multishot-ugc

TotalClaw 作者 totalclaw

从单个图像生成 10 个视角/角度变化,用于多镜头 UGC 视频。 ✅ 使用时间: - 拥有英雄形象并需要相机角度变化 - 创建多场景UGC视频(需要不同的镜头) - 想要来自一个来源的特写镜头、广角镜头、侧角镜头 - 构建带有场景变化的视频 ❌ 不要在以下情况下使用: - 还没有英雄形象 → 首先使用 morpheus-fashion-design - 需要完全不同的场景/地点→多次使用Morpheus - 只需要一张图片 → 跳过此步骤 - 想要手动编辑图像 → 使用 nano-banana-pro 输入:单张图像(拿着产品的人) 输出:10 个具有不同视角的 PNG 变体 典型流程:Morpheus → multishot-ugc → 选择最佳 4 个 → veed-ugc every → 远程编辑

安装 / 下载方式

TotalClaw CLI推荐
totalclaw install totalclaw:totalclaw~pauldelavallaz-multishot-ugc
cURL直接下载,无需登录
curl -fsSL https://skills.taituai.com/api/skills/totalclaw%3Atotalclaw~pauldelavallaz-multishot-ugc/file -o pauldelavallaz-multishot-ugc.md
## 概述(中文)

从单个图像生成 10 个视角/角度变化,用于多镜头 UGC 视频。
✅ 使用时间:
- 拥有英雄形象并需要相机角度变化
- 创建多场景UGC视频(需要不同的镜头)
- 想要来自一个来源的特写镜头、广角镜头、侧角镜头
- 构建带有场景变化的视频
❌ 不要在以下情况下使用:
- 还没有英雄形象 → 首先使用 morpheus-fashion-design
- 需要完全不同的场景/地点→多次使用Morpheus
- 只需要一张图片 → 跳过此步骤
- 想要手动编辑图像 → 使用 nano-banana-pro
输入:单张图像(拿着产品的人)
输出:10 个具有不同视角的 PNG 变体
典型流程:Morpheus → multishot-ugc → 选择最佳 4 个 → veed-ugc every → 远程编辑

## 原文

# Multishot-UGC

Generate 10 perspective variations of an image using ComfyDeploy's MULTISHOT-UGC workflow.

## Overview

Multishot-UGC takes a single image and generates 10 different variations exploring different perspectives, angles, and compositions. These variations are designed to be used in VEED lip-sync workflows to create dynamic UGC-style promotional videos with varied camera shots.

## API Details

**Endpoint:** `https://api.comfydeploy.com/api/run/deployment/queue`
**Deployment ID:** `9ccbb29a-d982-48cc-a465-bae916f2c7fd`

## Required Inputs

| Input | Description | Default |
|-------|-------------|---------|
| `input_image` | URL or path to the source image | Required |
| `text` | Description for exploration | "Explora distintas perspectivas de esta escena" |
| `resolution` | Output resolution | "2K" |
| `aspect_ratio` | Output aspect ratio | "9:16" |

## Usage

```bash
uv run ~/.clawdbot/skills/multishot-ugc/scripts/generate.py \
  --image "./person-with-product.png" \
  --output-dir "./multishot-output" \
  [--text "Custom exploration prompt"] \
  [--resolution 1K|2K|4K] \
  [--aspect-ratio 9:16|16:9|1:1|4:3|3:4]
```

### With URL:
```bash
uv run ~/.clawdbot/skills/multishot-ugc/scripts/generate.py \
  --image "https://example.com/image.png" \
  --output-dir "./variations"
```

## Output

The workflow generates 10 PNG images with variations:
- `1_00001_.png` through `10_00001_.png`

Each image explores a different perspective/angle of the original scene while maintaining subject identity and composition coherence.

## Workflow Integration

### Typical Pipeline

1. **Generate hero image with Morpheus/Ad-Ready**
   ```bash
   uv run morpheus... --output hero.png
   ```

2. **Create 10 angle variations**
   ```bash
   uv run multishot-ugc... --image hero.png --output-dir ./shots
   ```

3. **Select best variations for VEED lip-sync**
   ```bash
   # Review shots, then generate videos for chosen ones
   uv run veed-ugc... --image ./shots/3_00001_.png --brief "..."
   ```

## Notes

- Source image should be high quality (at least 1K resolution)
- Works best with images containing a clear subject/person
- Generation takes ~2-3 minutes for 10 variations
- All variations maintain the original aspect ratio unless specified