據 AlphaMoat 統計,截至 2026-09,UGC Campaign Pipeline 累計下載 2,240 次、獲 3 個星標,在全部 5,078 個 AI 技能中排名第 4,262 位,在「內容創作」分類(362 個)中排名第 283 位。
| 平台 | 累計安裝 | 累計下載 | 星標 | 評論 |
|---|---|---|---|---|
| ClawHub ↗ | 61 | 2205 | 2 | - |
| SkillHub ↗ | 168 | 2240 | 3 | — |
各平台的計數口徑不同,統計時間不一致。
該流程涵蓋主視覺圖、圖片變體、素材篩選、影片產生、剪輯成片等步驟,目標輸出為含場景、字幕和品牌標誌的 MP4。流程會呼叫引用的外部圖像或影片產生工具,產品圖片、品牌素材、提示詞與腳本可能傳送給這些服務;中間與成品檔案也可能保留在本機目錄。 流程需另行具備並設定 morpheus-fashion-design、multishot-ugc、veed-ugc 與 Remotion;開始前須審核各工具設定。
資料釋出於 2026-10-27
以下是根據技能文件生成的示例,不代表實際執行結果。
2026-09 起統計方法調整:同名技能按作者分別統計,並覆蓋全部技能。圖中虛線為調整前的月份,與之後的資料不直接比較,也不計算跨越調整的增長。
| 月份 | 累計下載 | 環比 | 星標 |
|---|---|---|---|
| 2026-09 | 2240 | — | 3 |
| 2026-08 | 2095 | +8.8% | 3 |
| 2026-07 | 1926 | +19.5% | 3 |
| 2026-06 | 1612 | +12.3% | 2 |
| 2026-05 | 1435 | — | 2 |
| 2026-03 | 1079 | — | 2 |
以下是作者釋出的原始檔案,按純文字顯示,本站未做修改或稽核;其中的連結不可點選,命令請核實後再執行。抓取於 2026-10-23。
--- name: ugc-campaign-pipeline description: | Complete UGC video campaign pipeline: product → hero image → variations → videos → edited final. ✅ USE WHEN: - User says "crear campaña UGC" or "pipeline completo" - Need end-to-end UGC video production - Starting from product image/URL → final edited video - Want the full Doritos-style workflow
❌ DON'T USE WHEN:
- Just need one step (use individual skills)
- Already have final videos, just editing → use Remotion
- Only need images, no video → use Morpheus only
OUTPUT: Edited MP4 video with multiple scenes + subtitles + logo
---
# UGC Campaign Pipeline
Complete workflow for creating UGC-style promotional videos from a product.
## Pipeline Overview
```
┌─────────────────────────────────────────────────────────────────────┐
│ UGC CAMPAIGN PIPELINE │
├─────────────────────────────────────────────────────────────────────┤
│ │
│ STEP 1: HERO IMAGE │
│ ├─ Input: Product image + model selection │
│ ├─ Tool: morpheus-fashion-design │
│ └─ Output: ~/clawd/outputs/{project}/morpheus/hero.png │
│ │
│ STEP 2: VARIATIONS │
│ ├─ Input: Hero image │
│ ├─ Tool: multishot-ugc │
│ └─ Output: ~/clawd/outputs/{project}/multishot/*.png (10 images) │
│ │
│ STEP 3: SELECTION │
│ ├─ Analyze all 11 images │
│ ├─ Criteria: variety, no errors, lip-sync friendly │
│ └─ Output: 4 best images selected │
│ │
│ STEP 4: SCRIPT │
│ ├─ Write 4-scene dialogue script │
│ ├─ Format: PURE DIALOGUE (no annotations) │
│ └─ Output: 4 lines of dialogue │
│ │
│ STEP 5: UGC VIDEOS │
│ ├─ Input: 4 images + 4 script lines │
│ ├─ Tool: veed-ugc (run 4 times) │
│ └─ Output: ~/clawd/outputs/{project}/ugc/*.mp4 (4 videos) │
│ │
│ STEP 6: FINAL EDIT │
│ ├─ Input: 4 videos + logo │
│ ├─ Tool: Remotion │
│ ├─ Add: subtitles, transitions, logo ending │
│ └─ Output: ~/clawd/outputs/{project}/final/video.mp4 │
│ │
└─────────────────────────────────────────────────────────────────────┘
```
## Execution Checklist
### Before Starting
- [ ] Product image received
- [ ] Brand/product name known
- [ ] Target audience understood
- [ ] Tone defined (casual, professional, energetic)
### Step 1: Hero Image (Morpheus)
```bash
# Select model from catalog
cat ~/clawd/models-catalog/catalog/catalog.json | jq '[.talents[] | select(.gender == "male/female") | {id, name, ethnicity}]'
# Generate hero image
COMFY_DEPLOY_API_KEY="..." uv run ~/.clawdbot/skills/morpheus-fashion-design/scripts/generate.py \
--product "product.jpg" \
--model "models-catalog/catalog/images/model_XX.jpg" \
--brief "..." \
--target "..." \
--aspect-ratio "9:16" \
--output "~/clawd/outputs/{project}/morpheus/hero.png"
