# Video breakdown report: final.mp4

> English translation of [report.md](report.md), the Markdown report rea-video generated for this video. Section order and content follow the tool output. Text recognized on screen (the "text track" lines) is kept in the original Chinese, since that is what appears in the frame.

**Verdict**: mixed production: 3D rendering 100%; overlays: text labels, captions, title (composited in code / an editor); audio: voice + BGM at 110.0 BPM (library music) + 5 sound effects

- Duration 66.542 s · 1080×1920 · 24.0 fps · 9:16
- 21 shots, 3.17 s average; transitions: cut×20, start×1
- Confidence (duration-weighted): 98%; method: deterministic forensics fused with a vision model (VLM)
- VLM active (gemini-3.1-pro-preview, gemini-3.1-pro-preview-pc, gcp-gemini-3-pro-preview)

![storyboard](storyboard.jpg)

## Production method share

| Underlying picture | Duration | Share |
|---|---|---|
| 3D rendering | 66.55 s | 100% |

## Remake plan

**21 3D-rendered shots → 10 generation units, 10 Seedance 2.5 calls**

- The default engine, Seedance 2.0 (including Mini / Fast), cannot produce several shots in one call; with it you would generate the 21 shots one by one. This plan assumes Seedance 2.5 (≤30 s per call, multi-shot in one call): 10 calls in total
- Grouping rules: AI shots in the same set joined by hard cuts are merged into one generation (S01); ≤30 s per call; consecutive hard cuts inside a unit should stay ≤8 s, at most 10 s (S06); split at dissolves / fades and set changes (S03)
- Audio: the model can generate action sound effects and ambience along with the picture; voice-over is better done as separate TTS and mixed in, so you can change the script and timing

### Generation units

| Unit | Shots | Time | Duration | Method | Continuity |
|---|---|---|---|---|---|
| G1 | #1 | 0.0–8.9 s | 8.9 s | single shot | First frame: generate an image with the composition of the original shot 1 first frame (shots/shot_001_first.jpg) |
| G2 | #2–#3 | 8.9–15.8 s | 7.0 s | multi-shot in one call | First frame: the last frame of G1, continuing the action (S05) |
| G3 | #4–#6 | 15.8–24.3 s | 8.5 s | multi-shot in one call | First frame: the last frame of G2, continuing the action (S05) |
| G4 | #7–#9 | 24.3–29.5 s | 5.1 s | multi-shot in one call | First frame: generate an image with the composition of the original shot 7 first frame (shots/shot_007_first.jpg) |
| G5 | #10 | 29.5–30.6 s | 1.2 s | single shot | First frame: generate an image with the composition of the original shot 10 first frame (shots/shot_010_first.jpg) |
| G6 | #11–#12 | 30.6–38.0 s | 7.4 s | multi-shot in one call | First frame: generate an image with the composition of the original shot 11 first frame (shots/shot_011_first.jpg) |
| G7 | #13 | 38.0–41.6 s | 3.5 s | single shot | First frame: the last frame of G6, continuing the action (S05) |
| G8 | #14–#16 | 41.6–49.6 s | 8.0 s | multi-shot in one call | First frame: generate an image with the composition of the original shot 14 first frame (shots/shot_014_first.jpg) |
| G9 | #17–#18 | 49.6–57.2 s | 7.6 s | multi-shot in one call | First frame: the last frame of G8, continuing the action (S05) |
| G10 | #19–#21 | 57.2–66.5 s | 9.3 s | multi-shot in one call | First frame: the last frame of G9, continuing the action (S05) |

**G1 · 3D rendering · shot #1 · 8.9 s**

- Reference material: `@image1` character reference (one frontal + one half-body per character) to keep faces consistent across shots; `@video1` the original 0.0–8.9 s segment as a reference for motion, camera and cut rhythm (reference only, do not copy the picture)
- First-frame image prompt: 3D cartoon style, a yellow fluffy little animal in a minimal green courtyard, a cracked-open salted duck egg in front of it leaking red oil, soft light, centered composition.
- Alternative route: a 3D project. Have an LLM write a Blender script or a Three.js scene, model the characters / props once and render each shot from its own camera; best for multiple angles and character consistency, and changing copy or shots is just a code change (R08)

```text
Vertical 9:16; frontal soft light (flat), soft light (almost no shadow edge), neutral (about 5000K), background clearly blurred (shallow depth of field / portrait mode). Character per @image1. Follow @video1 for motion, camera and cut rhythm.
3D animation, a yellow fluffy little animal cracks open a salted duck egg, red oil gushes out of the yolk, a brown fluffy little animal next to it watches in surprise, static camera, soft light.
```

<details><summary>English prompt (as generated by the tool)</summary>

```text
3D animation, a yellow furry little animal cracks open a salted duck egg, red oil gushes out from the yolk, a brown furry little animal watches in surprise nearby, static camera, soft lighting.
```

</details>

**G2 · 3D rendering · shots #2, #3 · 7.0 s**

- Reference material: `@image1` character reference (one frontal + one half-body per character) to keep faces consistent across shots; `@video1` the original 8.9–15.8 s segment as a reference for motion, camera and cut rhythm (reference only, do not copy the picture)
- First-frame image prompt: 3D cartoon style, three cute animal characters with fluffy fur stand around a white bowl holding a small clay pot. Simple geometric hill shapes in the background, solid-color backdrop. Soft top light, bright colors.
- Alternative route: a 3D project. Have an LLM write a Blender script or a Three.js scene, model the characters / props once and render each shot from its own camera; best for multiple angles and character consistency, and changing copy or shots is just a code change (R08)

```text
Vertical 9:16; frontal soft light (flat), soft light (almost no shadow edge), neutral (about 5000K), background clearly blurred (shallow depth of field / portrait mode). Character per @image1. Follow @video1 for motion, camera and cut rhythm. Hard cuts between shots.
Shot 1 (0.0–4.8 s): three fluffy cartoon animal characters stand in the frame, then suddenly turn into three white fur balls and roll out to the right together, leaving the bowl and pot in the middle. Static camera.
Shot 2 (4.8–7.0 s): static camera. In the center is a wooden bowl full of white powder. Several white balls with tiny eyes pop out of the powder one after another and rest on the surface, light and bouncy with an elastic settle.
```

<details><summary>English prompt (as generated by the tool)</summary>

```text
Shot 1 (0.0-4.8s): Three fluffy cartoon animal characters stand in the frame, then they suddenly turn into three white fur balls and roll out to the right side of the frame together, leaving the bowl and pot in the middle. Static camera.
Shot 2 (4.8-7.0s): Static camera. In the center is a wooden bowl filled with white powder. Several small white spheres with tiny eyes pop up sequentially from the powder, resting on the surface with a bouncy, elastic animation.
```

</details>

**G3 · 3D rendering · shots #4, #5, #6 · 8.5 s**

- Reference material: `@image1` character reference (one frontal + one half-body per character) to keep faces consistent across shots; `@video1` the original 15.8–24.3 s segment as a reference for motion, camera and cut rhythm (reference only, do not copy the picture)
- First-frame image prompt: 3D cartoon style, macro photography, a brown cylindrical flowerpot in the center with a white fluffy ball showing inside, a fluffy cartoon monkey peeking in from the right, soft solid-color background, Blender 3D render look, soft light
- Alternative route: a 3D project. Have an LLM write a Blender script or a Three.js scene, model the characters / props once and render each shot from its own camera; best for multiple angles and character consistency, and changing copy or shots is just a code change (R08)

```text
Vertical 9:16; frontal soft light (flat), soft light (almost no shadow edge), neutral (about 5000K), background clearly blurred (shallow depth of field / portrait mode). Character per @image1. Follow @video1 for motion, camera and cut rhythm. Hard cuts between shots.
Shot 1 (0.0–3.8 s): static eye-level shot, 3D cartoon scene, the fluffy object in the pot stays still, the little monkey plush on the right blinks its big eyes and watches it, tiny screws hover and drift slightly in the background, a smooth, clean 3D-rendered animation throughout
Shot 2 (3.8–5.2 s): 3D animation, a few tiny salt grains fly in from the upper right toward a pore in the shell of the duck-egg model, static camera.
Shot 3 (5.2–8.5 s): 3D educational animation, camera at eye level, fixed, pulling back slightly; on the desk a transparent cylinder of white liquid on the left, a cylinder of translucent yellow liquid on the right, a round semi-permeable membrane model with a wooden rim floating in between; several smooth white balls pass through the membrane ring from left to right at a smooth, even pace.
```

<details><summary>English prompt (as generated by the tool)</summary>

```text
Shot 1 (0.0-3.8s): Static eye-level shot, 3D cartoon scene, a fluffy white ball inside a brown cup, a cute furry monkey character on the right looking at it and blinking, soft studio lighting, Blender 3D render style.
Shot 2 (3.8-5.2s): 3D animation, tiny salt particles fly from the top right towards a pore on a duck egg model, static camera.
Shot 3 (5.2-8.5s): 3D scientific educational animation, camera slowly pulls back. On a clean desktop, a transparent cylinder with white liquid stands on the left, another cylinder with yellow liquid on the right, and a circular semi-permeable membrane frame suspended in the middle. Small smooth white spheres float and pass smoothly through the membrane from left to right, soft studio lighting.
```

