01Prompts
Make the subject, action, and camera specific, one at a time
When generating video from text, clarify what is shown, how it moves, and how the camera presents it.
Try it ↗Tips
Tips for better AI videos
From refining short instructions to new workflows using reference assets, video editing, and MCP. Check the examples and requirements, then try them yourself.
02Prompts
For image-to-video, describe the movement
The reference image communicates composition and appearance, so focus your prompt on movement and changes over time.
Try it ↗03Prompts
Change only one thing at a time when comparing
Record conditions and avoid large simultaneous changes to prompts and settings so you can track why a result improved.
Try it ↗04Generation methods
Give image, video, and audio references distinct roles
Models that support multiple reference assets may let you communicate appearance, camera work, and rhythm through different sources.
Try it ↗05Generation methods
Design a change using the first and last images
Modes that accept starting and ending images let you plan where the movement should finish.
Try it ↗06Generation methods
Fix a nearly usable shot with video editing generation
Video editing models let you try changing the necessary parts while preserving useful composition and movement.
Try it ↗07Production workflows
Test a short draft before moving to final quality
Checking composition and movement first can reduce the burden of repeated long or high-resolution prototypes.
Try it ↗08Production workflows
Connect your production agent and generation workflow with MCP
Once you have a working workflow, try a small integration that lets an agent operate the generation tool.
Try it ↗No tips match your filters.
8 tips