| name | lightx2v-ai-video-generation |
| description | Run LightX2V AI generation tasks from the command line via OpenAPI. Commands: lightx2v login, models, run, query, list, cancel, resume, delete, result, completion. Tasks: t2i, t2v, i2v, s2v, flf2v, t2av, i2av, vsr, animate, i2i. Use for: scriptable video/image generation, CI pipelines, agent automation, batch jobs. Triggers: lightx2v cli, lightx2v run, api key generation, text to video cli, text to image cli, download generated video, poll task status |
| allowed-tools | Bash(lightx2v *) |
Install skill: npx skills add ModelTC/LightX2V-Skills@lightx2v-ai-video-generation
Install CLI: curl -fsSL https://raw.githubusercontent.com/ModelTC/lightx2v-studio-cli/main/install.sh | sh
Skills directory: https://skills.sh/ModelTC/LightX2V-Skills
LightX2V AI Video Generation
Submit and download AI generation tasks via the LightX2V OpenAPI.
Online API docs: https://x2v.light-ai.top/api-docs
Quick Start
curl -fsSL https://raw.githubusercontent.com/ModelTC/lightx2v-studio-cli/main/install.sh | sh
lightx2v login
lightx2v models
lightx2v run t2i/Qwen-Image-2512 --prompt "a cute cat" --shape 512,512 -o out.png
Commands
| Command | Description |
|---|
lightx2v login [--base-url URL] | Save apikey_ to ~/.config/lightx2v/config.json |
lightx2v models [--json] | List task/model combinations |
lightx2v run <task>/<model_cls> ... | Submit, poll, download (main workflow) |
lightx2v query <task_id> [--json] | Query single task |
lightx2v list [--status] [--page] | Paginated task list |
lightx2v cancel <task_id> | Cancel running task |
lightx2v resume <task_id> | Resume failed/cancelled task |
lightx2v delete <task_id> | Delete finished task |
lightx2v result <task_id> [-o PATH] | Get result URL / download |
lightx2v completion bash or zsh | Shell completion script |
run options
| Flag | Description |
|---|
--prompt | Prompt text |
--input JSON or --input @file.json | Extra submit body fields |
--image / --video / --audio | Local file to base64 input |
--shape H,W | custom_shape e.g. 720,1280 |
--aspect-ratio | e.g. 16:9 |
--duration SEC | Seconds (t2av/i2av/s2v) |
--vsr-preset / --vsr-input-slot | VSR options |
-o PATH | Output file |
--quote | Pre-submit billing quote |
--json | Structured JSON event stream |
-q / --quiet | Minimal output |
--no-download | Return URL only |
Examples
Text-to-image
lightx2v run t2i/Qwen-Image-2512 --prompt "watercolor portrait" --aspect-ratio 16:9 -o portrait.png
Text-to-video
lightx2v run t2v/Wan2.2_T2V_A14B_distilled --prompt "waves at sunset" --shape 720,1280 -o beach.mp4
Image-to-video
lightx2v run i2v/Wan2.2_I2V_A14B_distilled --prompt "subject turns and smiles" --image ./ref.png -o animated.mp4
Digital human (s2v)
lightx2v run s2v/SekoTalk --prompt "natural speaking" --image ./portrait.png --audio ./speech.wav -o avatar.mp4
Task management
lightx2v query TASK_ID
lightx2v list --status SUCCEED --page 1 --page-size 20
lightx2v cancel TASK_ID
lightx2v result TASK_ID -o output.png
Notes
- API channel costs 2x web UI credits;
run prints quoted_credits.
- Config:
~/.config/lightx2v/config.json (mode 0600).
- Env:
LIGHTX2V_API_KEY, LIGHTX2V_CLOUD_API_KEY, LIGHTX2V_BASE_URL.
- Large media: prefer
url in --input JSON over base64.
Related
