🎬

Make Short Videos

Script → storyboard → clips → voice — a 45s vertical pipeline with reusable prompts

Task

Complete a 45s vertical pipeline: script → 4 shot prompts → clips → VO/captions → pre-export check

Time

~25 min

Suggested tool: HeyGen

Self-rate (1–5)

How this scenario works

Short-video fails usually come from expecting one-shot magic. This scene splits the job: 45s talking-head script (hook + 3 points + CTA) → 4-shot storyboard + Kling/Pika prompts → vertical clips → avatar or VO → caption sync → pacing/typo check before export. Good for: product teasers, knowledge VO drafts, vertical ad sketches. Bad for: impersonating real people, infringing assets, putting customer PII in scripts; read each platform’s ToS before commercial use. Example: topic=sugar-free sparkling water launch; audience=urban 25–35; length=45s vertical; tone=light and trustworthy. Acceptable: script with timecodes + visual cues; 4 copy-ready shot prompts; a draft you’d assemble in CapCut/Jianying. Tool pick: Kling for easier China access and motion; Pika for stylized clips; HeyGen for talking-head avatars. Verify pricing on official sites.

Practice steps (with checks)

1Set acceptance: length (e.g. 45s), aspect (9:16), audience, must-have selling points/CTA. Done when you can state “draft finished” in one sentence.
2Run “Talking-head script” with timecodes (hook → 3 points → CTA). Done when lines are ≤20 chars, conversational, with [visual] cues.
3Mark which shots are live vs AI-generated. Done when ≥4 shots each map to a spoken line.
4Run “Storyboard to video prompts” into 4 vertical shots (3–5s each) with camera motion. Done when prompts paste cleanly into Kling/Pika.
5Generate 2–4 clips; keep only the steadiest motion/composition. Done when subject is clear, no gibberish/watermarks, pacing fits VO.
6Use “Avatar VO” or TTS if needed; align captions in your editor. Done when voice is consistent and subtitles are typo-free.
7Run “Pacing & export check” against the script: hook in 3s, CTA clear. Done when you’d show the draft to a teammate.
8Save script + shot prompts; next time only swap [topic]/[audience]. Done when similar videos start in ≤15 minutes.

Practice Workbench

Video topicAudienceStyle / toneScriptVideo prompt draft

You are a short-video scriptwriter. Write a ~45s vertical talking-head script for sugar-free sparkling water launch.
Audience: urban professionals 25–35; style: light, trustworthy, not hyped

Rules:
1) Hook in 3s → 3 core points → CTA
2) Conversational; ≤20 chars per line
3) Bracket visual/subtitle cues per line
4) No greetings or filler

Output a Markdown table: Timecode | Line | Visual cue | Tone/pacing

Open Kling

Outcome checklist

Self rating (1–5)

More prompts (5)

Storyboard to video prompts

You storyboard for Kling and Pika. Split this script into 4 continuous vertical shots (9:16, 3–5s each) with copy-ready prompts.

Script:
[script]

Rules: subject + continuous action + camera move + light change; no static-only frames; no text/watermarks/fake real people.

Format each: Line | Chinese prompt | English prompt | Params (duration/aspect)

Single-shot iteration

The last shot failed. Rewrite one vertical prompt for [topic], style [style].

Issue + old prompt (in draft field):
[draft]
Line:
[script]

Enlarge subject motion; cleaner background; keep 9:16; avoid text, watermark, extra hands, clutter.
Output: Chinese prompt + English prompt + one-line negatives.

Avatar VO

You advise HeyGen-style avatar VO. Adapt this script for avatar read-aloud without changing facts.

Script:
[script]
Audience: [audience]; tone: [style]

Output:
1) Read-aloud script with clear pauses and pronunciation notes
2) Avatar look direction (formal/casual, M/F/neutral)
3) Vertical backdrop and framing
4) Five caption highlight words
Do not recommend products outside the listed tools.

Pacing & export check

You QA short-video pacing. Review my assembly plan against this script (timeline or shot list OK).

