Image to Video
Convert still images to animated videos via inference.sh CLI.
Quick Start
CODEBLOCK0
Install note: The install script only detects your OS/architecture, downloads the matching binary from dist.inference.sh, and verifies its SHA-256 checksum. No elevated permissions or background processes. Manual install & verification available.
Model Selection
| Model | App ID | Best For | Motion Style |
|---|
| Wan 2.5 i2v | INLINECODE1 | Realistic motion, natural movement | Photorealistic, subtle |
| Seedance 1.5 Pro |
bytedance/seedance-1-5-pro | Stylized, creative, animation-like | Artistic, expressive |
|
Seedance 1.0 Pro |
bytedance/seedance-1-0-pro | General purpose, good quality | Balanced |
|
Fabric 1.0 |
falai/fabric-1-0 | Cloth, fabric, liquid, flowing materials | Physics-based flow |
|
Grok Imagine Video |
xai/grok-imagine-video | General animation, text-guided | Versatile |
When to Use Each
| Scenario | Best Model | Why |
|---|
| Landscape with water/clouds | Wan 2.5 i2v | Best at natural, realistic motion |
| Portrait with subtle expression |
Wan 2.5 i2v | Maintains face fidelity |
| Product with fabric/cloth |
Fabric 1.0 | Specialized in material physics |
| Flag waving, curtain flowing |
Fabric 1.0 | Cloth simulation |
| Illustrated/artistic image |
Seedance | Matches stylized content |
| General "bring to life" |
Seedance 1.5 Pro | Good all-rounder |
| Quick test/iteration |
Seedance 1.0 Lite | Fastest, 720p |
Motion Types
Camera Movement
| Movement | Prompt Keyword | Effect |
|---|
| Push in / Dolly forward | "slow dolly forward", "camera pushes in" | Increasing intimacy/focus |
| Pull out / Dolly back |
"camera pulls back", "slow zoom out" | Reveal, context |
| Pan left/right | "camera pans slowly to the right" | Scanning, following |
| Tilt up/down | "camera tilts upward" | Revealing height |
| Orbit | "camera orbits around the subject" | 3D exploration |
| Crane up | "camera rises upward" | Grand reveal |
| Static | (no camera movement prompt) | Subject motion only |
Subject Motion
| Type | Prompt Examples |
|---|
| Natural elements | "water rippling", "clouds drifting", "leaves rustling in breeze" |
| Hair/clothing |
"hair blowing gently in wind", "dress fabric flowing" |
| Atmospheric | "fog slowly rolling", "dust particles floating in light beams" |
| Character | "person slowly turns to camera", "subtle breathing motion" |
| Mechanical | "gears turning", "clock hands moving" |
| Liquid | "coffee steam rising", "paint dripping", "water pouring" |
Prompting Best Practices
The Golden Rule: Subtle > Dramatic
AI video models produce better results with gentle, subtle motion than dramatic action. Requesting too much movement causes distortion and artifacts.