</details>

**G4 · 3D rendering · shots #7, #8, #9 · 5.1 s**

- Reference material: `@image1` character reference (one frontal + one half-body per character) to keep faces consistent across shots; `@video1` the original 24.3–29.5 s segment as a reference for motion, camera and cut rhythm (reference only, do not copy the picture)
- First-frame image prompt: 3D cartoon style, close-up of a salted duck egg cut open, translucent egg white wrapping a perfectly round orange yolk, a few round salt grains at the edge, clean solid-color background, soft studio lighting, C4D render, minimalist, high quality.
- Alternative route: a 3D project. Have an LLM write a Blender script or a Three.js scene, model the characters / props once and render each shot from its own camera; best for multiple angles and character consistency, and changing copy or shots is just a code change (R08)

```text
Vertical 9:16; frontal soft light (flat), soft light (almost no shadow edge), neutral (about 5000K), background clearly blurred (shallow depth of field / portrait mode). Character per @image1. Follow @video1 for motion, camera and cut rhythm. Hard cuts between shots.
Shot 1 (0.0–1.8 s): the camera pushes in slowly and steadily toward the orange yolk in the center, keeping the 3D cartoon look, lighting unchanged.
Shot 2 (1.8–3.5 s): 3D cartoon style, close-up, a salted duck egg cut open, a perfect orange spherical yolk in the middle wrapped in translucent egg white. Over time, blue droplet-shaped particle effects gradually appear in the egg white around the yolk. Camera stays fixed.
Shot 3 (3.5–5.1 s): static close-up, a 3D cross-section showing water being drawn out of the yolk, several translucent light-blue arrows flow outward from the yolk surface, the yolk firms up and contracts slightly, clean CG animation look throughout.
```

<details><summary>English prompt (as generated by the tool)</summary>

```text
Shot 1 (0.0-1.8s): The camera slowly and steadily pushes in towards the orange egg yolk in the center, maintaining the 3D cartoon texture and consistent lighting.
Shot 2 (1.8-3.5s): 3D cartoon style, close-up shot, a cross-section of a salted duck egg with a perfect orange spherical yolk in the center, surrounded by translucent egg white. Over time, blue water droplet-like particle effects gradually appear in the egg white around the yolk. The camera remains static.
Shot 3 (3.5-5.1s): Static macro 3D animation showing cross-section of an egg. Semi-transparent light blue arrows flow outward from the central yellow yolk representing osmotic dehydration, clean lighting.
```

</details>

**G5 · 3D rendering · shot #10 · 1.2 s**

- Reference material: `@image1` character reference (one frontal + one half-body per character) to keep faces consistent across shots; `@video1` the original 29.5–30.6 s segment as a reference for motion, camera and cut rhythm (reference only, do not copy the picture)
- First-frame image prompt: 3D render, a cute fluffy cartoon monkey stands on the left of the frame looking at a floating fluffy white ball on the right, a simple light-green geometric wall and a brown jar in the background, soft studio lighting bringing out the fur texture, Pixar style, vertical composition.
- Generate 4 s and trim to 1.2 s (very short shots are unstable when generated on their own)
- Alternative route: a 3D project. Have an LLM write a Blender script or a Three.js scene, model the characters / props once and render each shot from its own camera; best for multiple angles and character consistency, and changing copy or shots is just a code change (R08)

```text
Vertical 9:16; frontal soft light (flat), soft light (almost no shadow edge), neutral (about 5000K), background clearly blurred (shallow depth of field / portrait mode). Character per @image1. Follow @video1 for motion, camera and cut rhythm.
3D animation, a cute fluffy cartoon monkey stands on the left looking at a floating fluffy white ball on the right, the monkey breathes gently, the white ball bobs slightly, fixed camera, soft light.
```

<details><summary>English prompt (as generated by the tool)</summary>

```text
3D animation, a cute furry cartoon monkey standing on the left, looking at a floating furry white ball on the right. The monkey has slight breathing movements, the white ball floats slightly, fixed camera, soft lighting.
```

</details>

**G6 · 3D rendering · shots #11, #12 · 7.4 s**

- Reference material: `@image1` character reference (one frontal + one half-body per character) to keep faces consistent across shots; `@video1` the original 30.6–38.0 s segment as a reference for motion, camera and cut rhythm (reference only, do not copy the picture)
- First-frame image prompt: 3D render, close-up, a cutaway model of the inside of a duck egg. A smooth yellow sphere in the center stands for the yolk, wrapped in a translucent white shell, resting on a ring-shaped base of beige woven material. Light-green solid background, soft studio light, cartoon educational style, clean and bright.
- Alternative route: a 3D project. Have an LLM write a Blender script or a Three.js scene, model the characters / props once and render each shot from its own camera; best for multiple angles and character consistency, and changing copy or shots is just a code change (R08)

```text
Vertical 9:16; frontal soft light (flat), soft light (almost no shadow edge), neutral (about 5000K), background clearly blurred (shallow depth of field / portrait mode). Character per @image1. Follow @video1 for motion, camera and cut rhythm. Hard cuts between shots.
Shot 1 (0.0–3.1 s): 3D animation, fixed camera. A cutaway duck-egg model in the center of the frame; the yellow sphere (yolk) in the middle shrinks smoothly and proportionally over time while everything else stays still.
Shot 2 (3.1–7.4 s): several white fluffy ball creatures start out scattered on a wooden tabletop, then move toward the middle and huddle tightly together, and a few small orange balls pop out of them; camera stays fixed.
```

<details><summary>English prompt (as generated by the tool)</summary>

```text
Shot 1 (0.0-3.1s): 3D animation, fixed camera. In the center of the frame is a cross-section model of a duck egg structure. The yellow sphere (egg yolk) in the center gradually and smoothly shrinks in scale over time, while the other parts remain stationary.
Shot 2 (3.1-7.4s): Several white fluffy little ball creatures are initially scattered on a wooden table. Then they move towards the center and huddle tightly together. Next, several small orange spheres pop out from them. The camera remains static.
```

</details>

**G7 · 3D rendering · shot #13 · 3.5 s**

- Reference material: `@image1` character reference (one frontal + one half-body per character) to keep faces consistent across shots; `@video1` the original 38.0–41.6 s segment as a reference for motion, camera and cut rhythm (reference only, do not copy the picture)
- First-frame image prompt: 3D cutaway render, a cooked salted duck egg model cut open, a round orange-yellow yolk in the center, translucent gel-like egg white and a slice of shell outside, minimal soft studio lighting, centered composition, refined educational illustration style
- Alternative route: a 3D project. Have an LLM write a Blender script or a Three.js scene, model the characters / props once and render each shot from its own camera; best for multiple angles and character consistency, and changing copy or shots is just a code change (R08)

```text
Vertical 9:16; frontal soft light (flat), soft light (almost no shadow edge), neutral (about 5000K), background clearly blurred (shallow depth of field / portrait mode). Character per @image1. Follow @video1 for motion, camera and cut rhythm.
Static close-up of a 3D cutaway salted duck egg; glistening golden oil slowly seeps out of the top of the orange spherical yolk and gathers and flows, egg white and shell stay still, soft studio top light, a steady view of the oil separating out
```

<details><summary>English prompt (as generated by the tool)</summary>

```text
Static close-up 3D animation of a cross-sectioned salted duck egg, golden oil droplets slowly oozing and gathering on top of the smooth orange yolk, soft studio overhead lighting, realistic fluid simulation, educational animation style.
```

</details>

**G8 · 3D rendering · shots #14, #15, #16 · 8.0 s**

- Reference material: `@image1` character reference (one frontal + one half-body per character) to keep faces consistent across shots; `@video1` the original 41.6–49.6 s segment as a reference for motion, camera and cut rhythm (reference only, do not copy the picture)
- First-frame image prompt: 3D minimal educational render style, smooth gray background and light tabletop, a light gray-blue square towel with fine fuzz hangs in the center of the frame, threaded on a thin horizontal metal rod, soft even lighting, clean 3D model
- Alternative route: a 3D project. Have an LLM write a Blender script or a Three.js scene, model the characters / props once and render each shot from its own camera; best for multiple angles and character consistency, and changing copy or shots is just a code change (R08)

```text
Vertical 9:16; frontal soft light (flat), soft light (almost no shadow edge), neutral (about 5000K), background clearly blurred (shallow depth of field / portrait mode). Character per @image1. Follow @video1 for motion, camera and cut rhythm. Hard cuts between shots.
Shot 1 (0.0–3.7 s): fixed eye-level camera, 3D animation demo: a square towel with fine fuzz in the center is pushed inward by metal rods from left and right and compressed into a thin T shape; as it is squeezed, several crystal-clear water drops form in the air on both sides of the towel and fall; the background stays minimal and still
Shot 2 (3.7–5.8 s): 3D cartoon style, three cute fluffy little animals stand side by side, green in the middle, brown and gray on the sides, simple green geometric hills and a brown jar in the background. Fine yellow particles keep falling from the top of the frame. Static camera
Shot 3 (5.8–8.0 s): 3D cartoon style, close-up, a salted duck egg model cut open sits in a white container, the yolk orange-yellow. Over time more oil spills out over the top of the yolk, with twinkling star effects on the oil surface; camera stays fixed.
```

<details><summary>English prompt (as generated by the tool)</summary>

```text
Shot 1 (0.0-3.7s): Static eye-level shot, 3D educational animation: a light blue textured towel block suspended on a metal rod is squeezed inward from both sides by the rod mechanism, compressing into a tight T-shape, releasing transparent water droplets falling down on both sides, minimal studio background
Shot 2 (3.7-5.8s): 3D cartoon style, three fluffy cute little animals standing side by side, the middle one is green, the ones on the sides are brown and gray, the background is simple green geometric hills and a brown jar. Small yellow particles are continuously falling from the top of the screen. Static camera
Shot 3 (5.8-8.0s): 3D cartoon style, close-up shot, a cut salted duck egg model placed in a white container, the yolk is orange-yellow. Over time, more oil gradually overflows from the top of the yolk, with twinkling star effects on the oil surface, the camera remains fixed.
```

