> ## Documentation Index
> Fetch the complete documentation index at: https://docs.yarn.so/llms.txt
> Use this file to discover all available pages before exploring further.

# Video generation

> Generate video/image assets from prompts and insert the best results into your project.

## Overview

Use video generation when you need supporting visuals that are hard to capture manually.

This workflow is prompt -> generate -> review -> insert -> refine.

\[insert screenshot of generation panel with prompt field, model selector, and request status]

## Before you start

* Define one clear visual outcome per request.
* Match framing to your project aspect ratio.
* Plan for iteration: first result is a draft, not usually final.

## Prompt structure that performs better

Include:

* subject (what should appear)
* context (where/setting)
* style (visual tone)
* framing (close-up, wide, centered, etc.)
* motion intent (if video behavior matters)

## Steps

1. Write a focused prompt with the structure above.
2. Choose the generation model/settings.
3. Submit request and monitor status.
4. Review variations and shortlist candidates.
5. Insert the best option into the target scene.
6. Keep alternates for fast fallback if edits expose issues.

\[insert screenshot of generated variations with one selected for insertion]

## Check your result

* Generated asset supports the scene objective.
* Quality and framing are usable in timeline context.
* You have at least one fallback variation available.

## Troubleshooting

### Request accepted but no asset appears

Accepted and ready states are different. Wait for completion and refresh request status.

### Result quality is inconsistent

Narrow the prompt scope and regenerate with clearer visual constraints.

### Selected model changed automatically

Fallback routing can switch to compatible models when needed.

## Next

Arrange inserted assets using [Video layouts](/generative-videos/video-layouts).