CODEBLOCK1
Prompt Structure
CODEBLOCK2
Examples by Scenario
CODEBLOCK3
Duration Guidelines
| Duration | Quality | Use For |
|---|
| 2-3 seconds | Highest quality | GIFs, looping backgrounds, cinemagraphs |
| 4-5 seconds |
High quality | Social media posts, product reveals |
| 6-8 seconds | Good quality | Short clips, transitions |
| 10+ seconds | Quality degrades | Avoid unless stitching shorter clips |
Extending Duration
For longer videos, generate multiple short clips and stitch:
CODEBLOCK4
The Full Workflow
Still-to-Final-Video Pipeline
CODEBLOCK5
Cinemagraph Effect
A cinemagraph is a still photo where only one element moves (e.g., waterfall moving in an otherwise frozen scene). To achieve this:
- 1. Generate the still image with the motion element clearly defined
- Prompt for motion only in that specific element
- Keep to 2-4 seconds for seamless looping
CODEBLOCK6
Common Mistakes
| Mistake | Problem | Fix |
|---|
| Too much motion requested | Distortion, artifacts, warping | Subtle > dramatic, always |
| Wrong model for content type |
Poor results | Use selection guide above |
| Clips too long (10s+) | Quality degrades significantly | Keep to 3-5 seconds, stitch if needed |
| No camera movement specified | Random/unpredictable motion | Always specify camera behavior |
| Conflicting motion directions | Chaotic, unnatural | One primary motion direction |
| Low-res source image | Low-res video output | Start with highest quality source |
| Complex action scenes | Models can't handle | Keep motion simple and natural |
Related Skills
CODEBLOCK7
Browse all apps: INLINECODE6
图像转视频
通过 inference.sh CLI 将静态图像转换为动画视频。
快速开始
bash
curl -fsSL https://cli.inference.sh | sh && infsh login
生成静态图像
infsh app run falai/flux-dev-lora --input {
prompt: 日落时分的宁静山间湖泊，雪峰倒映在平静的水面，黄金时刻的光线，风景摄影,
width: 1248,
height: 832
}
将其动画化
infsh app run falai/wan-2-5-i2v --input {
prompt: 湖面泛起轻柔涟漪，云朵缓缓飘移，暖光渐变，远处有飞鸟,
image: path/to/lake-image.png
}
安装说明： 安装脚本 仅检测您的操作系统/架构，从 dist.inference.sh 下载匹配的二进制文件，并验证其 SHA-256 校验和。无需提升权限或后台进程。提供手动安装与验证。
模型选择
| 模型 | 应用 ID | 最佳用途 | 运动风格 |
|---|
| Wan 2.5 i2v | falai/wan-2-5-i2v | 逼真运动，自然动作 | 照片级真实，细腻 |
| Seedance 1.5 Pro |
bytedance/seedance-1-5-pro | 风格化、创意、类动画 | 艺术感，表现力强 |
|
Seedance 1.0 Pro | bytedance/seedance-1-0-pro | 通用，质量优秀 | 均衡 |
|
Fabric 1.0 | falai/fabric-1-0 | 布料、织物、液体、流动材质 | 基于物理的流动 |
|
Grok Imagine Video | xai/grok-imagine-video | 通用动画，文本引导 | 多功能 |
各场景适用模型
| 场景 | 最佳模型 | 原因 |
|---|
| 含水面/云朵的风景 | Wan 2.5 i2v | 自然逼真运动表现最佳 |
