$9.90 Pack
Reverse-Engineered Prompts for Kling, Wan 2.1, Sora & AvatarSay

Convert Any Reference Video into Shot-by-Shot AI Prompts

Single-frame image-to-prompt tools miss camera motion, subject trajectory, and temporal lighting shifts. ShotBreakdown feeds 3-frame Start/Mid/End Contact Sheets per cut into Gemini 2.5 VLM to write dense, motion-accurate English video prompts for every shot.

Or drag & drop an MP4 / MOV file — Max 100 MB · ≤3 min (10 credits/min · Free trial ≤45s)
Browse MP4 / MOV
✓ Gemini 2.5 VLM Calibrated (99.6% accuracy) ✓ Source video auto-deleted after breakdown ✓ One-time credits · Never expire
▼ Live Interactive Sample Output — AI Cinematic Cyberpunk Sci-Fi Micro-Trailer
Explore all 6 industry autopsies →
AI Video · Kling / Wan Prompt Pack 5 Shots · 12s · 864×486 (16:9) · 30 fps Calibrated by google/gemini-2.5-flash

AI Cinematic Cyberpunk Sci-Fi Micro-Trailer

3-Second Hook Analysis: The opening shot immediately grabs attention with a close-up of a person wearing futuristic neon glasses, set against a vibrant, low-light cyberpunk city. This visual intrigue is amplified by the OCR text 'NEO-KYOTO // SECTOR 09', hinting at a compelling sci-fi narrative.

00.00s / 12.00s
Avg Cut Pacing
2.4s (fast-cut)
Audio Tempo & Sync
99.4 BPM · 0% beat-cut
SFX Hits (BandIt v2)
6 triggers
Manual Prep Saved
~1.2 hrs
Interactive Shot Timeline (Click any cut to seek & inspect) 0.0s → 12s
♪ Beat & SFX Timeline (Dark, atmospheric synthwave with a driving beat and futuristic sound effects.) impact@4.17s · whoosh@7.16s · whoosh@9.57s · whoosh@11.06s · impact@11.37s · whoosh@11.67s
AI Video: 4 shots (80%) Code / Remotion: 1 shots (20%)
AI Video 98% conf
✦ VLM Calibrated: Offline guessed live_action → Fixed to ai_video
Forensic Evidence · Arbiter: FLASH “NEO-KYOTO // SECTOR 09”

This shot is clearly AI-generated due to the stylized, flawless rendering of the subject and environment, lacking the subtle imperfections of live-action footage. The offline label erred by classifying it as live_action.

Unnatural skin texture and lack of fine pores, typical of AI generation. Perfectly smooth, almost plastic-like reflections on the neon glasses. Slightly inconsistent facial features across frames, indicating AI interpolation. The 'NEO-KYOTO' text is perfectly integrated but lacks real-world optical distortion.
Framing & Angle CU · Eye-level
Camera Motion Slow push-in with a subtle right pan
Lighting Setup Neon-lit cityscape with strong magenta and cyan rim lighting, low-key fill
Gear / Engine Est. Kling 2.0 / Sora
Reverse-Engineered AI Video Prompt Best on: Kling 2.0 · 16:9

A cinematic close-up of a young person with short, dark hair, wearing futuristic, glowing magenta neon visor glasses. They are looking slightly to the right, with a stoic expression. The background is a blurred, vibrant cyberpunk city street at night, filled with neon signs in blues, purples, and pinks. The camera slowly pushes in and subtly pans right, emphasizing the glow of the glasses and the urban atmosphere. Cinematic, 8K, volumetric lighting, shallow depth of field.

AI Video 97% conf
✦ VLM Calibrated: Offline guessed code_ui_motion → Fixed to ai_video
Forensic Evidence · Arbiter: FLASH

Despite the offline label of code_ui_motion, this is an AI-generated video of a human-like character, evidenced by the artificial skin texture and stylized visual effects. It's not programmatic UI.

The subject's face exhibits AI-typical smoothness and lack of micro-detail. The 'glitter' effect on the cheeks appears painted on rather than naturally reflective. Background bokeh is too perfect and lacks real-world lens characteristics. Subtle shifts in facial symmetry and ear shape across frames.
Framing & Angle CU · Eye-level
Camera Motion Static locked, subtle head tilt
Lighting Setup Soft frontal key light with neon pink and blue background bokeh, subtle orange glow on headphones
Gear / Engine Est. Wan 2.1 / Sora
Reverse-Engineered AI Video Prompt Best on: Wan 2.1 · 16:9

A cinematic close-up of a young woman with dark hair, wearing large black headphones and a dark jacket with subtle orange neon accents. Her cheeks are adorned with glowing orange, pixelated glitter. She looks directly at the camera with a calm, intense gaze. The background is a soft, out-of-focus bokeh of colorful cyberpunk city lights. The shot is static, focusing on her expression and the intricate details of her face and glowing elements. Cinematic, high detail, volumetric lighting.