Script:
[script]

List:
A. Hook within 3s?
B. Each of 3 points has a spoken line?
C. Possible typo/awkward read lines
D. Over-promises on brand/claims
E. Five pre-export checks (resolution, safe area, silence, end CTA, platform-sensitive words)
QA only — don’t rewrite the whole script.

Prompt QA (chat model)

You edit AI video prompts. Critique my Kling/Pika prompt: missing observables, vague words, conflicting constraints. Then rewrite a sharper version.

Topic: [topic]; style: [style]

My draft:
[draft]

Recommended tools

Best fitFreemium

Kling

Kuaishou's AI video generation tool, creating high-quality short videos from text and images with realistic physics effects.

AlternativeFreemium

Pika

AI video generation tool by Pika Labs, supporting text/image-to-video with rich creative effects and easy operation.

Free-friendlyPaid

HeyGen

AI avatar video generation platform supporting virtual hosts, video translation, and lip-sync, widely used in marketing and education.

After you leave with templates,

ChooseNarrow tools with Choose LabBrowse guides & compares in Lab Related learning pathCreate Images & Short Videos Try another scenario →

Continue your learning path

Practice a scenario Choose Lab

How this scenario works

Practice steps (with checks)

1Set acceptance: length (e.g. 45s), aspect (9:16), audience, must-have selling points/CTA. Done when you can state “draft finished” in one sentence.

2Run “Talking-head script” with timecodes (hook → 3 points → CTA). Done when lines are ≤20 chars, conversational, with [visual] cues.

3Mark which shots are live vs AI-generated. Done when ≥4 shots each map to a spoken line.

4Run “Storyboard to video prompts” into 4 vertical shots (3–5s each) with camera motion. Done when prompts paste cleanly into Kling/Pika.

5Generate 2–4 clips; keep only the steadiest motion/composition. Done when subject is clear, no gibberish/watermarks, pacing fits VO.

6Use “Avatar VO” or TTS if needed; align captions in your editor. Done when voice is consistent and subtitles are typo-free.

7Run “Pacing & export check” against the script: hook in 3s, CTA clear. Done when you’d show the draft to a teammate.

8Save script + shot prompts; next time only swap [topic]/[audience]. Done when similar videos start in ≤15 minutes.

Practice Workbench

Video topicAudienceStyle / toneScriptVideo prompt draft

You are a short-video scriptwriter. Write a ~45s vertical talking-head script for sugar-free sparkling water launch. Audience: urban professionals 25–35; style: light, trustworthy, not hyped Rules: 1) Hook in 3s → 3 core points → CTA 2) Conversational; ≤20 chars per line 3) Bracket visual/subtitle cues per line 4) No greetings or filler Output a Markdown table: Timecode | Line | Visual cue | Tone/pacing

Outcome checklist

Self rating (1–5)

You storyboard for Kling and Pika. Split this script into 4 continuous vertical shots (9:16, 3–5s each) with copy-ready prompts. Script: [script] Rules: subject + continuous action + camera move + light change; no static-only frames; no text/watermarks/fake real people. Format each: Line | Chinese prompt | English prompt | Params (duration/aspect)

The last shot failed. Rewrite one vertical prompt for [topic], style [style]. Issue + old prompt (in draft field): [draft] Line: [script] Enlarge subject motion; cleaner background; keep 9:16; avoid text, watermark, extra hands, clutter. Output: Chinese prompt + English prompt + one-line negatives.

You advise HeyGen-style avatar VO. Adapt this script for avatar read-aloud without changing facts. Script: [script] Audience: [audience]; tone: [style] Output: 1) Read-aloud script with clear pauses and pronunciation notes 2) Avatar look direction (formal/casual, M/F/neutral) 3) Vertical backdrop and framing 4) Five caption highlight words Do not recommend products outside the listed tools.

You QA short-video pacing. Review my assembly plan against this script (timeline or shot list OK). Script: [script] List: A. Hook within 3s? B. Each of 3 points has a spoken line? C. Possible typo/awkward read lines D. Over-promises on brand/claims E. Five pre-export checks (resolution, safe area, silence, end CTA, platform-sensitive words) QA only — don’t rewrite the whole script.