Video Ad Deconstructor
AI-powered deconstruction of video ad creatives into actionable marketing insights.
What This Skill Does
- - Generate Summaries: Product, features, audience, CTA extraction
- Deconstruct Marketing Dimensions: Hooks, social proof, urgency, emotion, etc.
- Support Multiple Content Types: Consumer products and gaming ads
- Progress Tracking: Callback support for long analyses
- JSON Output: Structured data for downstream processing
Setup
1. Environment Variables
CODEBLOCK0
2. Dependencies
CODEBLOCK1
Usage
Basic Ad Deconstruction
CODEBLOCK2
Full Deconstruction
CODEBLOCK3
Output Structure
Summary Output
CODEBLOCK4
Deconstruction Output
CODEBLOCK5
Marketing Dimensions Deconstructed
| Dimension | What It Extracts |
|---|
| INLINECODE0 | Opening hooks from transcript |
| INLINECODE1 |
Attention-grabbing visuals |
|
text_hooks | On-screen text hooks |
|
social_proof | Testimonials, user counts, reviews |
|
urgency_scarcity | Limited time offers, stock warnings |
|
emotional_triggers | Fear, desire, belonging, etc. |
|
problem_solution | Pain points and solutions |
|
cta_analysis | Call-to-action effectiveness |
|
target_audience | Who the ad targets |
|
unique_mechanism | What makes product special |
Customizing Prompts
Edit prompts in prompts/marketing_analysis.md to customize:
- - What dimensions to analyze
- Output format
- Scoring criteria
- Gaming vs consumer product focus
Common Questions This Answers
- - "What hooks does this ad use?"
- "What's the emotional appeal?"
- "How does this ad create urgency?"
- "Who is this ad targeting?"
- "What social proof is shown?"
- "Deconstruct this competitor's ad"
视频广告解构器
基于AI的视频广告创意解构,转化为可执行的营销洞察。
该技能的功能
- - 生成摘要:提取产品、功能、受众、行动号召
- 解构营销维度:钩子、社会认同、紧迫感、情感等
- 支持多种内容类型:消费品和游戏广告
- 进度追踪:长分析的回调支持
- JSON输出:供下游处理的结构化数据
设置
1. 环境变量
bash
Gemini必需
GOOGLE
APPLICATIONCREDENTIALS=/path/to/service-account.json
2. 依赖项
bash
pip install vertexai
使用方法
基础广告解构
python
from scripts.deconstructor import AdDeconstructor
from scripts.models import ExtractedVideoContent
import vertexai
from vertexai.generative_models import GenerativeModel
初始化Vertex AI
vertexai.init(project=your-project-id, location=us-central1)
gemini_model = GenerativeModel(gemini-1.5-flash)
创建解构器
deconstructor = AdDeconstructor(gemini
model=geminimodel)
创建提取的内容(来自video-ad-analyzer或手动)
content = ExtractedVideoContent(
video_path=ad.mp4,
duration=30.0,
transcript=厌倦了凌乱的线缆?来认识一下CableFlow...,
text_timeline=[{at: 0.0, text: [今日五折优惠]}],
scene_timeline=[{timestamp: 0.0, description: 用户对缠绕的线缆感到沮丧}]
)
生成摘要
summary = deconstructor.generate_summary(
transcript=content.transcript,
scenes=0.0s: 用户对缠绕的线缆感到沮丧,
text_overlays=今日五折优惠
)
print(summary)
完整解构
python
解构所有营销维度
def on_progress(fraction, dimension):
print(f进度: {fraction*100:.0f}% - 已分析{dimension})
analysis = deconstructor.deconstruct(
extracted_content=content,
summary=summary,
is_gaming=False, # 游戏广告设为True
onprogress=onprogress
)
访问各维度
for dimension, data in analysis.dimensions.items():
print(f\n{dimension}:)
print(data)
输出结构
摘要输出
产品/应用: CableFlow线缆整理器
主要功能:
磁吸设计: 自动整理线缆
通用适配: 适用于所有线缆类型
优质材料: 耐用硅胶材质
目标受众: 对线缆管理感到困扰的技术用户
行动号召: 立即下单,享受五折优惠
解构输出
python
{
spoken_hooks: {
elements: [
{
hook_text: 厌倦了凌乱的线缆?,
timestamp: 0:00,
hook_type: 问题式提问,
effectiveness: 高 - 直击痛点
}
]
},
social_proof: {
elements: [
{
proof_type: 用户数量,
claim: 超过100万满意客户,
credibility_score: 7
}
]
},
# ... 更多维度
}
解构的营销维度
| 维度 | 提取内容 |
|---|
| spokenhooks | 开场钩子(来自转录文本) |
| visualhooks |
吸引注意力的视觉元素 |
| text_hooks | 屏幕文字钩子 |
| social_proof | 用户评价、用户数量、评论 |
| urgency_scarcity | 限时优惠、库存警告 |
| emotional_triggers | 恐惧、渴望、归属感等 |
| problem_solution | 痛点和解决方案 |
| cta_analysis | 行动号召有效性 |
| target_audience | 广告目标受众 |
| unique_mechanism | 产品独特之处 |
自定义提示词
编辑prompts/marketing_analysis.md中的提示词进行自定义:
- - 要分析的维度
- 输出格式
- 评分标准
- 游戏广告与消费品广告的侧重点
常见问题解答
- - 这个广告使用了什么钩子?
- 情感诉求是什么?
- 这个广告如何制造紧迫感?
- 这个广告的目标受众是谁?
- 展示了什么社会认同?
- 解构这个竞争对手的广告