| 含细微表情的人像 |
Wan 2.5 i2v | 保持面部保真度 |
| 含织物/布料的产品 |
Fabric 1.0 | 专精于材质物理 |
| 旗帜飘扬、窗帘飘动 |
Fabric 1.0 | 布料模拟 |
| 插画/艺术风格图像 |
Seedance | 匹配风格化内容 |
| 通用赋予生命 |
Seedance 1.5 Pro | 全能型 |
| 快速测试/迭代 |
Seedance 1.0 Lite | 最快，720p |
运动类型
摄像机运动
| 运动方式 | 提示关键词 | 效果 |
|---|
| 推近 / 向前推轨 | 缓慢向前推轨，摄像机推近 | 增强亲密感/聚焦 |
| 拉远 / 向后推轨 |
摄像机拉远，缓慢拉远 | 揭示，提供背景 |
| 左右摇摄 | 摄像机缓慢向右摇摄 | 扫描，跟随 |
| 上下俯仰 | 摄像机向上仰拍 | 揭示高度 |
| 环绕 | 摄像机环绕主体 | 3D 探索 |
| 升降 | 摄像机向上升起 | 宏大揭示 |
| 静态 | (无摄像机运动提示) | 仅主体运动 |
主体运动
| 类型 | 提示示例 |
|---|
| 自然元素 | 水面涟漪，云朵飘移，树叶在微风中沙沙作响 |
| 头发/衣物 |
头发在风中轻轻飘动，裙摆面料飘动 |
| 氛围 | 雾气缓缓弥漫，光束中漂浮的尘埃颗粒 |
| 角色 | 人物缓缓转向镜头，微弱的呼吸起伏 |
| 机械 | 齿轮转动，时钟指针移动 |
| 液体 | 咖啡热气升腾，颜料滴落，水在流淌 |
提示词最佳实践
黄金法则：细腻 > 剧烈
AI 视频模型在温和、细腻的运动下效果优于剧烈动作。要求过多运动会导致失真和伪影。
❌ 人物奔跑跳跃越过障碍物，摄像机同时旋转
✅ 人物缓慢向前行走，微风轻拂，摄像机跟随侧拍
❌ 爆炸，碎片四处飞溅
✅ 烛光轻轻摇曳，温暖的环境光渐变
❌ 快速变焦到眼睛，伴随剧烈摄像机抖动
✅ 缓慢向前推轨靠近主体，微妙的焦点变化
提示词结构
[摄像机运动] + [主体运动] + [氛围效果] + [情绪/节奏]
各场景示例
bash
风景动画
infsh app run falai/wan-2-5-i2v --input {
prompt: 摄像机缓慢向右摇摄，水面倒映移动的云朵，树木在微风中轻轻摇摆，温暖的金色光线，宁静而缓慢,
image: landscape.png
}
人像动画
infsh app run falai/wan-2-5-i2v --input {
prompt: 微弱的呼吸起伏，头部轻微转动，自然的眨眼，头发轻轻飘动，柔和的环境光渐变,
image: portrait.png
}
产品展示动画
infsh app run bytedance/seedance-1-5-pro --input {
prompt: 产品缓慢360度环绕，聚光灯柔和移动，微妙的反射变化，高端产品展示，流畅运动,
image: product.png
}
织物/布料动画
infsh app run falai/fabric-1-0 --input {
prompt: 布料在微风中飘动起伏，自然的布料物理，柔和运动,
image: fabric-scene.png
}
建筑可视化
infsh app run falai/wan-2-5-i2v --input {
prompt: 缓慢向前推轨穿过入口，摄像机轻微向上仰拍，环境光透过窗户过滤，光束中的尘埃颗粒,
image: building-interior.png
}
时长指南
| 时长 | 质量 | 用途 |
|---|
| 2-3 秒 | 最高质量 | GIF、循环背景、动态照片 |
| 4-5 秒 |
高质量 | 社交媒体帖子、产品展示 |
| 6-8 秒 | 良好质量 | 短片、转场 |
| 10+ 秒 | 质量下降 | 除非拼接短片，否则避免 |
延长时长
如需更长视频，生成多个短片并拼接：
bash
从同一图像生成 3 个片段，运动逐步推进
infsh app run falai/wan-2-5-i2v --input {
prompt: 缓慢向左摇摄，柔和的水面运动,
image: scene.png
} --no-wait
infsh app run falai/wan-2-5-i2v --input {
prompt: 继续摇摄，云朵移动，光线变化,
image: scene.png
} --no-wait
拼接
infsh app run infsh/media-merger --input {
media: [clip1.mp4, clip2.mp4]
}
完整工作流
静态图像到最终视频流水线
bash
1. 生成源图像（最佳质量）
infsh app run bytedance/seedream-4-5 --input {
prompt: 电影感风景，黎明时分的雾中山峦，前景湖泊，戏剧性云彩，黄金时刻，4K 画质，专业摄影,
size: 2K
}
2. 将图像动画化
infsh app run falai/wan-2-5-i2v --input {
prompt: 薄雾缓缓穿过山谷，湖面泛起涟漪，云朵缓慢移动，远处有飞鸟，温暖光线渐变,
image: landscape.png
}
3. 如有需要，提升视频分辨率
infsh app run falai/topaz-video-upscaler --input {
video: animated-landscape.mp4
}
4. 添加环境音效
infsh app run infsh/hunyuanvideo-foley --input {
video: animated-landscape.mp4,
prompt: 轻柔的自然环境声，远处的鸟鸣，微风，水波拍打
}
5.
