Back to skills
SKILL.md
Video Generation
ASecurityCreate AI videos with optimized prompts, motion control, and platform-ready output.
- 2 stars
- 0 votes
- 1 copy
- 1 view
- Added November 11, 2026
Works with
Security analysis
100/100Pro scans all 10 files and shows the line behind each finding
npx -y skills add johnalbertini14-glitch/openclaw-skills --skill video-generation --agent claude-codeAre you the author of Video Generation?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/johnalbertini14-glitch-video-generation)---
name: Video Generation
description: Create AI videos with optimized prompts, motion control, and platform-ready output.
metadata: {"clawdbot":{"emoji":"π¬","os":["linux","darwin","win32"]}}
---
# AI Video Generation
Help users create and promote AI-generated video content.
**Rules:**
- Ask what they want: text-to-video, image-to-video, or video editing
- Check provider files for setup: `runway.md`, `luma.md`, `kling.md`, `pika.md`, `stable-video-diffusion.md`, `replicate.md`
- Check `api-patterns.md` for async polling, backoff, and caching
- Check `promotion.md` for distribution and marketing strategies
- Start with lowest resolution/duration to validate prompt before scaling
---
## Platform Selection
| Use Case | Recommended |
|----------|-------------|
| Complex motion, professional | Runway Gen-4.5 |
| Human motion, longer clips | Kling V2 |
| Stylized, fast iterations | Pika |
| Camera movements, cinematic | Luma Dream Machine |
| Local/private, no API cost | Stable Video Diffusion |
| Quick testing, model variety | Replicate |
---
## Prompting for Video
- **Describe motion explicitly** β "waves crashing", "walking toward camera", not just scene descriptions
- **Include camera direction** β "camera slowly pans left", "dolly zoom", "tracking shot"
- **Front-load subject** β models weight early tokens heavily
- **Under 200 tokens** β longer prompts dilute motion guidance
- **Avoid impossible physics** β results look uncanny without proper setup
---
## Image-to-Video
- Source image quality matters more than prompt β artifacts amplify
- Match aspect ratio BEFORE generation β post-crop destroys quality
- Use images with implied motion (mid-stride, flowing hair)
- Remove text/logos β they warp unpredictably
---
## Aspect Ratios
- **16:9** β YouTube, standard horizontal
- **9:16** β TikTok, Reels, Shorts (vertical mandatory)
- **1:1** β Instagram feed, flexible
- Generate in target ratio β never crop after
---
## Cost Control
- Start with shortest duration (2-4s) to validate
- Preview at lowest resolution (480p/720p)
- Extend good clips rather than regenerating
- Download immediately β signed URLs expire
- Free tiers: Pika (150/mo), Kling (66/day), Luma (30/mo)
---
## Failure Modes
- **NSFW false positive** β rephrase or switch platform
- **Face morphing** β add "consistent face" to prompt
- **Text always broken** β add in post-production
- **Hands/fingers wrong** β frame to minimize visibility
- **Quality degrades >8-10s** β plan for cuts
---
### Current Setup
<!-- Provider: status -->
### Projects
<!-- What they're creating -->
### Preferences
<!-- Settings that work: resolution, duration, style -->
---
*Empty sections = not configured yet. Check providers.md for setup.*
Files in this skill
- SKILL.md
- _meta.json
- api-patterns.md
- kling.md
- luma.md
- pika.md
- promotion.md
- replicate.md
- runway.md
- stable-video-diffusion.md
Attribution
Related skills
View all in content-marketing βComments
Loading commentsβ¦
Newsletter
Get the weekly list of safe new skills, and the ones we pulled.