菜单
Convert images and thumbnails to short AI video clips using Kling, Hailuo, Luma, Runway, Veo, Sora, and 5+ more tools. Use when user asks to animate an image, create a video from a photo, or convert a thumbnail to video.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
| name | image-to-video |
| description | Convert images and thumbnails to short AI video clips using Kling, Hailuo, Luma, Runway, Veo, Sora, and 5+ more tools. Use when user asks to animate an image, create a video from a photo, or convert a thumbnail to video. |
| argument-hint | [image-path] |
Version: 1.0.0 | Updated: 2026-03-02 | Category: Content & Video
| Tool | Best Model (Mar 2026) | Max Length | I2V | Free Tier | Max Resolution | Native Audio | Best For |
|---|---|---|---|---|---|---|---|
| Runway | Gen-4.5 | 10s | Yes | 125 one-time credits (~25s Gen-4 Turbo) | 4K (upscale) | No | Cinematic consistency, character ref |
| Kling | Kling 3.0 / 2.6 Pro | 15s (3.0) / 10s (2.6) | Yes | 66 daily credits (360-540p, watermark) | 1080p (Master) | Yes (2.6+) | Motion control, product detail, fashion |
| Pika | Pika 2.5 | 10s | Yes | 80 monthly credits (480p, watermark) | 1080p+ (paid) | No | Creative effects (Pikaswaps, Pikadditions) |
| Luma | Ray3 / Ray3 Modify | 20s (720p+) | Yes | 30 gens/month (draft res, watermark) | 1080p | No | Long clips, start+end frame, cinematic |
| Sora | Sora 2 / Sora 2 Pro | 25s | Yes | None (Plus $20/mo minimum) | 1080p (Pro: 1792x1024) | Yes | Narrative scenes, physics, dialogue |
| Vidu | Vidu Q3 | 16s | Yes | 3 videos/month (720p, watermark) | 4K (Q3 Pro) | Yes (native) | Multi-shot sequences, synced audio |
| Hailuo/MiniMax | Hailuo 2.3 | 10s | Yes | Daily bonus credits (720p, watermark) | 1080p (paid) | Yes (2.6+) | Speed, social content, A/B testing |
| Google Veo | Veo 3.1 | 8s | Yes (Ingredients) | Limited (Gemini free: older model) | 4K (3840x2160) | Yes | 4K output, film language, camera control |
| Adobe Firefly | Firefly Video | 5s | Yes | Limited credits (with CC sub) | 2K native (up to 8K upscale) | No | Commercial-safe (IP indemnity), integration with CC |
| Seedance | Seedance 2.0 (ByteDance) | 15s | Yes | Free credits on signup | 1080p | Yes (native) | Multimodal input, fast generation |
| WAN | WAN 2.6 / 2.1 | 10s | Yes | Open source (run locally) | 1080p | No | Open source, self-hosted, general-purpose |
Current Models:
Image-to-Video Specifics:
Pricing:
Prompt Best Practices (Runway-Specific):
Sources: Runway Pricing | Gen-4 Research | Gen-4.5 Research
Current Models:
Image-to-Video Specifics:
Pricing:
Prompt Best Practices (Kling-Specific):
Sources: Kling AI | Kling 3.0 Guide | Kling 2.6 Motion Control
Current Models:
Key Features:
Pricing:
Prompt Best Practices (Pika-Specific):
Sources: Pika Pricing | Pika 2.5 Release
Current Models:
Image-to-Video Specifics:
Pricing:
Video Duration by Resolution:
Sources: Luma Pricing | Dream Machine | Ray3 Info
Current Models:
Image-to-Video Specifics:
Pricing:
Prompt Best Practices (Sora-Specific):
Sources: Sora 2 Guide | Sora Announcement
Current Models:
Key Features:
Pricing:
Sources: Vidu | Vidu Q3 Guide | Vidu Q3 on WaveSpeed
Current Models:
Pricing:
Prompt Best Practices (Hailuo-Specific):
Sources: Hailuo AI | MiniMax Hailuo 2.3
Current Models:
Key Features:
Pricing:
Prompt Best Practices (Veo-Specific):
Sources: Veo 3.1 4K Update | Veo 3.1 Blog | Google DeepMind Veo
Current Model: Firefly Video (Feb 2026)
Key Features:
Pricing:
Best For: Enterprise/agency use where IP indemnity matters. Integration with existing Adobe workflows.