```
### Step 2: Variations (Multishot)
```bash
COMFY_DEPLOY_API_KEY="..." uv run ~/.clawdbot/skills/multishot-ugc/scripts/generate.py \
--image "~/clawd/outputs/{project}/morpheus/hero.png" \
--output-dir "~/clawd/outputs/{project}/multishot" \
--resolution "2K" \
--aspect-ratio "9:16"
```
### Step 3: Selection Criteria
Analyze all 11 images for:
- ✅ Face clearly visible (frontal or 3/4)
- ✅ Mouth not obscured
- ✅ No distorted hands/fingers
- ✅ Product visible
- ✅ Different from other selected images
Select 4 images that are:
1. Different angles/perspectives
2. Suitable for lip-sync
3. Error-free
### Step 4: Script Writing
Write 4 lines of pure dialogue:
```
ESCENA 1: [Opening hook - grab attention]
ESCENA 2: [Problem/benefit - relate to audience]
ESCENA 3: [Feature highlight - specific value]
ESCENA 4: [CTA/brand mention - close strong]
```
**Rules:**
- Pure dialogue only
- NO annotations: [excited], (pause), *smiles*
- NO stage directions
- NO tone indicators
- Just what the person says
### Step 5: Generate Videos (VEED)
```bash
for i in 1 2 3 4; do
COMFY_DEPLOY_API_KEY="..." uv run ~/.clawdbot/skills/veed-ugc/scripts/generate.py \
--image "selected_image_$i.png" \
--script "Script line $i" \
--output "~/clawd/outputs/{project}/ugc/escena$i.mp4"
done
```
### Step 6: Final Edit (Remotion)
Create Remotion project with:
- All 4 video clips sequenced
- Animated subtitles for each scene
- Logo animation at end
- Render to final MP4
---
## Script Templates by Industry
### Snacks/Food
```
Escena 1: Che, tenés que probar esto.
Escena 2: [Sabor/textura highlight]. Te pega de una.
Escena 3: Y mirá, no es solo [feature], tiene ese gustito que te deja queriendo más.
Escena 4: [Brand + Product]. Una vez que arrancás, no parás.
```
### Tech/Gadgets
```
Escena 1: Mirá lo que me llegó.
Escena 2: Esto cambia todo. [Key feature].
Escena 3: Y lo mejor? [Secondary benefit].
Escena 4: [Brand]. Ya no vuelvo atrás.
```
### Beauty/Skincare
```
Escena 1: Ok, necesito contarte algo.
Escena 2: Este [producto] es increíble. [Result].
Escena 3: Lo uso hace [tiempo] y mirá la diferencia.
Escena 4: [Brand]. Tu piel te lo va a agradecer.
```
### Fashion
```
Escena 1: Encontré mi nueva obsesión.
Escena 2: Mirá este [prenda]. [Quality/style].
Escena 3: Combina con todo y es súper [comfort/style].
Escena 4: [Brand]. Estilo sin esfuerzo.
```
---
## Output Structure
```
~/clawd/outputs/{project}/
├── morpheus/
│ └── hero.png # Original hero image
├── multishot/
│ ├── 1_00001_.png # 10 variations
│ ├── 2_00001_.png
│ └── ...
├── ugc/
│ ├── escena1.mp4 # Individual scene videos
│ ├── escena2.mp4
│ ├── escena3.mp4
│ └── escena4.mp4
├── assets/
│ └── logo.png # Brand logo
└── final/
└── video.mp4 # Final edited video
```
---
## Quality Checklist (Before Delivery)
- [ ] Hero image: product visible, model looks natural
- [ ] Variations: 4 selected are distinct and lip-sync ready
- [ ] Script: matches brand tone, pure dialogue format
- [ ] Videos: lip-sync quality is good, no artifacts
- [ ] Final: subtitles readable, transitions smooth, logo appears
- [ ] Audio: voice quality clear, timing natural
識別並去掉文字里的 AI 寫作痕跡,讓文字讀起來更像人寫。
把文字里 AI 寫作的典型痕跡挑出來並改寫得更自然,同時保留原意的寫作編輯技能。
識別並去除 AI 生成文字痕跡的中文文字編輯技能。
wechat-search 檢索公眾號熱門文章,並按相關性、熱度和時效整理結果。
從公開短影片分享連結提取標題、簡介和口播文案,並給出改寫版本。
humanizer 為中英文文字提供更自然的表達和風格編輯指導。
寫網文章節並按檔案落盤,聊天裡只回摘要與指標。
把一組行銷方法模組收進同一個技能,按請求路由到對應模組取清單與可交付稿。
資料月份:2026-09 · 下載/星標/安裝數聚合自公開技能註冊庫(ClawHub、SkillHub),按月更新。口徑說明見資料方法。