SGLang-Diffusion Video Generation

Generate videos using a local SGLang-Diffusion server (Wan2.2, Hunyuan, FastWan, etc.). Use when: user asks to generate, create, or render a video with a loc...

Generate videos via a local SGLang-Diffusion server's OpenAI-compatible API.

Video generation is asynchronous and takes several minutes. The script handles submission, polling, and download automatically.

Prerequisites

• SGLang-Diffusion server running a video model (default: http://127.0.0.1:30000)
• Supported models: Wan2.2-T2V, Wan2.2-I2V, FastWan, Hunyuan
• If the server was started with --api-key, set SGLANG_DIFFUSION_API_KEY env var

Generate a video

python3 {baseDir}/scripts/generate_video.py --prompt "a curious raccoon exploring a garden"

Useful flags

python3 {baseDir}/scripts/generate_video.py --prompt "ocean waves at sunset" --size 1280x720
python3 {baseDir}/scripts/generate_video.py --prompt "city timelapse" --negative-prompt "blurry, low quality"
python3 {baseDir}/scripts/generate_video.py --prompt "dancing robot" --steps 50 --guidance-scale 7.5 --seed 42
python3 {baseDir}/scripts/generate_video.py --prompt "flying through clouds" --seconds 8 --fps 24 --out ./my-video.mp4
python3 {baseDir}/scripts/generate_video.py --prompt "flying through clouds" --server http://192.168.1.100:30000 --out ./my-video.mp4
python3 {baseDir}/scripts/generate_video.py --prompt "cat playing" --poll-interval 15 --timeout 1800
python3 {baseDir}/scripts/generate_video.py --prompt "animate this scene" --input-image /tmp/scene.png

API key (optional)

Only needed if the SGLang-Diffusion server was started with --api-key. Set SGLANG_DIFFUSION_API_KEY, or pass --api-key directly:

python3 {baseDir}/scripts/generate_video.py --prompt "hello" --api-key sk-my-key

Or configure in ~/.openclaw/openclaw.json:

{
 skills: {
   "sglang-diffusion-video": {
     env: { SGLANG_DIFFUSION_API_KEY: "sk-my-key" },
   },
 },
}

Notes

• The script prints a MEDIA: line for OpenClaw to auto-attach on supported chat providers.
• Output defaults to timestamped MP4 in /tmp/.
• Video generation typically takes 5-15 minutes depending on GPU and model size.
• Do not read the video back; report the saved path only.