If you have ever opened an AI video tool, typed a long prompt, and got back something that looks nothing like what you imagined, you are not alone. Most bad results are not a model problem. They are a prompt-structure problem.
I sell AI short-video prompt packs, but I also keep a free starter set. This post is about the free set, and how to get something usable out of it in one sitting.
The free pack
The starter pack has 9 templates. Download it here:
https://zzstudio.itch.io/ai-video-prompt-starter-pack
Each template separates three things that beginners usually mash together:
- Subject — what is in the shot
- Camera move — how the camera behaves
- Pacing — how fast the action and cuts feel
Change one variable at a time. If you change all three at once, you will never know what fixed — or broke — the shot.
A 10-minute test you can run today
Pick one template. Generate the same idea three times:
- Version A: exactly as written
- Version B: same subject and pacing, different camera move
- Version C: same subject and camera, different pacing
Keep the version that looks most intentional, not the version that looks most random. AI video rewards a clear, boringly specific instruction more than a long, poetic one. Name the subject, the action, the light, and the camera in that order, and stop adding adjectives once the shot is already clear.
That is the whole method behind the paid packs too. The free set teaches the structure. The full Vol.1 set (36 templates) just gives you more ready-made combinations to swap variables in, and the Kling UGC ad pack (20 image-to-video workflows) applies the same structure to product ads.
- Full Vol.1: https://zzstudio.itch.io/ai-video-prompt-pack-vol1
- Code and examples: https://github.com/zz583/ai-video-prompt-pack
What to try next
If you try the free pack, tell me which template you used and what you changed first — subject, camera, or pacing. That one choice tells you more about your style than ten random generations.
No sign-up is needed for the free pack. Start with one template, change one variable, and keep the version you would actually post.
Top comments (0)