Project focus
CASE STUDY 02
AIGC Short-Video Workflow | From Structure Breakdown to Script, Shots, and Cut
Break the source viral video into hook, pacing, structure, and conversion paths, then combine that with a product brief to generate reusable scripts, asset mapping, and rough-cut previews.
01
Overview
This is an AI workflow experiment for short-video production. It takes a viral video link, a product brief, and optional assets, then runs link diagnosis, information extraction, and structural analysis before generating product-fit scripts and rough-cut previews.
What is already stable is analysis, scripting, and preview generation rather than a one-click final-video system.
Within the portfolio, it proves how platform insight, product translation, and AIGC execution can be organized into a real workflow.
02
Problem and insight
Problem
Viral-video analysis often ends at observation without becoming a repeatable production method.
At the same time, product content often lacks platform-native structure and hooks.
Insight
The reusable layer is not the exact wording, but the structure: hook, pacing, tension, and CTA placement.
Once the structure is clear, it can be rewritten for a product scenario without becoming imitation.
03
Video pairs
Shows how structure is transferred from source viral videos into output videos.
Pair 01
Original → OutputOriginal viral videoMP4
Original viral video A
The reference input used to analyze the hook, pacing, structure, and conversion path.
→Original → Output
Output videoMP4
Output video A
A productized preview generated from the product brief and structural mapping to validate the script and shot plan.
Pair 02
Original → OutputOriginal viral videoMP4
Original viral video B
The second reference input used to test whether the workflow works beyond a single source clip.
→Original → Output
Output videoMP4
Output video B
A second rough-cut preview generated through the same workflow to test structural transfer under another reference style.
04
Productized short-video workflow
Start from a reference video link and a product brief, then move through extraction, structural analysis, script generation, rough-cut preview, and iteration.
01
INPUT BRIEF
Input the video link and product brief with any optional assets.
02
STRUCTURE DIAGNOSIS
Diagnose the source link and extract available material plus basic metadata.
03
SCRIPT REWRITE
Break down the hook, structure, desire/pain points, pacing, and conversion path.
04
VISUAL MAPPING
Map the structure onto product scenarios such as Citely or Literfy and reposition the value points.
05
PREVIEW & REVIEW
Generate scripts across different product types and content templates.
01
INPUT BRIEF
Input the video link and product brief with any optional assets.
→
02
STRUCTURE DIAGNOSIS
Diagnose the source link and extract available material plus basic metadata.
→
03
SCRIPT REWRITE
Break down the hook, structure, desire/pain points, pacing, and conversion path.
→
04
VISUAL MAPPING
Map the structure onto product scenarios such as Citely or Literfy and reposition the value points.
→
05
PREVIEW & REVIEW
Generate scripts across different product types and content templates.
05
Results and takeaways
Output A
Reusable scripts
Generated multiple short-video scripts from the viral structure and the product brief.
Output B
Rough-cut previews
Created reviewable rough-cut videos from assets and the preview plan.
Output C
Iteration records
Tracked each preview round, the issues found, and the next iteration tasks.
What was validated
After breaking viral structures into hooks, pacing, and conversion beats, the structure can be translated into product scripts more reliably and checked early through rough-cut previews.
Next step
Continue adding structure labels, asset-fit rules, and checklists for different video types to reduce ineffective generation and make each revision round more predictable.