</details>

**G9 · 3D rendering · shots #17, #18 · 7.6 s**

- Reference material: `@image1` character reference (one frontal + one half-body per character) to keep faces consistent across shots; `@video1` the original 49.6–57.2 s segment as a reference for motion, camera and cut rhythm (reference only, do not copy the picture)
- First-frame image prompt: 3D cartoon style, vertical. A fluffy green monkey character on the left, a small brown pot on a white base on the right, the pot open at the top showing a yellow yolk. Simple light-green geometric hill shapes in the background. Soft, even light overall.
- Alternative route: a 3D project. Have an LLM write a Blender script or a Three.js scene, model the characters / props once and render each shot from its own camera; best for multiple angles and character consistency, and changing copy or shots is just a code change (R08)

```text
Vertical 9:16; frontal soft light (flat), soft light (almost no shadow edge), neutral (about 5000K), background clearly blurred (shallow depth of field / portrait mode). Character per @image1. Follow @video1 for motion, camera and cut rhythm. Hard cuts between shots.
Shot 1 (0.0–5.4 s): 3D animation. The green fluffy character on the left stays still. On the right, what was a white fur ball instantly turns into a small brown pot. The pot opens at the top and a stream of pink liquid winds down from above in an S curve, landing on the yellow yolk inside the pot. Fixed camera.
Shot 2 (5.4–7.6 s): 3D animation, three fluffy monkeys; the middle one raises both hands to lift the bowl and salted duck egg on its head, looking happy, the two monkeys beside it sway slightly, camera fixed.
```

<details><summary>English prompt (as generated by the tool)</summary>

```text
Shot 1 (0.0-5.4s): 3D animation. A green fuzzy character on the left remains still. On the right, a white fuzzy ball instantly transforms into a small brown pot. The top of the pot opens, and a stream of pink liquid meanders down from above in an S-shape, landing on the yellow yolk inside the pot. Fixed camera.
Shot 2 (5.4-7.6s): 3D animation, three furry monkey characters, the middle monkey raises its hands holding a bowl and a salted duck egg on its head, happy expression, the two monkeys on the side sway slightly, fixed camera.
```

</details>

**G10 · 3D rendering · shots #19, #20, #21 · 9.3 s**

- Reference material: `@image1` character reference (one frontal + one half-body per character) to keep faces consistent across shots; `@video1` the original 57.2–66.5 s segment as a reference for motion, camera and cut rhythm (reference only, do not copy the picture)
- First-frame image prompt: 3D-rendered fluffy cartoon monkey plushies, dense realistic fur all over, big smiling eyes, one yellow and one brown, a salted duck egg cut open and leaking oil and a small brown jar on the table, minimal clean background, soft studio lighting, vertical composition.
- Alternative route: a 3D project. Have an LLM write a Blender script or a Three.js scene, model the characters / props once and render each shot from its own camera; best for multiple angles and character consistency, and changing copy or shots is just a code change (R08)

```text
Vertical 9:16; frontal soft light (flat), soft light (almost no shadow edge), neutral (about 5000K), background clearly blurred (shallow depth of field / portrait mode). Character per @image1. Follow @video1 for motion, camera and cut rhythm. Hard cuts between shots.
Shot 1 (0.0–3.5 s): fixed camera, two fluffy cartoon monkeys stand at a table, the brown monkey blinks and raises a hand slightly and waves, a red prohibition sign slides in from the right, the characters' fur is fine and flicker-free, the whole frame steady and lifelike.
Shot 2 (3.5–5.8 s): fixed eye-level camera, a 3D-rendered wooden balance scale. A small yolk ball drops straight down from above and merges into the large yolk ball on the left pan, the beam tilts down to the left, then a few sparkling four-point stars pop up above it, and the scale comes to rest tilted.
Shot 3 (5.8–9.3 s): camera fixed, three fluffy cartoon animal characters stand side by side, blinking and swaying slightly, the geometric spheres in the background stay floating and still, lighting steady.
```

<details><summary>English prompt (as generated by the tool)</summary>

```text
Shot 1 (0.0-3.5s): Static camera shot, two fluffy 3D cartoon monkeys standing in front of a table with salted duck eggs, the brown monkey blinks and moves slightly, a red prohibition sign enters from the right, realistic hair particle rendering, soft studio lighting.
Shot 2 (3.5-5.8s): Static eye-level shot, 3D rendered wooden scale. A small egg yolk ball falls from above and merges into a larger yolk ball on the left tray, causing the scale to tilt down to the left, followed by sparkling four-point star graphics popping up above, ending in a tilted rest.
Shot 3 (5.8-9.3s): Static camera, three fluffy cartoon animal characters standing side by side, characters slightly blinking and swaying their bodies, geometric spheres in the background remain suspended and still, stable lighting.
```

</details>

## Framing, lighting and editing

- Framing: medium shot 49%, close-up 25%, extreme close-up 23%; eye-level angle; subject is the sharpest region of the frame (no face, so shot size is an estimate)
- Camera: fixed (tripod / stand / on a table) 78%, moving (gimbal / handheld pan) 22%; common move: push-in
- Lighting: frontal soft light (flat). Key light frontal (from the camera direction, both sides of the subject equally bright), at subject height; contrast ratio about 1.0:1 (0.0 stops); medium softness; neutral color temperature (about 5000K); background clearly blurred (shallow depth of field / portrait mode)
- (The picture is not live footage: the above describes the light in the frame, for writing generation prompts, not a shooting plan)
- Grade: Japanese pastel (bright, low saturation, lifted blacks); low saturation, low contrast, lifted blacks (grayish shadows), neutral overall
- Editing: 18.0 cuts per minute, 3.17 s average shot; no live-action shots

| Dimension | Item | Conclusion | Source |
|---|---|---|---|
| Framing | Shot size | Medium shot | Measured: medium shot; VLM: medium shot |
| Framing | Angle | Eye level | VLM: eye level |
| Framing | Composition | Subject centered, vertical | VLM: subject centered, vertical |
| Camera | Support | Tripod / stand, fixed | Measured: fixed (tripod / stand / on a table); VLM: tripod / stand, fixed |
| Camera | Lens | Virtual standard lens | VLM: virtual standard lens |
| Camera | Setups / scene | Pure 3D virtual camera: character wide / medium interaction shots + microscopic duck-egg cross-section views + towel-wringing / balance-scale metaphor views | VLM: same |
| Camera | Likely gear | No camera equipment, pure 3D digital rendering (e.g. Blender / C4D) | VLM: same |
| Lighting | Scene | Indoor | VLM: indoor |
| Lighting | Key | Soft skylight | VLM: soft skylight |
| Lighting | Softness | Medium | Measured: medium; VLM: soft |
| Lighting | Fill | Even fill | VLM: even fill |
| Lighting | Rim | None | VLM: none |
| Lighting | Catchlight | Highlights in the characters' eyes | VLM: highlights in the characters' eyes |
| Lighting | Visible sources | None | VLM: none |
| Lighting | Color temperature | Neutral (about 5000K) | Measured: neutral (about 5000K); VLM: neutral |
| Lighting | Setup | Even soft lighting | VLM: even soft lighting |
| Lighting | Overall | Bright, soft 3D cartoon lighting, slightly warm, key frontal and slightly to one side, soft natural shadows, diffuse reflection with a plush-material feel. | VLM: same |
| Grade | Style | Japanese pastel (bright, low saturation, lifted blacks) | Measured: Japanese pastel (bright, low saturation, lifted blacks) |

- Note: contrast ratio, softness, color temperature and shot size are measured on the subject's pixels (the face when there is one, otherwise the sharpest region, which is an estimate); platform transcoding crushes shadows, so softness and fill read conservative; angle, light source type, catchlight and rim come from visual review. Where measurement and visual review disagree, both are listed.

## Metadata clues

- Encoder: `Lavf61.7.100`

## Audio breakdown

![audio](audio.png)

- Overall loudness −17.3 dBFS, peak −1.6 dBFS
- **Voice**: present, 76% of duration, fundamental about 319.6 Hz (reads female)
- **Background music**: 97% of duration (1.5–66.0 s), about 110.0 BPM (or 55.0 / 219.9), key D major (confidence 0.866), section changes at 4.0 s, 8.5 s, 13.0 s, 17.0 s, 26.5 s, 42.5 s, 46.5 s, 51.0 s
- **Cuts vs beat**: 45% of cuts land on a beat (random baseline about 25%)

**Audio source: AI-generated or existing material**

- **Conclusion**: video type: narration / voice-over + music → voice: TTS; music: library music; sound effects: added in post from a sound library / editor (leaning, to be confirmed)
- **Separated stems**: `stems/original.wav` (original), `stems/vocals.wav` (dry voice; transient effects land here too), `stems/music.wav` (music + ambience)
- Video form: explainer; AI picture 0%, live footage 0%
- **Background music**: library music (existing_library 67%, ai_generated 21%, popular_song 12%)
  - The music is a bed under the picture: for an ordinary video the bed is most likely from a library (prior)
  - The voice is speech / narration rather than singing (sustained-pitch ratio 0.15): this is the ordinary "narration + BGM" form
  - VLM listening judgment: library music
  - Music and video end together: cut to length or custom-made (either source is possible, weak evidence)
  - How to confirm: on the Douyin / Xiaohongshu page check the music tag: a song title → existing track; "original sound by @author" → uploaded by the author (for music videos this is usually an AI song)
  - How to confirm: run Soda Music / QQ Music song recognition or ACRCloud: a hit means an existing song
  - How to confirm: listen: AI songs often have slurred diction, thin metallic highs and lyrics that track the picture closely; library beds are mostly instrumental, standard structure, cut off or faded at the end
- **Voice**: TTS (tts 81%, human_recording 14%, ai_video_native 5%)
  - VLM judged it a synthesized TTS voice
  - Voice and background sound are continuous across cuts: the whole narration was laid on separately in post (TTS or a human voice-over)
- **Sound effects**: added in post from a sound library / editor (library_editor 57%, ai_native 29%, live_sync 14%)
  - Cartoon effects in 3D / animated videos are also commonly synthesized procedurally or added from a library (for example the capsule-cinema plush recipe synthesizes them procedurally)
- Per-segment native-audio check: background change 10.4 dB at cuts, 9.6 dB within shots (ratio 1.08) → sound is continuous across cuts, laid as one bed in post

