Home / Tools / AI Video Prompt Generator
AI Video Prompt Generator
A free AI video prompt generator: write the subject, action and scene, tap preset camera moves, shot sizes, lighting and pacing, and get a video prompt for Sora, Veo, Kling, Runway, Hailuo or Wan, as a paragraph, a shot list or the official MiniMax H3 format.
Everything happens in your browser. Nothing you type is uploaded.
Quick templates
Tap one to fill everything in, then change the text to your own idea.
1. Describe what you want
2. Tap the presets
How to use it
- Write the subject and what happens. In a video the action matters most, like "the kitten jumps onto the windowsill".
- Tap a shot size, camera movement, lighting and pacing. One camera move per clip is the most reliable.
- Pick the format, length and aspect ratio, press Copy, and paste it into your AI video tool.
Tips for better prompts
- Write the action as a clear sequence: "first..., then...". One or two actions are plenty for a 5 to 10 second clip.
- Choose one camera move. Push-in plus orbit plus handheld shake at once often warps the picture.
- For dialogue, put it in "Sound / dialogue" and say who speaks. Not every tool generates sound.
- Make a 5-second test first. Once the direction is right, render the longer or higher-quality version and save credits.
Common questions
What is the difference between Paragraph and Shot list?
Paragraph is one block of text that most video tools take as it is. Shot list puts style, scene, action and camera on separate lines, which is handy when you want to tweak each part or hand it to someone else.
Will the tool always follow the length and aspect ratio?
Not always. Many tools set length and aspect ratio in their own menus, and the prompt only backs that up. Pick the same settings in the tool too.
What is different about the MiniMax H3 format?
H3 generates picture and sound together, and its official guide splits the prompt into three parts: integrated_multimodal_description (picture, action, camera, dialogue), overall_soundscape (ambient and action sounds) and non_diegetic_music (score only the audience hears). Dialogue is tagged with a speaker (S1) and its language and kept word for word. The H3 style templates are adapted from the style rules in the official MiniMax Skills.
Is what I type saved anywhere?
No. The prompt is put together in your browser and never sent to a server.