- Home
- Skills
- Cdeistopened
- Skill Stack
- Video Generator
video-generator_skill
- TypeScript
1
GitHub Stars
1
Bundled Files
2 months ago
Catalog Refreshed
4 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 cdeistopened/skill-stack --skill video-generator- SKILL.md13.7 KB
Overview
This skill generates short-form AI videos using two providers: Google VEO 3.1 for native synchronized audio and OpenAI Sora for higher visual fidelity and longer clips. It exposes model, duration, aspect, resolution, negative prompts, and variation controls so you can iterate from fast drafts to final-quality outputs. The workflow emphasizes one-shot prompts, clear camera language, and separate audio layering for best results.
How this skill works
You provide a concise cinematic prompt and choose a provider/model; the tool calls the chosen API to create MP4 outputs and polls until the async job completes. VEO supports native audio, negative prompts, and configurable resolutions; Sora focuses on visual quality, artistic styles, and clips up to 12 seconds. Multiple variations can be generated and saved to a specified output directory for review and iteration.
When to use it
- Choose VEO when you need synchronized dialogue or sound effects and 4–8s clips.
- Choose Sora for higher visual quality, artistic motion, or 12s clips.
- Use VEO Fast or short durations for rapid drafting and prompt refinement.
- Generate 2–4 variations to pick the best take before finalizing.
- Use vertical 9:16 for social (Reels/TikTok) and 16:9 for headers or product reveals.
Best practices
- Keep prompts 150–300 characters and describe one continuous action per shot.
- Structure prompts with shot, setting, subject/action, audio (VEO), then style/grade.
- Treat audio as its own layer; use colon syntax for dialogue in VEO and keep lines under seven words.
- Start with fast/draft mode, iterate 2–3 times, then render final quality and duration.
- Include a minimal negative prompt (e.g., "text, watermarks") for VEO to avoid overlays.
Example use cases
- Social teaser vertical clip: close-up motion with strong texture and short SFX (use Sora or VEO for audio).
- Podcast header: ambient 16:9 clip with subtle motion and soft room sounds (VEO recommended).
- Product reveal: slow orbit, dramatic studio lighting, 4–8s, use VEO for 4K or Sora for stylized look.
- Extended cinematic scene: 12s Sora clip for contemplative walking or landscape sequences.
- Iterative drafts: VEO fast with short duration to refine camera move and subject action before final render.
FAQ
No. Sora does not produce native audio. Use VEO when you need dialogue, sound effects, or ambient sound mixed into the clip.
What durations and resolutions are supported?
VEO supports 4, 6, or 8s and resolutions up to 4K; Sora supports 4, 8, or 12s with fixed sizes tied to aspect ratio. Use 9:16 for vertical social and 16:9 for widescreen.