**Sound effects** (5; source: vlm = heard by the model, deterministic = signal-detected candidate, low confidence)

| Time | Type | Description | Matching picture | Source |
|---|---|---|---|---|
| 1.50 s | pop | the salted duck egg cracking and oil bursting out | action | vlm |
| 18.00 s | pop | tiny pops of salt grains hopping and burrowing in | action | vlm |
| 31.00 s | whoosh | the yolk dehydrating and shrinking | action | vlm |
| 43.00 s | other | wringing water out of a towel | action | vlm |
| 55.00 s | ding | success chime for the oily salted duck egg | text_pop | vlm |

**Audio remake plan**

- voice (A01-narration): TTS narration, a sweet young-female / children's-educational voice, crisp clear diction, moderate pace with a storytelling rhythm, inspiring and approachable → capsule-cinema DoubaoTTSTool / UniversalTTSTool (or a MiniMax cloned voice); use duration-controlled TTS (A02) when you must fit the original timing
- music (A03-music): light, lively, childlike cartoon score; relaxed, cheerful, full of curiosity and exploration; 110.0 BPM; D major → the original leans library: search a library by BPM / mood / key, or generate a royalty-free track in a similar style with AI
- sfx (A04-sfx): 5 effects → place them at the time points: capsule-cinema SoundEffectsTool / a sound library, or synthesize them procedurally like the plush recipe (pitch aligned to the BGM key)
- ambience (A04-sfx): no location ambience → ambience stock or a video-to-audio model
- Mix: voice prominent and clear, BGM as a bed with light sidechain ducking under the voice, effects at moderate level matched to the action points.

## Structure and copy (VLM)

- Summary: cute 3D plush cartoon characters and explanatory animation give a fun science explainer on why salted duck eggs ooze oil, the science behind the curing.
- Form: explainer; hook: opens with a puzzle, "why does a salted duck egg ooze oil when not a drop of oil was added?", paired with a close-up of a plush duckling cracking an egg that bursts with red oil, which grabs curiosity right away.
- Narration: TTS, a sweet young-female / children's-educational voice, crisp clear diction; music: light, lively, childlike cartoon score; captions: a persistent large yellow outlined title at the top, centered black explainer captions with white outline in a cute font at the bottom, small step-indicator labels in the middle
- Overall judgment (VLM): the whole piece is 3D character animation and 3D diagram rendering (plush animal characters and internal cross-section demos made in C4D / Blender or similar), composited in post with TTS voice-over and library BGM: a pure CG animated science short.
  - 0.0–8.88 s · the question and the visual · In a small courtyard in the morning, the duckling cracks open a cured salted duck egg and bright red oil pours out. Not a drop of oil was ever added, and the cat stares wide-eyed: where on earth does this oil come from?
  - 8.88–15.83 s · the hidden hand revealed · It turns out the director behind this oily show is plain ordinary salt.
  - 15.83–21.04 s · step 1: salt enters the pores · Step one: in the brine, salt creeps in grain by grain through the pores in the shell.
  - 21.04–33.75 s · step 2: salt enters the yolk and dehydrates it · Step two: salt fills the egg white first, then crosses the yolk membrane and sinks deep into the yolk. The water stored in the yolk gets pulled out by the salt. The dehydrated yolk shrinks into a ball, smaller and smaller, harder and harder.
  - 33.75–45.25 s · step 3: the proteins squeeze the oil out · Step three: the high salt makes the proteins in the yolk tighten up and clump together. The oil that was wrapped up tight suddenly all comes loose. Like wringing a towel: tighten your grip and the water is squeezed out.
  - 45.25–57.21 s · step 4: wait for the oil, and the result · Step four: leave the rest to time. Home curing takes 30 to 40 days. Crack one open after enough days and red oil pools on the yolk, trembling and glistening. That is an oily salted duck egg!
  - 57.21–66.0 s · wrap-up: where the oil really comes from · But this oil is not some mysterious substance someone sneaked in. It was always the fat in the yolk, not a drop more, not a drop less.

## Shot list

| # | Time | Duration | Underlying picture | Confidence | Material | Overlays | Camera | Remake route |
|---|---|---|---|---|---|---|---|---|
| 1 | 00:00.000 | 8.88 s | 3D rendering | 99% | Rendered by the producer | Captions, title | Static | R08-product-multiview → capsule_cinema |
| 2 | 00:08.875 | 4.79 s | 3D rendering | 99% | Rendered by the producer | Captions, title | Static | R08-product-multiview → capsule_cinema |
| 3 | 00:13.667 | 2.17 s | 3D rendering | 99% | Rendered by the producer | Text, title | Static | R08-product-multiview → capsule_cinema |
| 4 | 00:15.833 | 3.75 s | 3D rendering | 99% | Rendered by the producer | Captions, title | Static | R08-product-multiview → capsule_cinema |
| 5 | 00:19.583 | 1.46 s | 3D rendering | 99% | Rendered by the producer | Text, title | Smooth move | R08-product-multiview → capsule_cinema |
| 6 | 00:21.042 | 3.29 s | 3D rendering | 99% | Rendered by the producer | Captions, title | Static | R08-product-multiview → capsule_cinema |
| 7 | 00:24.333 | 1.79 s | 3D rendering | 99% | Rendered by the producer | Text, title | Smooth move | R08-product-multiview → capsule_cinema |
| 8 | 00:26.125 | 1.67 s | 3D rendering | 99% | Rendered by the producer | Text, title | Smooth move | R08-product-multiview → capsule_cinema |
| 9 | 00:27.792 | 1.67 s | 3D rendering | 99% | Rendered by the producer | Text, title | Static | R08-product-multiview → capsule_cinema |
| 10 | 00:29.458 | 1.17 s | 3D rendering | 99% | Rendered by the producer | Text, title | Smooth move | R08-product-multiview → capsule_cinema |
| 11 | 00:30.625 | 3.12 s | 3D rendering | 99% | Rendered by the producer | Captions, title | Smooth move | R08-product-multiview → capsule_cinema |
| 12 | 00:33.750 | 4.29 s | 3D rendering | 99% | Rendered by the producer | Captions, title | Static (subject moving) | R08-product-multiview → capsule_cinema |
| 13 | 00:38.042 | 3.54 s | 3D rendering | 90% | Rendered by the producer | Captions, title | Static | R08-product-multiview → capsule_cinema |
| 14 | 00:41.583 | 3.67 s | 3D rendering | 99% | Rendered by the producer | Captions, title | Static | R08-product-multiview → capsule_cinema |
| 15 | 00:45.250 | 2.08 s | 3D rendering | 99% | Rendered by the producer | Text, title | Static | R08-product-multiview → capsule_cinema |
| 16 | 00:47.333 | 2.25 s | 3D rendering | 99% | Rendered by the producer | Text, title | Smooth move | R08-product-multiview → capsule_cinema |
| 17 | 00:49.583 | 5.38 s | 3D rendering | 99% | Rendered by the producer | Captions, text, title | Static | R08-product-multiview → capsule_cinema |
| 18 | 00:54.958 | 2.25 s | 3D rendering | 99% | Rendered by the producer | Text, captions, title | Static | R08-product-multiview → capsule_cinema |
| 19 | 00:57.208 | 3.54 s | 3D rendering | 90% | Rendered by the producer | Captions, title | Static | R08-product-multiview → capsule_cinema |
| 20 | 01:00.750 | 2.25 s | 3D rendering | 99% | Rendered by the producer | Text, title | Static | R08-product-multiview → capsule_cinema |
| 21 | 01:03.000 | 3.54 s | 3D rendering | 99% | Rendered by the producer | Captions, title | Smooth move | R08-product-multiview → capsule_cinema |

## Shot details

### #1 00:00.000–00:08.875 · 3D rendering + code / editor overlays (captions, title)

![shot_001](shots/shot_001.jpg)

- Probabilities: 3D rendering 99%, 2D animation 0%, code / template rendering 0% (fused)
- Text on objects in the frame (not post captions): 咸
- Material: rendered by the producer
- Likely tool: Blender / C4D
- In-shot motion beats: 2.62 s, 3.67 s, 4.12 s, 7.00 s
- Rendered texture: flat color blocks 53%, 95% of pixels use only 50 colors, horizontal / vertical straight edges 37%
- [VLM] Characters and set are all 3D models, with distinct fur material and a cartoon render style
- [VLM] The oil bursting from the salted egg is a 3D fluid or particle effect
- [VLM] Geometry is stable, no AI-generated warping or flicker
- Picture: 3D animation: a yellow plush character cracks open a salted duck egg, red oil pours out, a brown character watches from the side. Fixed camera, several shot sizes intercut, soft light.
- Framing / light: medium shot; eye level; virtual fixed camera; outdoor overcast / shade soft light; key frontal (from the camera direction, both sides of the subject equally bright); contrast about 1.2:1 (0.2 stops); medium softness
- Text track (captions / bottom): 鸭宝磕开一枚腌好的咸鸭蛋 / 明明没往里加一滴油 / 花猫看得眼睛都直了
- Text track (text / middle): 咸
- Text track (title / top): 咸鸭蛋为什么会流油？
- Remake: 3D rendering: first/last-frame AI generation gives a close approximation; for exact multi-angle work go 3D modeling + Blender rendering
  - First-frame image prompt: 3D cartoon style, a yellow fluffy little animal in a minimal green courtyard, a cracked-open salted duck egg in front of it leaking red oil, soft light, centered composition.
  - Video prompt: 3D animation, a yellow fluffy little animal cracks open a salted duck egg, red oil gushes out of the yolk, a brown fluffy little animal next to it watches in surprise, static camera, soft light.