Sources: Adobe Firefly Pricing | Firefly Blog
Current Model: Seedance 2.0
Key Features:
Pricing:
Sources: Seedance 2.0 | Seedance on fal.ai
Current Models:
Key Features:
Pricing: Free (open source). Hardware costs only.
Best For: Self-hosted workflows, privacy-sensitive projects, unlimited generation without credits
Sources: WAN GitHub | WAN on HuggingFace
[Camera movement], [pace/speed], [subject action], [environmental motion/details]
Example:
Slow push-in, steady cinematic pace, the developer's fingers type on the glowing keyboard,
holographic UI panels float and pulse with soft blue light around the workspace
For consistent cinematic outputs, cover these 8 elements:
| Element | What to Specify | Example |
|---|---|---|
| 1. Subject | Who/what is the focus | "A developer at a desk" |
| 2. Emotion/Mood | Tone of the scene | "focused, intense concentration" |
| 3. Optics/Framing | Shot type and lens | "medium close-up, 35mm lens" |
| 4. Motion | Camera + subject movement | "slow dolly in, subtle typing motion" |
| 5. Lighting | Light source and quality | "cool monitor glow, purple ambient neon" |
| 6. Style | Visual aesthetic | "cinematic, dark moody, tech noir" |
| 7. Audio (if supported) | Sound design | "mechanical keyboard clicks, ambient hum" |
| 8. Continuity | What stays constant | "face remains still, same expression" |
| Movement | Description | Prompt Keywords | Best For |
|---|---|---|---|
| Static/Locked | Camera stays still, subject moves | "static shot", "locked camera", "fixed frame" | Subtle expressions, product focus |
| Pan | Horizontal swivel from fixed point | "pan left", "pan right", "slow pan", "sweeping pan" | Revealing landscapes, following action |
| Tilt | Vertical angle up/down from fixed point | "tilt up", "tilt down", "slow tilt", "dramatic tilt" | Height emphasis, outfit reveal |
| Push-in/Dolly In | Camera moves toward subject | "push in", "dolly in", "slow push-in", "intimate push" | Building tension, product detail |
| Pull-back/Dolly Out | Camera moves away from subject | "pull back", "dolly out", "reveal pull-back" | Context reveal, scene endings |
| Truck | Camera moves parallel to subject | "truck left", "truck right", "lateral movement" | Walking scenes, shelf scanning |
| Pedestal | Camera moves vertically (elevator-like) | "pedestal up", "pedestal down", "rising reveal" | Revealing hidden elements |
| Tracking | Camera follows moving subject | "tracking shot", "follow shot", "match pace" | Action sequences, character walk |
| Orbit/Arc | Camera circles around subject | "orbit clockwise", "orbit counterclockwise", "arc around", "slow orbit" | Hero shots, product showcase, dramatic |
| Crane/Boom | Sweeping vertical + horizontal move | "crane up", "crane shot", "boom up", "sweeping crane" | Epic establishing shots, crowd reveals |
| Rack Focus | Focus shifts between planes | "rack focus", "shift focus", "focus pull" | Attention redirection |
| Crash Zoom | Very fast dramatic zoom | "crash zoom", "snap zoom", "whip zoom" | Action beats, comedy, emphasis |
| Zoom | Lens zooms in/out (not physical move) | "zoom in", "zoom out", "slow zoom" | Drawing attention, reveal |
| Handheld | Slight natural shake | "handheld", "shaky cam", "documentary style" | Realism, urgency, immediacy |
| FPV/First-Person | Camera IS the subject | "FPV", "first person view", "POV shot" | Immersive, gaming content |
You can combine camera movements for complex shots:
"Slow dolly in while panning slightly right, the camera rises gently"
"Crane up and orbit counterclockwise, revealing the full workspace"
"Tracking shot following the subject with a slight handheld shake"
| Modifier | Effect | Keywords |
|---|---|---|
