🦞 小龙虾品牌

Created by 小龙虾 AI 工作室

"小龙虾，有大钳（前）途！"

Contact for custom services: +86 15805942886

Need custom AI video workflows, automation setup, or bulk content generation? Reach out!

Video Generator + Auto Post

Generate AI videos locally and publish them to social media platforms automatically.

Requirements

Hardware

• - GPU: NVIDIA RTX 3060 12GB or better (recommended)
• RAM: 16GB+ system memory
• Storage: 50GB+ free space on D: drive (recommended)

Software

• - ComfyUI Desktop - Install from https://comfyui.org
• Python 3.10+ - For scripts
• Node.js 18+ - For CLI tools

Quick Start

Step 1: Install ComfyUI

CODEBLOCK0

Step 2: Install Video Models

In ComfyUI Manager, install:

• - Stable Video Diffusion (SVD)
• AnimateDiff
• ModelScope Text-to-Video

Step 3: Configure Output Paths (D: Drive)

Recommended: Store videos on D: drive to save C: space

Edit comfyui/settings.json:
CODEBLOCK1

Or use command line:
CODEBLOCK2

Step 4: Generate Your First Video

1. 1. Open ComfyUI
2. Load "Text to Video" workflow
3. Enter your prompt
4. Click "Generate"
5. Wait 2-5 minutes (depends on GPU)

Workflows

Text-to-Video (SVD)

Best for: Short clips (2-4 seconds), realistic motion

CODEBLOCK3

Settings:

• - Resolution: 1024x576 or 576x1024
• Frames: 25
• FPS: 6
• Motion bucket: 127 (adjust for more/less movement)

Image-to-Video

Best for: Bringing static images to life

CODEBLOCK4

Tips:

• - Use high-quality source images
• Add motion hints in prompt
• Experiment with motion bucket (100-150)

AnimateDiff (Longer Videos)

Best for: 5-15 second animated clips

CODEBLOCK5

Settings:

• - Context length: 16 frames
• Context overlap: 4 frames
• Model: mmsdv15_v2.ckpt

Auto-Post to Social Media

Platform Specifications

Automated Posting Script

See scripts/auto-post.py for automation.

Setup:

1. 1. Install dependencies: INLINECODE2
2. Configure API keys in INLINECODE3
3. Run: INLINECODE4

Content Strategy

For each video:

1. 1. Generate 3-5 caption variations
2. Add relevant hashtags (5-10)
3. Schedule posts at optimal times
4. Engage with comments

See references/caption-templates.md for examples.

Batch Processing

Generate multiple videos at once:

CODEBLOCK6

Optimization Tips

Speed Up Generation

1. 1. Use FP16 precision - 2x faster, minimal quality loss
2. Reduce frames - 16 frames instead of 25 for drafts
3. Lower resolution - 512x512 for testing, upscale later
4. Batch multiple prompts - Process in parallel

Improve Quality

1. 1. Use better prompts - Be specific about motion
2. Adjust motion bucket - Higher = more movement
3. Add negative prompts - Reduce artifacts
4. Upscale output - Use ESRGAN or similar

Troubleshooting

Out of Memory (OOM)

Fix:

• - Reduce resolution
• Decrease batch size to 1
• Close other GPU applications
• Use --lowvram flag

Slow Generation

Fix:

• - Check GPU utilization (should be 90%+)
• Update NVIDIA drivers
• Use INLINECODE7
• Close background apps

Poor Video Quality

Fix:

• - Use more detailed prompts
• Try different models (SVD vs AnimateDiff)
• Adjust motion settings
• Generate at higher resolution, then compress

Scripts

• - scripts/generate-video.py - Single video generation
• INLINECODE9 - Batch processing
• INLINECODE10 - Auto-publish to social media
• INLINECODE11 - Compress for platforms

References

• - references/prompt-guide.md - Writing effective video prompts
• INLINECODE13 - Detailed platform requirements
• INLINECODE14 - Social media caption templates
• INLINECODE15 - ComfyUI workflow examples

Assets

• - assets/default-workflow.json - Basic text-to-video workflow
• INLINECODE17 - Recommended settings per platform
• INLINECODE18 - Curated hashtag collections

Commands / Triggers

Use this skill when:

• - "Generate a video of..."
• "Create AI video for social media"
• "Auto-post this video to..."
• "Batch generate videos from prompts"
• "Optimize video for TikTok/Instagram"

📊 Performance Benchmarks (RTX 3060 12GB)