### #2 00:08.875–00:13.667 · 3D rendering + code / editor overlays (captions, title)

![shot_002](shots/shot_002.jpg)

- Probabilities: 3D rendering 98%, 2D animation 0%, code / template rendering 0% (fused)
- Material: rendered by the producer
- Likely tool: Blender / C4D
- In-shot motion beats: 9.62 s, 10.46 s, 11.46 s
- Rendered texture: flat color blocks 48%, 95% of pixels use only 65 colors, straight edges 34%
- 67% of sampled frames have a frontal face with a fixed camera, and the audio track has voice
- [VLM] Characters have the standard 3D fur-render look, clean edges, stable lighting
- [VLM] At 00:10 the characters are instantly swapped for fur balls that roll away: typical 3D keyframe / effects work, no AI-generated random warping
- Picture: three plush cartoon animals around a pot in a bowl in the middle, then the animals turn into white fur balls and roll out of frame to the right. Fixed camera, soft lighting.
- Framing / light: medium shot; fixed (tripod / stand / on a table); frontal soft light (flat); key frontal (from the camera direction, both sides of the subject equally bright); contrast about 1.2:1 (0.3 stops); hard light (clear shadow edge)
- Text track (captions / bottom): 这油 / 到底是打哪儿冒出来的？ / 原来 / 这场流油的大戏
- Text track (title / top): 咸鸭蛋为什么会流油？
- Remake: 3D rendering: first/last-frame AI generation gives a close approximation; for exact multi-angle work go 3D modeling + Blender rendering
  - First-frame image prompt: 3D cartoon style, three cute animal characters with fluffy fur stand around a white bowl holding a small clay pot. Simple geometric hill shapes in the background, solid-color backdrop. Soft top light, bright colors.
  - Video prompt: three fluffy cartoon animal characters stand in the frame, then suddenly turn into three white fur balls and roll out to the right together, leaving the bowl and pot in the middle. Static camera.

### #3 00:13.667–00:15.833 · 3D rendering + code / editor overlays (text labels, title)

![shot_003](shots/shot_003.jpg)

- Probabilities: 3D rendering 99%, code / template rendering 0%, 2D animation 0% (fused)
- Material: rendered by the producer
- Likely tool: Blender / C4D
- Rendered texture: flat color blocks 75%, 95% of pixels use only 18 colors, straight edges 41%
- [VLM] Geometry edges are clean and sharp, no AI-generated warping or flicker
- [VLM] Materials (wood grain, matte white balls) and lighting (soft shadow under the bowl) show typical 3D-render traits
- [VLM] The balls pop out on precise, rigid paths, consistent with 3D keyframe animation
- Picture: minimal 3D scene, a wooden bowl holds white powder and a few white balls with eyes (anthropomorphic salt grains) pop out of the powder one by one. Fixed camera, soft lighting.
- Framing / light: extreme close-up; slightly high angle; tripod / stand, fixed; high-key (bright overall, almost no shadows); key frontal (from the camera direction, both sides of the subject equally bright); contrast about 1.0:1 (0.0 stops); soft light (almost no shadow edge)
- Text track (text / bottom): 幕后导演是普普通通的盐
- Text track (title / top): 咸鸭蛋为什么会流油？
- Remake: 3D rendering: first/last-frame AI generation gives a close approximation; for exact multi-angle work go 3D modeling + Blender rendering
  - First-frame image prompt: 3D cartoon style, minimal two-tone background, a small light-wood bowl in the center of the frame full of white powder, several white balls with little black eyes peeking out of the powder. Soft top light, matte finish, vertical centered composition.
  - Video prompt: static camera. In the center is a wooden bowl full of white powder. Several white balls with tiny eyes pop out of the powder one after another and rest on the surface, light and bouncy with an elastic settle.

### #4 00:15.833–00:19.583 · 3D rendering + code / editor overlays (captions, title)

![shot_004](shots/shot_004.jpg)

- Probabilities: 3D rendering 93%, code / template rendering 6%, 2D animation 0% (fused)
- Material: rendered by the producer
- Likely tool: Blender
- In-shot motion beats: 16.88 s
- Rendered texture: flat color blocks 52%, 95% of pixels use only 35 colors, straight edges 39%
- [VLM] Entirely 3D assets: fur particle systems (the plush doll, the fur ball in the pot), a smooth plastic / ceramic flowerpot, a wooden-handled screw prop
- [VLM] Clean, smooth global diffuse lighting with soft ambient occlusion (AO)
- [VLM] Objects in the shot move with keyframe dynamics, geometry edges stable with no flicker or warping
- Picture: a 3D cartoon educational scene, a brown pot holds a white fur ball, next to it a fluffy little doll character, a simple slope and a hanging screw in the background, camera fixed at eye level.
- Framing / light: medium shot; eye level; tripod / stand, fixed; frontal soft light (flat); key frontal (from the camera direction, both sides of the subject equally bright); contrast about 1.0:1 (0.1 stops); soft light (almost no shadow edge)
- Text track (captions / bottom): 第一步腌蛋的盐水里 / 盐顺着蛋壳的气孔
- Text track (title / top): 咸鸭蛋为什么会流油？ / 第1步·盐钻气孔
- Remake: 3D rendering: first/last-frame AI generation gives a close approximation; for exact multi-angle work go 3D modeling + Blender rendering
  - First-frame image prompt: 3D cartoon style, macro photography, a brown cylindrical flowerpot in the center with a white fluffy ball showing inside, a fluffy cartoon monkey peeking in from the right, soft solid-color background, Blender 3D render look, soft light
  - Video prompt: static eye-level shot, 3D cartoon scene, the fluffy object in the pot stays still, the little monkey plush on the right blinks its big eyes and watches it, tiny screws hover and drift slightly in the background, a smooth, clean 3D-rendered animation throughout

### #5 00:19.583–00:21.042 · 3D rendering + code / editor overlays (text labels, title)

![shot_005](shots/shot_005.jpg)

- Probabilities: 3D rendering 98%, 2D animation 0%, AI-generated video 0% (fused)
- Material: rendered by the producer
- Likely tool: Blender / C4D
- Rendered texture: flat color blocks 31%, 95% of pixels use only 51 colors, straight edges 30%
- Per-frame sharpness oscillates periodically (period 2 frames, correlation 0.64): the picture went through frame interpolation / rate conversion; the original frame rate was lower
- Duplicate frames follow an interpolated pattern (0%, period None): source was 24/25 fps then transcoded, typical of AI video, 3D / animation renders or film footage
- [VLM] Geometry edges are perfectly smooth, no real-world physical flaws
- [VLM] Materials (translucent shell, matte yolk, woven base) are uniform, consistent with 3D rendering
- [VLM] The salt grains move on precise paths, consistent with 3D keyframe animation
- [VLM] No AI-generated warping or flicker, no live-footage noise
- Picture: a 3D-rendered duck-egg cross-section model on a base, a few salt grains fly toward a pore in the shell. Fixed camera, soft lighting.
- Framing / light: extreme close-up; eye level; moving (gimbal / handheld pan); frontal soft light (flat); key frontal (from the camera direction, both sides of the subject equally bright); contrast about 1.1:1 (0.1 stops); soft light (almost no shadow edge)
- Text track (text / bottom): 一粒粒往里钻
- Text track (title / top): 咸鸭蛋为什么会流油？ / 第1步·盐钻气孔
- Text track (title / top): 气孔
- Remake: 3D rendering: first/last-frame AI generation gives a close approximation; for exact multi-angle work go 3D modeling + Blender rendering
  - First-frame image prompt: 3D render, vertical, a translucent cutaway duck-egg model on a woven base, a yellow yolk inside, fine hairs on the shell, solid-color background, soft studio light, cartoon educational style.
  - Video prompt: 3D animation, a few tiny salt grains fly in from the upper right toward a pore in the shell of the duck-egg model, static camera.

### #6 00:21.042–00:24.333 · 3D rendering + code / editor overlays (captions, title)

![shot_006](shots/shot_006.jpg)

- Probabilities: 3D rendering 93%, code / template rendering 6%, 2D animation 0% (fused)
- Material: rendered by the producer
- Likely tool: Blender
- Rendered texture: flat color blocks 67%, 95% of pixels use only 20 colors, straight edges 43%
- [VLM] The cylindrical glass containers and the round yolk-membrane filter are regular 3D geometry
- [VLM] The white balls (salt molecules) move on smooth, precise paths, none of the warping or flicker common in AI generation
- [VLM] Clean uniform ambient diffuse lighting and soft shadows, typical of Blender / C4D educational renders
- Picture: a 3D educational animation: the camera looks at a transparent glass cylinder model on a white desk at eye level, slightly from above, then pulls back to show the egg white, the yolk-membrane filter and the yolk model; white salt-particle balls pass through the round yolk-membrane filter in the middle; explanatory title and captions at the top and bottom.
- Framing / light: medium shot; slightly high angle; fixed (tripod / stand / on a table); backlit / background brighter than subject; key frontal (from the camera direction, both sides of the subject equally bright); contrast about 1.0:1 (0.0 stops); soft light (almost no shadow edge)
- Text track (captions / bottom): 第二步盐先灌满蛋清 / 再穿过蛋黄膜
- Text track (title / top): 咸鸭蛋为什么会流油？ / 第2步·盐进蛋黄
- Text track (title / top): 蛋黄膜
- Remake: 3D rendering: first/last-frame AI generation gives a close approximation; for exact multi-angle work go 3D modeling + Blender rendering
  - First-frame image prompt: 3D scientific diagram, minimal studio background, light-green solid wall and white desk, a transparent glass cylinder in the middle holding white liquid, on a shallow round light-wood tray, soft diffuse lighting, high-quality 3D render.
  - Video prompt: 3D educational animation, camera at eye level, fixed, pulling back slightly; on the desk a transparent cylinder of white liquid on the left, a cylinder of translucent yellow liquid on the right, a round semi-permeable membrane model with a wooden rim floating in between; several smooth white balls pass through the membrane ring from left to right at a smooth, even pace.

