lcy362 /agnes-video-generator

About agnes-video-generator

Feed it a text idea or a long article, and it builds a multi-scene video. TTS narration (a computer voice reading your script) and word-level SRT subtitles come built in. It runs on your laptop, but the model work happens in the cloud, so no GPU is needed. You'll need a free Agnes AI API key, and the video model still makes unstable frames.

Best for: Video makers who want a free, self-hosted pipeline from script to narrated short without a local GPU.

• Splits a prompt or a pasted article into scenes and renders one clip at a time
• Adds free TTS narration and word-level SRT subtitles automatically
• Makes text-to-video, image-to-video, keyframes animation and digital anchor clips
• Offers three scene-chaining modes and resumes from a checkpoint after a stop
• Exports 9:16, 16:9 or 1:1 clips up to 20s each, with unlimited scenes
• Rotates extra API keys on a 429 and polls progress over a REST API

1. 1
   Copy the template and add your key: cp .env.example .env, then edit AGNES_API_KEY
2. 2
   Add more keys as AGNES_API_KEY_2, AGNES_API_KEY_3 … (numbering must be contiguous) to raise the rate limit
3. 3
   Start it with start.sh, or run it with Docker or docker-compose
4. 4
   Or run the packaged version: npx free-short-video
5. 5
   Open the Web UI, pick a video mode, paste a prompt or manuscript, and generate