| Very slow | Dreamy, contemplative | "very slow", "glacial pace", "barely moving" |
| Slow | Cinematic, elegant | "slow", "steady", "gentle", "smooth" |
| Medium | Natural, documentary | "natural pace", "moderate speed" |
| Fast | Energetic, dynamic | "fast", "dynamic", "brisk", "energetic" |
| Whip | Sudden, dramatic | "whip", "snap", "lightning fast", "sudden" |
| Level | Description | Keywords | Use Case |
|---|---|---|---|
| Minimal | Almost imperceptible | "barely perceptible movement", "very subtle", "still with micro-motion" | Thumbnails, portrait-style |
| Subtle | Natural idle motion | "gentle sway", "subtle breathing", "slight movement", "soft idle" | Professional headshots, calm scenes |
| Moderate | Clear but controlled | "natural movement", "smooth gesture", "controlled action" | Product demos, presentations |
| Dynamic | Active, energetic | "active movement", "energetic", "fluid motion" | Action scenes, sports |
| Dramatic | Maximum motion | "explosive motion", "dramatic action", "intense movement" | Music videos, trailers |
Hair/Clothing:
"hair gently moves as if from a light breeze"
"coat fabric ripples softly"
"scarf billows in the wind"
Eyes/Face (CAREFUL — most distortion-prone):
"eyes blink naturally"
"subtle smile forms"
"gaze shifts to the right"
WARNING: Keep facial animation minimal to avoid distortion. "Natural blinking" and "subtle expression" are safest.
Hands/Typing:
"fingers move across keyboard with natural rhythm"
"hands gesture subtly while speaking"
"subtle finger movement on the trackpad"
Environment/Background:
"particles float gently in the air"
"screen content scrolls slowly"
"ambient light pulses softly"
"clouds drift across the sky"
| Cause | Symptom | Fix |
|---|---|---|
| Re-describing image content in prompt | Reduced motion, visual artifacts | Prompt should ONLY describe motion, not the scene |
| Too many actions at once | Chaotic, incoherent motion | ONE hero action + hint secondary motion |
| Abstract/conceptual language | Unpredictable results | Use specific physical descriptions |
| Hidden limbs in source image | Extra fingers, hallucinated hands | Ensure all limbs visible in source |
| Wide-angle lens in source | Perspective distortion during motion | Use neutral focal length (35-85mm framing) |
| Too many adjectives | Quality degradation | 3-4 descriptive elements per component max |
| Mismatched aspect ratios | Stretching, cropping artifacts | Match source image to output aspect ratio |
| Excessive facial animation | Face warping, identity drift | Keep face motion minimal ("subtle", "natural blink") |
| Low-resolution source image | Blurry, unstable output | Use 720p+ source images minimum |
| Contradictory instructions | Confused model output | Review prompt for conflicts |
Place critical exclusions first (models weight earlier terms more):
"blurry, low resolution, distorted, warped face, extra fingers, glitchy text,
unnatural movements, chaotic cuts, morphing features, flickering"
AI video models struggle with text and logos. They frequently warp, blur, or morph text during motion. This is a fundamental limitation of current diffusion models.
Minimize motion near text areas:
"the text/logo remains perfectly stationary in the frame"
"camera movement avoids the text area"
"text stays sharp and readable throughout"
Use static camera for text-heavy areas: If text must be visible, keep the camera locked and animate only non-text elements.
Specify high contrast text:
"bold high-contrast text", "clear sans-serif text", "readable block letters"
Post-production approach (recommended for important text):
Kling 2.6 for logos: Best at preserving edges and logos in product shots.