### #7 00:24.333–00:26.125 · 3D rendering + code / editor overlays (text labels, title)

![shot_007](shots/shot_007.jpg)

- Probabilities: 3D rendering 98%, 2D animation 0%, AI-generated video 0% (fused)
- Material: rendered by the producer
- Likely tool: Blender / C4D
- [VLM] The yolk is a perfect geometric sphere
- [VLM] The egg white is a typical 3D translucent subsurface-scattering material
- [VLM] The salt grains at the edge are regular sphere models
- [VLM] Lighting is clean and noise-free, consistent with 3D rendering
- Picture: a 3D-rendered close-up of a salted duck egg cross-section, the camera pushes in slowly toward the orange yolk in the center.
- Framing / light: extreme close-up; eye level; moving (gimbal / handheld pan); frontal soft light (flat); key frontal (from the camera direction, both sides of the subject equally bright); contrast about 1.0:1 (0.0 stops); soft light (almost no shadow edge)
- Text track (text / bottom): 一路潜进蛋黄深处
- Text track (text / middle): 蛋黄
- Text track (title / top): 咸鸭蛋为什么会流油？ / 第2步·盐进蛋黄
- Remake: 3D rendering: first/last-frame AI generation gives a close approximation; for exact multi-angle work go 3D modeling + Blender rendering
  - First-frame image prompt: 3D cartoon style, close-up of a salted duck egg cut open, translucent egg white wrapping a perfectly round orange yolk, a few round salt grains at the edge, clean solid-color background, soft studio lighting, C4D render, minimalist, high quality.
  - Video prompt: the camera pushes in slowly and steadily toward the orange yolk in the center, keeping the 3D cartoon look, lighting unchanged.

### #8 00:26.125–00:27.792 · 3D rendering + code / editor overlays (text labels, title)

![shot_008](shots/shot_008.jpg)

- Probabilities: 3D rendering 98%, 2D animation 0%, AI-generated video 0% (fused)
- Material: rendered by the producer
- Likely tool: Blender / C4D
- [VLM] The yolk is a perfect geometric sphere
- [VLM] The egg white is a uniform translucent 3D material
- [VLM] The blue droplet particles appear with smooth, precise animation, no AI warping or flicker
- [VLM] Lighting and materials are very clean overall, typical of 3D rendering
- Picture: a 3D-rendered close-up of a salted duck egg cross-section, the orange spherical yolk centered, blue droplet-shaped particles gradually appearing around it, camera fixed.
- Framing / light: extreme close-up; eye level; moving (gimbal / handheld pan); frontal soft light (flat); key frontal (from the camera direction, both sides of the subject equally bright); contrast about 1.0:1 (0.0 stops); soft light (almost no shadow edge)
- Text track (text / bottom): 蛋黄里存着的水分
- Text track (title / top): 咸鸭蛋为什么会流油？ / 第2步·盐进蛋黄
- Remake: 3D rendering: first/last-frame AI generation gives a close approximation; for exact multi-angle work go 3D modeling + Blender rendering
  - First-frame image prompt: 3D cartoon style, close-up, a salted duck egg cut open, a perfect orange spherical yolk in the middle wrapped in translucent egg white, on a light plush base, solid-color background, soft lighting, minimal clean materials.
  - Video prompt: 3D cartoon style, close-up, a salted duck egg cut open, a perfect orange spherical yolk in the middle wrapped in translucent egg white. Over time, blue droplet-shaped particle effects gradually appear in the egg white around the yolk. Camera stays fixed.

### #9 00:27.792–00:29.458 · 3D rendering + code / editor overlays (text labels, title)

![shot_009](shots/shot_009.jpg)

- Probabilities: 3D rendering 94%, code / template rendering 4%, 2D animation 0% (fused)
- Material: rendered by the producer
- Likely tool: Blender
- [VLM] The shell cross-section, egg white and yolk are all geometrically precise 3D objects
- [VLM] The arrow motion and the dehydration-shrink animation match keyframe interpolation
- [VLM] Smooth diffuse shading and subsurface-scattering render look
- Picture: a 3D cross-section of the salted-egg curing process. The egg white is enclosed by a transparent shell with the yolk at the center; translucent blue arrows show water being drawn out; camera fixed and centered.
- Framing / light: extreme close-up; eye level; tripod / stand, fixed; frontal soft light (flat); key frontal (from the camera direction, both sides of the subject equally bright); contrast about 1.0:1 (0.0 stops); soft light (almost no shadow edge)
- Text track (text / bottom): 被盐一把拉了出去
- Text track (title / top): 咸鸭蛋为什么会流油？ / 第2步·盐进蛋黄
- Remake: 3D rendering: first/last-frame AI generation gives a close approximation; for exact multi-angle work go 3D modeling + Blender rendering
  - First-frame image prompt: 3D cutaway of a salted duck egg, a white shell rig on the outside, a plump round yellow yolk inside translucent egg white, soft even studio lighting, minimal tech-educational style.
  - Video prompt: static close-up, a 3D cross-section showing water being drawn out of the yolk, several translucent light-blue arrows flow outward from the yolk surface, the yolk firms up and contracts slightly, clean CG animation look throughout.

### #10 00:29.458–00:30.625 · 3D rendering + code / editor overlays (text labels, title)

![shot_010](shots/shot_010.jpg)

- Probabilities: 3D rendering 99%, 2D animation 0%, code / template rendering 0% (fused)
- Material: rendered by the producer
- Likely tool: Blender / C4D
- Rendered texture: flat color blocks 48%, 95% of pixels use only 37 colors, straight edges 36%
- [VLM] A typical 3D cartoon render style
- [VLM] The character and the white ball have high-quality simulated 3D fur, clean edges, no AI-generated flicker or warping
- [VLM] Geometry edges are smooth and lighting is stable and consistent, as in conventional 3D software renders
- Picture: a 3D animated scene, a fluffy cartoon monkey stands on the left looking at a floating fluffy white ball on the right, a simple geometric wall and a brown jar in the background.
- Framing / light: close-up; eye level; moving (gimbal / handheld pan); frontal soft light (flat); key frontal (from the camera direction, both sides of the subject equally bright); contrast about 1.2:1 (0.3 stops); soft light (almost no shadow edge)
- Text track (text / bottom): 失了水的蛋黄
- Text track (title / top): 咸鸭蛋为什么会流油？ / 第2步·盐进蛋黄
- Remake: 3D rendering: first/last-frame AI generation gives a close approximation; for exact multi-angle work go 3D modeling + Blender rendering
  - First-frame image prompt: 3D render, a cute fluffy cartoon monkey stands on the left of the frame looking at a floating fluffy white ball on the right, a simple light-green geometric wall and a brown jar in the background, soft studio lighting bringing out the fur texture, Pixar style, vertical composition.
  - Video prompt: 3D animation, a cute fluffy cartoon monkey stands on the left looking at a floating fluffy white ball on the right, the monkey breathes gently, the white ball bobs slightly, fixed camera, soft light.

### #11 00:30.625–00:33.750 · 3D rendering + code / editor overlays (captions, title)

![shot_011](shots/shot_011.jpg)

- Probabilities: 3D rendering 98%, AI-generated video 0%, 2D animation 0% (fused)
- Material: rendered by the producer
- Likely tool: Blender / C4D
- [VLM] The egg structure, translucent shell, smooth yolk and woven base all show perfect geometry and the typical 3D-render look
- [VLM] The shrinking of the central yellow sphere is a precise parametric scale, with no physical deformation or AI-generated flicker
- [VLM] Lighting is clean and stable overall, none of the noise or flaws of a real camera
- Picture: a 3D-rendered cutaway model of the inside of a duck egg. A yellow sphere (the yolk) floats inside a translucent white shell on a woven base. The animation shows the central yellow sphere gradually shrinking.
- Framing / light: extreme close-up; eye level; moving (gimbal / handheld pan); frontal soft light (flat); key frontal (from the camera direction, both sides of the subject equally bright); contrast about 1.0:1 (0.0 stops); soft light (almost no shadow edge)
- Text track (captions / bottom): 缩成一团 / 越缩越小越缩越硬
- Text track (title / top): 咸鸭蛋为什么会流油？ / 第2步·盐进蛋黄
- Remake: 3D rendering: first/last-frame AI generation gives a close approximation; for exact multi-angle work go 3D modeling + Blender rendering
  - First-frame image prompt: 3D render, close-up, a cutaway model of the inside of a duck egg. A smooth yellow sphere in the center stands for the yolk, wrapped in a translucent white shell, resting on a ring-shaped base of beige woven material. Light-green solid background, soft studio light, cartoon educational style, clean and bright.
  - Video prompt: 3D animation, fixed camera. A cutaway duck-egg model in the center of the frame; the yellow sphere (yolk) in the middle shrinks smoothly and proportionally over time while everything else stays still.

