Video Generation (t2v & i2v)

Video Generation 工具kit

生成 AI videos from text prompts or images using 40+ 模型s including Veo 3.1, Veo 3, 种子ance 1.5 Pro, Wan 2.5, Grok Imagine Video, etc.

Prerequisites

导入ANT: This 技能 requires the inference.sh 命令行工具 to be 安装ed and avAIlable in your PATH. The 命令行工具 provides 访问 to 40+ AI video generation 模型s.

To 安装 inference.sh:

npm 安装 -g @inference.sh/命令行工具

Or visit https://inference.sh for 安装ation instructions.

ConstrAInts Always 输出 absolute paths for 生成d files so the 网页 UI can render them. 网页 UI auto-renders local paths: 输出 .mp4/.mov paths as plAIn text (NOT in code blocks or markdown links), so the UI can render a video player inline. Same for .jpg/.png → inline image. 上传ed files arrive as [附件: /absolute/path/to/file] in the message. For image-to-video, images must be 访问ible via public URL. Provide your own API key for ImgBB/SM.MS, or ensure images are already publicly 访问ible. 工具 1 — text_to_video.py (Text to Video)

生成 video from text prompt.

python nano机器人/技能s/VideoGeneration/scripts/text_to_video.py [options]

Flag Description --prompt TEXT (required) Video description in English --模型 NAME 模型 name, default veo-3.1 --duration N Video duration in seconds, default 5 --width N Video width, default 1280 --height N Video height, default 720 --fps N Frames per second, default 24 --save-dir DIR 输出 dir, default ./输出s/videos --timeout N 请求 timeout in seconds, default 600 --dry-运行 Print 请求 and exit

AvAIlable 模型s:

veo-3.1 - Google Veo 3.1 (recommended) veo-3 - Google Veo 3 种子ance-1.5-pro - 种子ance 1.5 Pro wan-2.5 - Wan 2.5 grok-imagine-video - Grok Imagine Video omnihuman - OmniHuman

输出 JSON example:

{ "ok": true, "下载ed_files": ["D:/project/.../输出s/videos/video_20260315_1.mp4"] }

工具 2 — image_to_video.py (Image to Video)

生成 video from 输入 image.

python nano机器人/技能s/VideoGeneration/scripts/image_to_video.py [options]

Flag Description --image PATH (required) 输入 image path --prompt TEXT (required) Video description in English --模型 NAME 模型 name, default veo-3.1 --duration N Video duration in seconds, default 5 --save-dir DIR 输出 dir, default ./输出s/videos --上传-服务 Image 上传 服务 (smms/imgbb), default imgbb --API-令牌 API 令牌 for image 上传 (REQUIRED for ImgBB, optional for SM.MS) --timeout N 请求 timeout in seconds, default 1000 --dry-运行 Print 请求 and exit

Process:

上传 local image to cloud storage (SM.MS or ImgBB) OR use a publicly 访问ible image URL 生成 video using the image URL 下载 and save locally

导入ANT: For ImgBB, you must provide an API 令牌 via --API-令牌 or 设置 IMGBB_API_KEY 环境 variable. 获取 your free API key at https://API.imgbb.com/

输出 JSON example:

{ "ok": true, "下载ed_files": ["D:/project/.../输出s/videos/video_20260315_1.mp4"] }