Short duration helps: Text stays more stable in 3-5s clips than 10s+.
This section is specifically designed for thumbnails with: dark backgrounds, purple/teal neon lighting, desk/workspace scenes, MacBook/monitors, developer character, floating UI elements, holographic panels.
Recommended (in order of effectiveness):
Slow Push-In (BEST for thumbnails):
"Slow push-in toward the developer's workspace, steady cinematic pace,
ambient particles float gently, monitor screens glow softly"
Why: Creates focus, minimal distortion, keeps face stable.
Subtle Orbit (dramatic, good for hero shots):
"Very slow orbit clockwise around the developer at the desk,
neon reflections shift on the monitor surface, floating UI elements rotate gently"
Why: Adds depth and drama without disrupting the subject.
Static + Ambient Motion (safest for face stability):
"Static locked camera, the developer sits motionless at the desk,
holographic panels pulse with soft light, particles drift upward,
screen content scrolls slowly"
Why: Zero face distortion. All motion is environmental.
Gentle Pedestal Up (reveal shot):
"Slow pedestal up from the keyboard level, rising to reveal the developer's face
and floating holographic displays, purple ambient light pulses"
Dolly Back + Reveal:
"Slow dolly backward revealing the full workspace setup,
multiple monitors glow, floating code snippets hover in the air"
"Translucent holographic panels float around the workspace, slowly rotating and pulsing"
"Code snippets hover in mid-air with a soft cyan glow, gently bobbing up and down"
"Floating UI windows orbit the developer, each displaying different data visualizations"
"Semi-transparent screens drift slowly, reflecting purple and blue neon light"
"Holographic interface elements materialize one by one around the desk"
Key descriptors for floating elements:
"Holographic panels emit a soft pulsating blue-purple glow"
"Screens pulse rhythmically with cyan light, intensity rising and falling"
"Neon edges of the floating panels flicker with electric energy"
"Warm glow radiates from the holographic displays, casting colored light on the developer's face"
"Panels glow brighter momentarily before dimming back, creating a breathing light effect"
Light behavior keywords:
"Fingers move naturally across the backlit keyboard, screen content scrolls upward"
"Subtle typing motion, code appearing on the main monitor line by line"
"The MacBook screen displays scrolling code with a soft green-on-black terminal"
"Cursor blinks on the screen, new lines of code appear gradually"
"Multiple monitors show different live data — one scrolling code, one showing metrics"
This is the #1 challenge. Here is the priority order:
Face = STATIC, Everything else = MOVING:
"The developer sits perfectly still, face unchanged, steady gaze at the screen.
Around him, holographic panels pulse with light, particles float upward,
keyboard keys glow softly, ambient neon light shifts between purple and blue"
Minimal face motion only:
"The developer blinks naturally, otherwise perfectly still.
Ambient environment has floating particles, pulsing lights, and drifting UI elements"
Identity locks: Always include: "same face throughout, consistent facial features, no face morphing"
Camera choice matters:
Tool selection for face stability:
Slow push-in, smooth cinematic pace. A developer sits at a dark workspace,
face perfectly still with natural subtle blinking. Holographic UI panels float
around the desk, pulsing with soft [purple/blue/cyan] neon light. Particles
drift gently upward. The MacBook screen displays scrolling code. Ambient
lighting shifts subtly between [purple and teal]. The background remains
perfectly static while floating elements orbit slowly.
Slow orbit clockwise around [product], smooth cinematic pace. The [product]
sits on a dark reflective surface. Ambient particles float in the air.
Soft studio lighting highlights edges and details. The background is
[dark/gradient]. Camera completes a quarter rotation over [5-10] seconds.
Static locked camera, [duration] seconds. [Person description] stands/sits
in [environment]. Face remains perfectly still. Hair moves gently from a
subtle breeze. Atmospheric particles float in the background. Volumetric
light rays stream through [window/source]. The mood is [cinematic/dramatic/calm].
