🎬 X2V-AI-Images-Videos-Skill — Free AI Image & Video Skill

中文说明: README_zh.md

Use Cases

Left: virtual boyfriend video (TTS + i2v/s2v). Middle: AI co-creation video from script. Right: feature overview (t2i, i2i, t2v, i2v, s2v, flf2v, TTS, voice clone).

Free AI image, video, TTS, and voice clone — use it for full AI comic drama, AI MV, AI digital-human voiceover, or deploy on OpenClaw as a virtual boyfriend/girlfriend. Powered by the open-source LightX2V free API; try the cloud at x2v.light-ai.top.
👋 Join our WeChat group! LightX2V Rotbot WeChat ID: random42seed

✨ Features

• 🖼️ Image — Text-to-image (t2i), image editing (i2i)
• 🎞️ Video — Text-to-video (t2v), image-to-video (i2v), digital human / talking head (s2v), first–last frame (flf2v)
• 🎤 TTS — Text-to-speech with preset voices
• 🎙️ Voice clone — Clone a voice from audio, then synthesize with it
• 💕 Virtual boyfriend/girlfriend (end-to-end complex case) — Persona (SOUL.md) → avatar (avatar/) → scene image (i2i/t2i) → TTS/talking video (s2v) or silent motion (i2v); see virtual-boyfriend-video.md

Use this skill for the cloud API. For restricted content or local unlimited inference, self-host by deploying models using LightX2V locally and running it as a service.

📋 Supported models (cloud)

Available via GET /api/v1/model/list. Use the model_cls value when submitting tasks:

Task	model_cls
t2i (text-to-image)	Qwen-Image-2512
i2i (image edit)	Qwen-Image-Edit-2511
t2v (text-to-video)	Wan2.2_T2V_A14B_distilled
i2v (image-to-video)	Wan2.2_I2V_A14B_distilled
s2v (digital human)	SekoTalk
flf2v (first-last frame)	Wan2.2_I2V_A14B_distilled
animate	wan2.2_animate

TTS and voice clone use separate endpoints (see SKILL.md); no model_cls needed.

📌 Requirements

• OpenClaw installed and configured
• LightX2V cloud access token — how to get it: Get token (Feishu doc) | x2v.light-ai.top
• Free tier: 100 task submissions per day

🚀 Install

Install from source (git + install.sh)

git clone https://github.com/xinyiqin/lightx2v_ai_generation_skills.git
cd lightx2v_ai_generation_skills
./install.sh

The installer will copy the skill to your OpenClaw workspace and prompt for your cloud token (optional at install time; you can set it later in openclaw.json).

⚙️ Configuration

Set in ~/.openclaw/openclaw.json under skills.entries.lightx2v.env (or in the environment):

Variable	Required	Description
LIGHTX2V_CLOUD_TOKEN	Yes	API token — see how to get token (Feishu)
LIGHTX2V_CLOUD_URL	No	Base URL (default: https://x2v.light-ai.top)

Example:

"skills": {
  "entries": {
    "lightx2v": {
      "enabled": true,
      "env": {
        "LIGHTX2V_CLOUD_TOKEN": "your-token",
        "LIGHTX2V_CLOUD_URL": "https://x2v.light-ai.top"
      }
    }
  }
}

Auto-load: The scripts (e.g. lightx2v_submit_and_poll.sh, tts_generate.sh) read LIGHTX2V_CLOUD_TOKEN and LIGHTX2V_CLOUD_URL from ~/.openclaw/openclaw.json when they are not already set in the environment, so you do not need to export them in the shell every time. Configure once in openclaw.json and run the scripts directly.

Ensure lightx2v is in skills.allow if you use an allowlist.

💡 Usage

After installation, the OpenClaw agent can use the skill to:

• Generate or edit images from text or reference images
• Generate videos (t2v, i2v, digital human s2v)
• Generate speech (TTS) or clone a voice from an audio sample

📚 Complex case examples

Step-by-step guides for multi-step workflows (e.g. digital-human-video, storyboard-video, my_boyfriend):

Example	Description	Doc
🎞️ First–last frame (flf2v)	First frame (t2i or user image) → i2i last frame → flf2v → short transition video	first-last-frame-video.md
🎬 Digital human video	Scene image + TTS → s2v → talking-head video	digital-human-video.md
🎥 One-sentence to video	t2i/i2i → one key image → i2v → short video	one-sentence-to-video.md
💕 Virtual boyfriend/girlfriend	Persona (SOUL.md) → avatar → scene image + TTS/s2v or i2v → companion video or voice; OpenClaw: set bot personality first	virtual-boyfriend-video.md

⚠️ Data & privacy

All data you submit (images, audio, text, generated outputs) is not used for secondary model training or commercial profit. It is only used to serve your requests for personal entertainment. This is to reduce infringement risk and to address concerns that your original characters (OCs) or assets might be used for AI blending or training elsewhere. You keep ownership of your inputs and outputs; use the service for your own creative and non-commercial purposes.

🗑️ Uninstall

Remove the skill directory and the lightx2v entry from openclaw.json:

rm -rf ~/.openclaw/workspace/skills/lightx2v
# Then edit ~/.openclaw/openclaw.json to remove skills.entries.lightx2v and lightx2v from skills.allow

📄 License

MIT