosting makes sense and when a hosted tool wins.
Kokoro TTS: We Ran the 82M Open-Source Voice Model on a Laptop CPU
A hands-on teardown of Kokoro, the 82-million-parameter open-source text-to-speech model. We ran it on an Apple M3 Pro CPU, generated real voice samples you can play, hit the espeak-ng packaging bug (and fixed it), and lay out honestly when Kokoro is enough and when you need a hosted talking-video tool.
LivePortrait: The Open-Source Portrait Animation Model, Explained
A practical guide to LivePortrait, the open-source AI model that animates a still portrait from a driving video or audio — what it does, how to run it, and when a hosted talking-avatar tool like VisionStory is the faster choice.