AI Video 98% conf
✦ VLM Calibrated: Offline guessed live_action → Fixed to ai_video
Forensic Evidence · Arbiter: FLASH “WHEN MEMORY BECOMES CODE”

This shot is an AI-generated video, not live-action, due to the artificial digital effects seamlessly integrated onto the face and the overall stylized, rendered appearance of the subject and environment. The offline label erred.

The 'digital glitch' effect on the face is too uniform and lacks organic interaction with skin. Skin texture is smooth and lacks the imperfections of real human skin. Background elements are stylized and lack the nuanced depth of live-action. Facial features remain consistent but lack micro-expressions, suggesting AI generation.
Framing & Angle MCU · Eye-level
Camera Motion Slow push-in
Lighting Setup Dramatic low-key lighting with blue and magenta neon reflections, digital glitch overlay on face
Gear / Engine Est. Kling 2.0 / Sora
Reverse-Engineered AI Video Prompt Best on: Kling 2.0 · 16:9

A cinematic medium close-up of a young man with short dark hair, wearing a dark hooded jacket. Digital blue and white glitching particles appear to be forming and dissolving on the left side of his face. He looks directly at the camera with a serious, intense expression. The background is a blurred, dark cyberpunk street with subtle neon lights. The camera slowly pushes in, enhancing the dramatic effect of the digital transformation. Cinematic, high contrast, volumetric lighting.

AI Video 97% conf
✦ VLM Calibrated: Offline guessed code_ui_motion → Fixed to ai_video
Forensic Evidence · Arbiter: FLASH

This shot is clearly an AI-generated fantasy character, not programmatic UI, due to its highly stylized aesthetic, perfect rendering, and lack of real-world physics. The offline label erred.

The character's features are highly stylized and perfectly symmetrical, indicative of AI art. Hair strands are too uniform and lack natural frizz or flyaways. The staff held by the character has an unnaturally smooth, rendered texture. The background is painterly and lacks photographic realism.
Framing & Angle MCU · Eye-level
Camera Motion Slow right pan with subtle push-in
Lighting Setup Soft, ethereal lighting with a gentle glow, naturalistic fantasy atmosphere
Gear / Engine Est. Wan 2.1 / Midjourney Video
Reverse-Engineered AI Video Prompt Best on: Wan 2.1 · 16:9

A cinematic medium close-up of a beautiful fantasy elf woman with long, flowing white hair, piercing blue eyes, and delicate facial features. She wears intricate, ornate jewelry around her neck and ears, and holds a detailed, ancient-looking staff. Her expression is serene and wise. The background is a softly blurred, enchanted forest with dappled sunlight. The camera slowly pans right and subtly pushes in, revealing more of her intricate attire and the magical setting. Fantasy art style, high detail, ethereal lighting.

Code / Remotion 99% conf
Forensic Evidence · Arbiter: FLASH “CHRONO ECHOES — AI Micro-Series”

This shot is definitively code_ui_motion, characterized by its crisp typography, geometric design, and programmatic animation, perfectly matching the offline heuristic.

Sharp, vector-based typography and geometric shapes. Clean, programmatic animation of text and bounding boxes. Consistent color palette and precise alignment, typical of UI. Absence of any organic textures or camera artifacts.
Framing & Angle WS · Eye-level
Camera Motion Code spring slide, text reveal
Lighting Setup Flat UI canvas with subtle glow effects
Gear / Engine Est. Remotion 4 / After Effects
Reverse-Engineered AI Video Prompt Best on: Remotion React · 16:9

A clean, minimalist title card animation. The text 'KLING 2.0 + WAN 2.1' appears first in a small, glowing purple rectangle. Below it, 'CHRONO ECHOES' animates in with a subtle spring effect, followed by 'AI Micro-Series · Coming This Fall' fading in. All text is white, set against a dark, deep purple background. The overall aesthetic is sleek, futuristic, and programmatic, with precise timing and smooth transitions.

Inspect Remotion Component & Spring Physics Spec (JSON)
{
  "componentLayout": "Vertical stack of text elements within a dark purple bounding box, centered on screen.",
  "springPhysics": {
    "fps": 60,
    "damping": 15,
    "stiffness": 100
  },
  "paletteHexes": [
    "#1A0033",
    "#FFFFFF",
    "#9933FF",
    "#6600CC"
  ],
  "textAnimationKeyframes": [
    {
      "text": "KLING 2.0 + WAN 2.1",
      "animation": "fade_in_slide_up",
      "delayMs": 0
    },
    {
      "text": "CHRONO ECHOES",
      "animation": "spring_scale_in",
      "delayMs": 200
    },
    {
      "text": "AI Micro-Series · Coming This Fall",
      "animation": "fade_in",
      "delayMs": 500
    }
  ]
}
Edit in ChatCut ↗
Copied to clipboard!