[Camera movement], steady pace. Close-up of a MacBook Pro screen showing
[code editor/terminal/dashboard]. Code scrolls upward naturally. The cursor
blinks. Ambient keyboard glow pulses softly. Shallow depth of field blurs
the background workspace. [Monitor reflections shift/bokeh lights drift].
Slow crane up from [starting point], sweeping reveal. Rising from [desk level/
ground level] to reveal [full scene/workspace/city]. Atmospheric fog drifts
through the scene. [Neon/ambient] lights illuminate the environment.
The scale of the scene becomes apparent as the camera rises.
| Parameter | Type | Required | Description |
|---|---|---|---|
| source_image | Image (720p+ recommended) | Yes | The still image to animate |
| prompt | String | Yes | Motion description (camera + subject + environment) |
| tool | Enum | Yes | Which AI tool to use |
| duration | Integer (3-25s) | No | Target video length |
| aspect_ratio | Enum (16:9, 9:16, 1:1, 4:3) | No | Output aspect ratio |
| resolution | Enum (480p-4K) | No | Output resolution (tool-dependent) |
| motion_reference | Video | No | Motion reference video (Kling Motion Control only) |
| Parameter | Type | Description |
|---|---|---|
| video | MP4 | Generated video file |
| duration | Integer | Actual video length in seconds |
| resolution | String | Output resolution |
| credits_used | Integer | Credits consumed |
| Skill | How |
|---|---|
thumbnail-generator | Generate thumbnail image -> animate with this skill |
nano-banana-images | Generate source image -> animate |
video-edit | Post-process: trim, add text overlay, music |
pan-3d-transition | Combine I2V clips with 3D transitions |
title-variants | Generate titles -> overlay on video |
recreate-thumbnails | Face-swap source image -> animate |
Need 4K output?
-> Google Veo 3.1 or Vidu Q3 Pro
Need face stability (thumbnail/portrait)?
-> Kling Motion Control (best) or Static camera on any tool
Need native audio?
-> Sora 2, Vidu Q3, Seedance 2.0, Kling 2.6+, or Veo 3.1
Need free / no budget?
-> Kling free (66 daily credits) or Hailuo free (daily bonus)
-> WAN 2.1 (open source, run locally)
Need commercial IP safety?
-> Adobe Firefly (IP indemnity)
Need creative effects (swaps, additions)?
-> Pika 2.5 (Pikaswaps, Pikadditions, Pikaframes)
Need longest output?
-> Sora 2 (25s) or Luma Ray3 (20s) or Vidu Q3 (16s)
Need product/fashion detail?
-> Kling 2.6 Pro (preserves edges, logos, fabric)
Need fast iteration?
-> Hailuo 2.3 (speed) or Runway Gen-4 Turbo (cheap credits)
Need multi-shot sequences?
-> Vidu Q3 (Smart Cuts) or Veo 3.1 (Ingredients)
Need open source / self-hosted?
-> WAN 2.1/2.6 (GitHub, HuggingFace)
Last updated: 2026-03-02 | Research covers tools available as of early March 2026
DEPRECATED. Koyeb removed its free tier. Use github-actions-deploy or n8n-workflow-deploy instead.
Deploy any automation to run on schedule using GitHub Actions. Use when user asks to deploy with cron, schedule an automation, run code automatically, or deploy for free.
Deploy serverless Python functions and AI workloads to Modal. Use when user needs GPU, AI inference, serverless Python, or scheduled AI tasks.
Deploy automations as n8n workflows — visual, webhook-driven, multi-step. Use when user asks to create a workflow, connect services visually, or deploy to n8n.
Deploy frontends, dashboards, webhooks, and API endpoints to Vercel. Use when user needs a web UI, dashboard, webhook receiver, or Next.js/React frontend.
Add 6-layer guardrail pipeline to any AI agent — RBAC, injection defense, output filtering, monitoring