### #12 00:33.750–00:38.042 · 3D rendering + code / editor overlays (captions, title)

![shot_012](shots/shot_012.jpg)

- Probabilities: 3D rendering 99%, code / template rendering 0%, 2D animation 0% (fused)
- Material: rendered by the producer
- Likely tool: Blender / C4D
- Rendered texture: flat color blocks 55%, 95% of pixels use only 28 colors, straight edges 43%
- [VLM] The fluffy characters' fur is rendered finely and stably, none of the flicker or warping common in AI video
- [VLM] The physical collisions and deformation as the characters squeeze together are very regular and as expected
- [VLM] The orange balls pop out on precise paths, consistent with conventional keyframed 3D animation
- [VLM] Lighting is consistent and materials are clean overall, typical of 3D software renders
- Picture: a 3D cartoon frame, several white fluffy big-eyed creatures on a wooden board move toward the middle and huddle together, then a few orange balls pop out of them, camera fixed.
- Framing / light: close-up; eye level; tripod / stand, fixed; frontal soft light (flat); key frontal (from the camera direction, both sides of the subject equally bright); contrast about 1.0:1 (0.0 stops); soft light (almost no shadow edge)
- Text track (captions / bottom): 第三步 / 高盐让蛋黄里的蛋白质 / 纷纷收紧抱团
- Text track (title / top): 咸鸭蛋为什么会流油？ / 第3步·蛋白挤油 / 第了步·蛋白挤油
- Remake: 3D rendering: first/last-frame AI generation gives a close approximation; for exact multi-angle work go 3D modeling + Blender rendering
  - First-frame image prompt: 3D cartoon style, several white fluffy ball creatures with big eyes lying on a wooden tabletop, solid-color background, soft light, a macro-photography feel.
  - Video prompt: several white fluffy ball creatures start out scattered on a wooden tabletop, then move toward the middle and huddle tightly together, and a few small orange balls pop out of them; camera stays fixed.

### #13 00:38.042–00:41.583 · 3D rendering + code / editor overlays (captions, title)

![shot_013](shots/shot_013.jpg)

- Probabilities: 3D rendering 89%, AI-generated video 9%, 2D animation 0% (fused)
- Material: rendered by the producer
- Likely tool: Blender
- [VLM] Perfectly regular 3D geometry and a translucent material cross-section
- [VLM] Realistic, regular fluid / oil-separation physics simulation on the yolk surface
- [VLM] Clean, noise-free 3D lighting and smooth transitions
- Picture: a 3D cross-section of the inside of a salted duck egg, transparent egg white wrapping an orange-yellow yolk, golden oil gradually separating out at the top of the yolk, fixed camera, soft studio light.
- Framing / light: extreme close-up; eye level; tripod / stand, fixed; frontal soft light (flat); key frontal (from the camera direction, both sides of the subject equally bright); contrast about 1.0:1 (0.0 stops); soft light (almost no shadow edge)
- Text track (captions / bottom): 原本被裹得严严实实的油 / 一下子全散开了
- Text track (title / top): 咸鸭蛋为什么会流油？ / 第了步·蛋白挤油
- Remake: 3D rendering: first/last-frame AI generation gives a close approximation; for exact multi-angle work go 3D modeling + Blender rendering
  - First-frame image prompt: 3D cutaway render, a cooked salted duck egg model cut open, a round orange-yellow yolk in the center, translucent gel-like egg white and a slice of shell outside, minimal soft studio lighting, centered composition, refined educational illustration style
  - Video prompt: static close-up of a 3D cutaway salted duck egg; glistening golden oil slowly seeps out of the top of the orange spherical yolk and gathers and flows, egg white and shell stay still, soft studio top light, a steady view of the oil separating out

### #14 00:41.583–00:45.250 · 3D rendering + code / editor overlays (captions, title)

![shot_014](shots/shot_014.jpg)

- Probabilities: 3D rendering 96%, code / template rendering 3%, 2D animation 0% (fused)
- Material: rendered by the producer
- Likely tool: Blender
- Rendered texture: flat color blocks 67%, 95% of pixels use only 20 colors, straight edges 45%
- [VLM] A 3D fur-material block and metal rods, geometry stable and precise
- [VLM] The mechanical squeeze deformation and falling drops are standard 3D dynamics / keyframe simulation
- [VLM] Minimal solid-color 3D set with smooth diffuse lighting
- Picture: a 3D educational animation. Against a minimal light-gray background, a fuzzy light-blue square towel is squeezed tight by the metal rod running through it, transparent drops fall on both sides, simulating wringing water out of a towel.
- Framing / light: medium shot; eye level; fixed (tripod / stand / on a table); backlit / background brighter than subject; key frontal (from the camera direction, both sides of the subject equally bright); contrast about 1.0:1 (0.0 stops); soft light (almost no shadow edge)
- Text track (captions / bottom): 就像拧毛巾 / 手一收紧 / 水就被挤了出来
- Text track (title / top): 咸鸭蛋为什么会流油？ / 第了步·蛋白挤油
- Remake: 3D rendering: first/last-frame AI generation gives a close approximation; for exact multi-angle work go 3D modeling + Blender rendering
  - First-frame image prompt: 3D minimal educational render style, smooth gray background and light tabletop, a light gray-blue square towel with fine fuzz hangs in the center of the frame, threaded on a thin horizontal metal rod, soft even lighting, clean 3D model
  - Video prompt: fixed eye-level camera, 3D animation demo: a square towel with fine fuzz in the center is pushed inward by metal rods from left and right and compressed into a thin T shape; as it is squeezed, several crystal-clear water drops form in the air on both sides of the towel and fall; the background stays minimal and still

### #15 00:45.250–00:47.333 · 3D rendering + code / editor overlays (text labels, title)

![shot_015](shots/shot_015.jpg)

- Probabilities: 3D rendering 98%, 2D animation 0%, code / template rendering 0% (fused)
- Material: rendered by the producer
- Likely tool: Blender / C4D
- Rendered texture: flat color blocks 44%, 95% of pixels use only 65 colors, straight edges 33%
- [VLM] The subject is three fluffy cartoon figures with clear 3D modeling and fur rendering
- [VLM] The background is a set of simple geometric shapes, lighting stable, no AI warping
- [VLM] Fine particles fall from the top at a constant speed, consistent with a 3D particle system
- Picture: three fluffy 3D cartoon animals stand side by side in a simple geometric set, a jar in the background, fine particles falling from the top of the frame. Fixed camera.
- Framing / light: full shot; eye level; fixed (tripod / stand / on a table); frontal soft light (flat); key frontal (from the camera direction, both sides of the subject equally bright); contrast about 1.0:1 (0.0 stops); soft light (almost no shadow edge)
- Text track (text / bottom): 第四步剩下的交给时间
- Text track (title / top): 咸鸭蛋为什么会流油？ / 第4步·静待出油
- Remake: 3D rendering: first/last-frame AI generation gives a close approximation; for exact multi-angle work go 3D modeling + Blender rendering
  - First-frame image prompt: 3D cartoon style, three cute fluffy little animals stand side by side, green in the middle, brown and gray on the sides, simple green geometric hills and a brown jar in the background, soft light, vertical composition
  - Video prompt: 3D cartoon style, three cute fluffy little animals stand side by side, green in the middle, brown and gray on the sides, simple green geometric hills and a brown jar in the background. Fine yellow particles keep falling from the top of the frame. Static camera

### #16 00:47.333–00:49.583 · 3D rendering + code / editor overlays (text labels, title)

![shot_016](shots/shot_016.jpg)

- Probabilities: 3D rendering 98%, AI-generated video 0%, 2D animation 0% (fused)
- Material: rendered by the producer
- Likely tool: Blender / C4D
- [VLM] The egg, yolk, oil and container are all 3D models with perfect geometry and uniform materials
- [VLM] The oil overflow is a typical 3D fluid or deformation animation with precise motion
- [VLM] Lighting is consistent, without the complex reflections and refractions of the real world
- [VLM] Twinkling star effects at the top of the frame are post compositing or a 3D particle effect
- Picture: a close-up of a salted duck egg model cut open, oil gradually spilling over the top of the yolk with twinkling star effects, solid-color background.
- Framing / light: medium shot; moving (gimbal / handheld pan); frontal soft light (flat); key frontal (from the camera direction, both sides of the subject equally bright); contrast about 1.0:1 (0.0 stops); soft light (almost no shadow edge)
- Text track (text / bottom): 家庭腌制要30~40天
- Text track (title / top): 咸鸭蛋为什么会流油？ / 第4步·静待出油
- Remake: 3D rendering: first/last-frame AI generation gives a close approximation; for exact multi-angle work go 3D modeling + Blender rendering
  - First-frame image prompt: 3D cartoon style, close-up, a salted duck egg model cut open sits in a white container, the yolk orange-yellow with a little oil on top, light-green solid background, soft lighting.
  - Video prompt: 3D cartoon style, close-up, a salted duck egg model cut open sits in a white container, the yolk orange-yellow. Over time more oil spills out over the top of the yolk, with twinkling star effects on the oil surface; camera stays fixed.

