Agent Skills
› SamurAIGPT/Generative-Media-Skills
› muapi-freeze-effect-video
muapi-freeze-effect-video
GitHub基于参考图片生成电影级“时间静止”特效视频。用户指定场景、冻结瞬间细节及角色,模型利用图像到视频技术实现主角穿越静止世界并恢复时间的视觉效果。
Trigger Scenarios
需要制作具有时间暂停和恢复特效的创意短视频
希望将静态人物照片转化为动态电影感视频
Install
npx skills add SamurAIGPT/Generative-Media-Skills --skill muapi-freeze-effect-video -g -y
SKILL.md
Frontmatter
{
"name": "muapi-freeze-effect-video",
"slug": "muapi-freeze-effect-video",
"version": "1.0.0",
"description": "Generate a cinematic \"freeze effect\" video where time stops mid-scene, the subject walks through the frozen world, then time resumes with a snap.",
"acceptLicenseTerms": true
}
Freeze Effect Video
Generate a cinematic "freeze effect" video where time stops mid-scene, the subject walks through the frozen world, then time resumes with a snap.
Inputs
| Name | Type | Required | Default | Description |
|---|---|---|---|---|
photo |
image_url | yes | — | Reference photo of the main character (face/body) whose appearance must remain consistent across all shots. |
scene |
text | no | a packed, dimly lit sports bar with neon accents, blurred TVs showing a championship celebration | The setting where time will freeze (e.g. "a busy nightclub", "a wedding reception", "a stadium crowd"). |
freeze_moment |
text | no | golden arcs of beer suspended midair, popcorn floating motionless, people frozen mid-cheer | The signature frozen visual — what hangs in the air when time stops. |
closing_line |
text | no | perfect | A short word/phrase the character whispers before resuming time. |
aspect_ratio |
text | no | 16:9 | Output aspect ratio — "16:9" for cinematic, "9:16" for vertical/social. |
duration |
text | no | 15 | Target duration in seconds (10 or 15 recommended). |
Steps
Phase A — Generate the Freeze Effect Video
Submit the plan with ONE step:
- Freeze effect video —
muapi video from-image(model=seedance-v2.0-i2v, quality=high, generate_audio=true):- Reference image:
{{photo}} - Aspect ratio:
{{aspect_ratio}} - Duration:
{{duration}} - Resolution:
720p - Prompt:
Ultra-realistic, shot on Arri Alexa Mini, 35mm lens, {{scene}}, volumetric haze, dynamic hard shadows, shallow depth of field. Main Character: The person from @image1. Facial features and body proportions remain consistent across all shots. [0:00–0:03] {{scene}}. Time flows normally. Steadicam frontal medium shot tracking the person walking confidently through the crowd. The crowd erupts in euphoria around them. As the person walks, they raise their right hand and snap. [0:03–0:06] At the snap, a subtle spherical shockwave bursts from their fingertips — air distortion and light refraction ripple outward. Everything freezes mid-motion. {{freeze_moment}}. Neon light catches dust and liquid in the air. Absolute silence. [0:06–0:09] Only the person moves. Soft, echoing footsteps. Camera tracks backward as they walk through the frozen scene, observing calmly. They reach out and pluck a single suspended detail from the air. [0:09–0:11] They stop in front of a frozen onlooker, face locked in an ecstatic expression. The person tilts their head, gently observes. Softly: "{{closing_line}}". [0:11–0:15] They turn, face the camera, smirk, and snap again. A stronger reverse shockwave ripples outward. Motion instantly resumes — sounds explode back, people land mid-jump. The person walks away as the camera pushes through the celebrating crowd. Fade to black. Sound Design: Deafening ambient celebration → snap → deep shockwave/bass drop → absolute silence → footsteps → soft whisper → snap → reverse shockwave → deafening celebration returns.
- Reference image:
After generation, present the video to the user.
Notes
- The character's identity is anchored by
@image1— the reference photo must clearly show the main subject's face. - For best results, choose a
scenewith crowd energy and lots of small objects (liquid, confetti, sparks) that can plausibly "freeze" in midair. - If the model rejects realistic human likeness, switch to the Global tier model (
seedance-2-image-to-video-fast) which allows looser identity matching. - For vertical/social delivery, set
aspect_ratio=9:16and keepduration=10for tighter pacing. - Always keep
generate_audio=true— the snap → silence → snap audio arc is the signature of this effect and should not be muted.
Trigger Keywords
freeze effect, time freeze video, time stop video, snap freeze, frozen world, bullet time crowd, time stand still, pause time effect
Notes for the Executing Agent
- This recipe is LLM-orchestrated: read each phase, gather any missing inputs from the user, then call
muapiCLI commands. Usemuapi auth configurefirst ifMUAPI_API_KEYis unset. - For model IDs without a CLI alias yet, fall back to the raw endpoint via
curl -X POST https://api.muapi.ai/api/v1/<endpoint> -H "x-api-key: $MUAPI_API_KEY" -H 'content-type: application/json' -d '{...}'and poll withmuapi predict wait <request_id>. - Substitute
{{input_name}}placeholders with the user's actual inputs before issuing each call.
Version History
- d4ce293 Current 2026-07-25 11:20


