2.5k
GitHub Stars
2
Bundled Files
2 months ago
Catalog Refreshed
3 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstrat where the catalogue uses aiagentskills.
npx veilstrat add skill openclaw/skills --skill skill-ad-creative-engine- _meta.json299 B
- SKILL.md2.1 KB
Overview
This skill generates polished 1080×1920 short-form video ads (TikTok / Reels / Shorts) from product clips and images. It produces three built-in viral styles — Clean, Meme, and UGC — and runs locally using Python and ffmpeg so no cloud APIs or external services are required. The output is ready-to-publish H.264 video with warm color grade and mobile-optimized framing.
How this skill works
You provide a JSON config that lists scenes (video clips or images), durations, transitions, music, and a chosen style. The Python scripts stitch assets together with overlays, fonts, and transitions, encode to 1080×1920 at 30 FPS, and export H.264 + AAC. Optionally enable beat-sync (requires librosa) to snap cuts to detected beats. Fonts and basic color grading are applied automatically based on the selected style.
When to use it
- Create mobile-first product ads quickly from existing clips and photos
- Prototype multiple ad variations in different viral styles
- Produce privacy-friendly assets entirely on local machines (no cloud)
- Generate social creatives for launches, promos, or organic growth experiments
Best practices
- Prepare high-quality 9:16 clips or portrait-cropped images for best framing
- Keep hooks short and bold for Clean or Meme styles; UGC favors minimal overlay
- Provide a single music track and enable beat_sync only if you want precise cut timing
- Use the included fonts (Anton, Montserrat) or supply matching TTFs in the fonts folder
- Test outputs on a phone before publishing to confirm text size and safe areas
Example use cases
- Turn demo clips and product photos into a 15–30s TikTok ad with a bold hook overlay
- Create humorous meme-style cuts with ALL-CAPS captions for viral testing
- Produce a UGC-style vertical ad that only shows a creator username pill and feels authentic
- Batch-render several style variants for A/B testing creative performance on Reels
FAQ
No. The tool runs locally using Python and ffmpeg; no cloud APIs or accounts are needed.
What formats and resolutions are supported for input assets?
Input supports common video and image formats readable by ffmpeg. For best results use portrait (9:16) video or high-resolution images to avoid upscaling artifacts.
How do I enable beat-synced edits?
Set "beat_sync": true in the config and install librosa. Provide a music track; the script will detect beats and snap cuts to timestamps.