Why Teams Switch to Automated Shot Forensics

Replace manual frame-scrubbing and guesswork with deterministic cut detection + multimodal Gemini 2.5 VLM calibration.

Old Manual Workflow

Feeding a single screenshot into ChatGPT only describes a static photo—completely missing push-ins, whip pans, and physics motion.

With ShotBreakdown

3-keyframe Start / Mid / End Contact Sheet analysis captures true camera movement, lens focal feel, motion intensity, and negative prompts.

Old Manual Workflow

Dumping a 30-second multi-cut video into a generic chat model yields one vague paragraph instead of per-shot generation prompts.

With ShotBreakdown

Deterministic FFmpeg scene-cut splitting isolates every cut first, giving you an exact, copy-ready prompt per shot.

Old Manual Workflow

Wasting expensive Kling or Sora credits trying to lip-sync dialogue shots that belong in a dedicated talking-character engine.

With ShotBreakdown

Automatically detects Avatar / 3D Cartoon Talking shots and gives you a 1-click remake shortcut to AvatarSay (no monthly subscription).

How to Reverse-Engineer AI Video Prompts from Any Reference Video

01

Upload Reference Clip or Paste Social Link

Drop an AI trailer, commercial, or reference reel (MP4/MOV <= 100 MB) or paste a Shorts/Reels/TikTok URL.

02

Scene Splitting + 3-Frame Contact Sheet VLM Calibration

Each shot is sliced cleanly and mapped onto a Start/Mid/End keyframe grid so Gemini 2.5 VLM observes exact subject motion and lighting evolution.

03

Copy English Prompt + Negative Prompt into Kling, Wan, or Sora

Click "Copy Prompt" on any shot card—complete with aspect ratio, motion intensity tag, and recommended generator—or export all prompts in VIDEO.md.

One-Time Credit Packs · Zero Subscriptions

Test 1 clip (≤45s) free after sign-in, then pay once for credits that never expire.

Try Before Paying

Free Trial

$0 · Sign in once
  • ✓ Browse all 6 live breakdowns without login
  • ✓ 1 free video slot (≤45s) after sign-in
  • ✓ Full cut timeline + audio BPM beat grid
  • ✓ Full VLM forensics & AI prompts on Shots #1–#3
  • — Shots #4+ blurred · VIDEO.md / PDF / SRT locked
Most Popular

Creator Pack

$9.90 one-time · never expires

300 Credits (~30 short videos at 10 credits/min)

  • ✓ 300 non-expiring credits (10 credits / min, ~30 short videos)
  • ✓ Direct TikTok, Instagram Reels & YouTube Shorts URL paste
  • ✓ Full-clip Gemini VLM forensics & copy-ready English AI prompts
  • ✓ Unlock all exports: VIDEO.md, Shot List PDF, SRT & Remotion JSON
  • ✓ 30-day cloud report history + 1-click public share links
  • ✓ 100% automatic credit refund if any job fails
Save 27% · Team & Ad Spy

Agency Pack

$29.00 one-time · never expires

1,200 Credits (~120 short videos at 10 credits/min)

  • ✓ 1,200 non-expiring credits (~120 short videos, $0.24 / min)
  • ✓ Everything in Creator Pack included
  • ✓ Batch Ad Spy Queue: analyze & compare up to 10 URLs at once
  • ✓ White-label Client PDF Shot List export (zero ShotBreakdown watermark)
  • ✓ Priority serverless GPU & VLM inference queue
  • ✓ 100% automatic credit refund guarantee
Have a Reviewer, Partner, or Launch Promo Code? Redeem non-expiring credits & full export access instantly after signing in.

Frequently Asked Questions

Which AI video generators are the reverse-engineered prompts tuned for? +

Each shot prompt is structured in clean cinematic English (Subject + Action Trajectory + Camera Movement + Lens & Lighting + Atmosphere) alongside a Negative Prompt, working directly in Kling 2.0, Wan 2.1, OpenAI Sora, Runway Gen-4, Seedance, and Pollo.ai.

Does Video-to-Prompt work on live-action videos too, not just AI clips? +

Yes! Creators frequently upload real cinema or commercial B-roll to get accurate camera-movement and lighting prompts so they can recreate a similar aesthetic in AI video generators.

How much does it cost to extract prompts from a 60-second video? +

Just 10 credits (~$0.33 on the $9.90 Creator Pack, or ~$0.24 on the $29.00 Agency Pack). Plus, your first <=45s clip after signing in is free to test.