### #17 00:49.583–00:54.958 · 3D rendering + code / editor overlays (captions, text labels, title)

![shot_017](shots/shot_017.jpg)

- Probabilities: 3D rendering 99%, 2D animation 0%, code / template rendering 0% (fused)
- Material: rendered by the producer
- Likely tool: Blender / C4D
- In-shot motion beats: 50.38 s, 51.29 s
- Rendered texture: flat color blocks 52%, 95% of pixels use only 51 colors, straight edges 39%
- [VLM] Geometry edges of characters and objects are extremely clean and regular
- [VLM] Fur materials (character and white ball) and smooth materials (the pot) show the typical 3D-render look
- [VLM] The red oil flows down along a precise curved path, no physical deformation or AI flicker
- [VLM] Lighting stays absolutely consistent throughout the animation
- Picture: a 3D animated scene: the green plush character on the left watches the change on the right. The white fur ball on the right turns into a small brown pot, the pot opens, and pink liquid (red oil) flows down in an S curve onto the yolk. Fixed camera.
- Framing / light: close-up; eye level; fixed (tripod / stand / on a table); backlit / background brighter than subject; key frontal (from the camera direction, both sides of the subject equally bright); contrast about 1.2:1 (0.3 stops); soft light (almost no shadow edge)
- Text track (captions / bottom): 腌足日子再敲开 / 红油汪在蛋黄上 / 颤颤的亮亮的
- Text track (text / middle): 红油
- Text track (title / top): 咸鸭蛋为什么会流油？ / 第4步·静待出油
- Remake: 3D rendering: first/last-frame AI generation gives a close approximation; for exact multi-angle work go 3D modeling + Blender rendering
  - First-frame image prompt: 3D cartoon style, vertical. A fluffy green monkey character on the left, a small brown pot on a white base on the right, the pot open at the top showing a yellow yolk. Simple light-green geometric hill shapes in the background. Soft, even light overall.
  - Video prompt: 3D animation. The green fluffy character on the left stays still. On the right, what was a white fur ball instantly turns into a small brown pot. The pot opens at the top and a stream of pink liquid winds down from above in an S curve, landing on the yellow yolk inside the pot. Fixed camera.

### #18 00:54.958–00:57.208 · 3D rendering + code / editor overlays (text labels, captions, title)

![shot_018](shots/shot_018.jpg)

- Probabilities: 3D rendering 98%, 2D animation 0%, code / template rendering 0% (fused)
- Material: rendered by the producer
- Likely tool: Blender / C4D
- Rendered texture: flat color blocks 41%, 95% of pixels use only 89 colors, straight edges 39%
- 83% of sampled frames have a frontal face with a fixed camera, and the audio track has voice
- [VLM] Character fur is rendered cleanly and stably, no flicker
- [VLM] Geometry edges are sharp, lighting physically consistent
- [VLM] Motion is precise, none of the warping or floating common in AI
- Picture: three 3D plush monkey characters, the middle one lifts the bowl and salted duck egg on its head upward, minimal geometric mountains in the background, fixed camera.
- Framing / light: close-up; eye level; fixed (tripod / stand / on a table); frontal soft light (flat); key frontal (from the camera direction, both sides of the subject equally bright); contrast about 1.1:1 (0.1 stops); soft light (almost no shadow edge)
- Text track (text / bottom): 这就是流油咸鸭蛋
- Text track (captions / middle): 流油咸鸭蛋！ / 流油咸鸭蛋！+
- Text track (title / top): 咸鸭蛋为什么会流油？ / 成了！
- Remake: 3D rendering: first/last-frame AI generation gives a close approximation; for exact multi-angle work go 3D modeling + Blender rendering
  - First-frame image prompt: 3D cartoon style, three fluffy monkey characters, the middle one balancing a white bowl on its head with a salted duck egg in it, simple geometric mountains in the background, soft lighting, centered composition.
  - Video prompt: 3D animation, three fluffy monkeys; the middle one raises both hands to lift the bowl and salted duck egg on its head, looking happy, the two monkeys beside it sway slightly, camera fixed.

### #19 00:57.208–01:00.750 · 3D rendering + code / editor overlays (captions, title)

![shot_019](shots/shot_019.jpg)

- Probabilities: 3D rendering 89%, AI-generated video 9%, talking avatar 1% (fused)
- Material: rendered by the producer
- Likely tool: Blender
- In-shot motion beats: 58.33 s, 59.54 s
- 100% of sampled frames have a frontal face with a fixed camera, and the audio track has voice
- [VLM] The characters have a clear 3D fur particle system (fluffy geometric detail is precise, no AI flicker)
- [VLM] The jar, bowl and red prohibition sign behind them are 3D-modeled materials with rendered lighting, geometry stable
- [VLM] Model deformation during the animation strictly follows keyframes and the rigged skeleton
- Picture: a 3D cartoon scene, two fluffy cartoon monkeys stand at a table with a salted duck egg cut open and oozing oil and a pickling jar beside them, a red prohibition sign appears on the right, camera fixed at eye level.
- Framing / light: close-up; eye level; tripod / stand, fixed; frontal soft light (flat); key frontal (from the camera direction, both sides of the subject equally bright); contrast about 1.3:1 (0.3 stops); soft light (almost no shadow edge)
- Text track (captions / bottom): 不过这油不是谁偷偷加的 / 神秘物质
- Text track (text / middle): 咸
- Text track (title / top): 咸鸭蛋为什么会流油？ / 注意！
- Remake: 3D rendering: first/last-frame AI generation gives a close approximation; for exact multi-angle work go 3D modeling + Blender rendering
  - First-frame image prompt: 3D-rendered fluffy cartoon monkey plushies, dense realistic fur all over, big smiling eyes, one yellow and one brown, a salted duck egg cut open and leaking oil and a small brown jar on the table, minimal clean background, soft studio lighting, vertical composition.
  - Video prompt: fixed camera, two fluffy cartoon monkeys stand at a table, the brown monkey blinks and raises a hand slightly and waves, a red prohibition sign slides in from the right, the characters' fur is fine and flicker-free, the whole frame steady and lifelike.

### #20 01:00.750–01:03.000 · 3D rendering + code / editor overlays (text labels, title)

![shot_020](shots/shot_020.jpg)

- Probabilities: 3D rendering 96%, code / template rendering 3%, 2D animation 0% (fused)
- Material: rendered by the producer
- Likely tool: Blender
- Rendered texture: flat color blocks 61%, 95% of pixels use only 31 colors, straight edges 41%
- [VLM] A perfectly clean 3D wooden balance scale and yolk sphere models, no live-footage noise or lens flaws
- [VLM] The wood-grain texture maps regularly; the scale tilt and falling sphere are smooth physics animation with no AI flicker or warping
- [VLM] A 2D composited glowing star-burst effect above
- Picture: a 3D cartoon-style scene. A wooden balance scale sits in the center with a yolk ball on the left pan; a small yolk ball drops from above and the left side sinks, and twinkling stars appear above the scale. Fixed view, soft top light.
- Framing / light: medium shot; eye level; tripod / stand, fixed; frontal soft light (flat); key frontal (from the camera direction, both sides of the subject equally bright); contrast about 1.0:1 (0.0 stops); soft light (almost no shadow edge)
- Text track (text / bottom): 这些油本来就是蛋黄里的脂肪
- Text track (title / top): 咸鸭蛋为什么会流油？ / 注意！
- Remake: 3D rendering: first/last-frame AI generation gives a close approximation; for exact multi-angle work go 3D modeling + Blender rendering
  - First-frame image prompt: 3D cartoon render, minimal gray-green gradient background and white tabletop. A wooden two-pan balance scale in the exact center, a smooth round orange-yellow yolk ball on the left pan, the scale tilted slightly to the left. Soft diffuse light, clean minimal style.
  - Video prompt: fixed eye-level camera, a 3D-rendered wooden balance scale. A small yolk ball drops straight down from above and merges into the large yolk ball on the left pan, the beam tilts down to the left, then a few sparkling four-point stars pop up above it, and the scale comes to rest tilted.

### #21 01:03.000–01:06.542 · 3D rendering + code / editor overlays (captions, title)

![shot_021](shots/shot_021.jpg)

- Probabilities: 3D rendering 98%, 2D animation 0%, AI-generated video 0% (fused)
- Material: rendered by the producer
- Likely tool: Blender / C4D
- Rendered texture: flat color blocks 47%, 95% of pixels use only 74 colors, straight edges 35%
- [VLM] Character fur is rendered finely with sharp edges, none of the flicker or smearing common in AI video
- [VLM] Background geometry (mountains, spheres) has smooth edges and stable, consistent lighting
- [VLM] Character motion (blinking, small movements) is precisely controlled, with no deformation at all
- Picture: three fluffy cartoon animal characters stand side by side, minimal geometric mountains and floating spheres in the background, the characters blink and sway slightly.
- Framing / light: medium shot; moving (gimbal / handheld pan); frontal soft light (flat); key frontal (from the camera direction, both sides of the subject equally bright); contrast about 1.0:1 (0.0 stops); soft light (almost no shadow edge)
- Text track (captions / bottom): 一滴没多 / 一滴也没少
- Text track (title / top): 咸鸭蛋为什么会流油？ / 注意！
- Remake: 3D rendering: first/last-frame AI generation gives a close approximation; for exact multi-angle work go 3D modeling + Blender rendering
  - First-frame image prompt: 3D cartoon style, three cute fluffy animal characters (green, brown, gray) stand side by side, minimal geometric mountains and two floating pink spheres in the background, soft top light, high-quality 3D render, rich fur detail.
  - Video prompt: camera fixed, three fluffy cartoon animal characters stand side by side, blinking and swaying slightly, the geometric spheres in the background stay floating and still, lighting steady.

## Notes

- Deterministic forensics are reliable for "is it moving, is it a whole-image push, is the texture rendered, was there a 24→30 frame-rate conversion"; telling live footage, AI and 3D apart relies on the VLM and human review.
- Platform transcoding usually strips metadata; missing metadata does not mean there was none.