Source: the first 20 seconds of 4 highly liked videos from Douyin "Li Yifan," analyzed on 2026-10-27. See samples at
references/samples.md. This Skill only distills movement, rhythm, and structure; it does not include the other person's appearance, clothing, scene, or verbatim lines. When generating, do not feed the original video to the model as a reference video.
In one sentence
The shot is cut tight, the hands are below the lower edge of the frame most of the time, and they enter the frame for less than 3 seconds total within 20 seconds. Emotion comes entirely from the face: the first frame of the opening is already a smiling face, eyebrows raise and the body leans forward on questions, and the eyes narrow briefly at the end of a sentence.
Suitable / Not suitable
- Suitable: AI videos, tutorials, pain-point openings; the reference image has no hands in frame; first time making a digital human and wanting to prioritize stability
- Not suitable: openings that need strong physical expressiveness (switch to
talking-head-qiuzhi2046's small amount of gestures) - Digital human difficulty:low. It has the lowest hand-glitch risk of the five styles
Camera position and composition
- Seated, tight close-up above the chest: head and shoulders take up most of the frame, with the lower edge cutting at the chest
- Clip a conspicuous lavalier mic at the center of the collar to add realism
- Background: a work environment such as a desk, monitor, green plants, or a wooden pegboard, with warm natural light. Describe it according to your own scene
- Fixed camera position
Hand strategy
| Code | Action | Used for which type of line | Position | Duration |
|---|---|---|---|---|
| L1 | Both hands spread low, palms up, with only the palms and wrists showing from the lower edge of the frame | Question line: "Are you also..." | Lower edge of the frame, below the chest | 0.25–0.5 seconds, a fleeting flash |
| L4 | Both hands held in a loose circle in front of the chest, fingers slightly spread | Promise line: "I've helped you summarize..." | Chest height, just entering the frame | About 1 second |
| L3 | Hands do not enter the frame at all | Most declarative sentences | — | — |
Expression and head
- The first frame of the opening is already a toothy smile (3 of 4 videos)
- Raise eyebrows on question lines, lean the body slightly forward, with a tone of "I understand your pain"
- Narrow or close the eyes briefly at the end of a sentence (about 0.3 seconds)
- The self-introduction line has the brightest smile, with the head gently tilted to one side
- Head nodding is smaller than in the high-energy style; overall, "stays seated"
Opening structure template
| Time | Visual | Action | Line function |
|---|---|---|---|
| 0–5s | Full-screen real person | Smiling face → raised eyebrows and forward lean, L1 low hand spread | Pain-point question |
| 5–18s | B-roll or picture-in-picture | Only facial expression in the small window | Use case footage to act out the pain point |
| 18–20s | Return to full screen | Brightest smile, head tilted | One-sentence self-introduction |
| After 20s | Real person full screen → full-screen caption title | L4 empty hug | Promise: “I’ve summarized the N steps for you in this episode” “All the materials are organized” |
Please write the lines yourself; only learn the function of the sentences.
Digital human generation suggestions
- By default, keep hands out of frame throughout; only add L1 or L4 to 1–2 question or promise sentences
- Explicitly write “both hands below the lower edge of the frame” in the generation description to prevent the model from adding gestures on its own
- Write expressions down to specific moments: at which second to raise an eyebrow, on which word to squint
- Do the B-roll segment from 5–18 seconds at the editing layer to cover the seams
Prohibited actions
- Do not raise hands above the chest
- Do not make any gestures that require the fingers to be seen clearly
- Do not shake the head, do not shrug
- Do not sway the body left and right, do not lean too far away from the chair back
Output format
## 基础姿态
## 逐句动作
| 时间 | 台词 | 画面 | 手（默认画外 / L1 / L4） | 表情与头部 |
|---|---|---|---|---|
## 禁止动作
## 给视频模型的描述
Generation description example (original, can be rewritten)
Close-up, the person sits directly facing the camera, the lower edge of the frame cuts at the chest, both hands are outside the frame. The camera is fixed. At second 0, the person is already smiling with teeth showing, looking at the camera. When starting to speak at second 1, the eyebrows raise, and the body leans forward about 5 degrees. At second 3, both palms face up and only emerge from the lower edge of the frame for about half a second before being withdrawn. Squint slightly on the last word of each sentence. Do not make any other gestures throughout.