AI Media Generator lets you create videos, images, and audio from text prompts using different AI models.
How to access AI Media Generator
To open AI Media Generator in Video Editor, do the following:
Open a project in Video Editor.
In the sidebar, click the Generate icon.
The generation panel will appear with three tabs: Video, Image, and Audio.
How to generate media
To create new assets using the AI Media Generator, do the following:
In the generation panel, select the tab for the type of media you want to create (Video, Image, or Audio).
Click the Model dropdown to select your preferred AI engine (e.g., Google Veo, Sora, etc.).
In the Prompt box, type a description of what you want to generate.
Adjust the available parameters for your selected model. Depending on the type of asset you create and the model, you may be able to select the aspect ratio (e.g., 9:16, 1:1), duration, quality, and some other options.
-
If the selected model supports image inputs, you can upload photos from your device to guide the generation process. To add a file, simply tap the corresponding image slot. Depending on the model, different slots may be available:
Start image: The image will serve as the first frame of your asset.
End image: The image will serve as the final frame of your asset.
Reference image: The image will be used to guide the style, character, or composition of the generated content.
Click the Generate button. For models that require an image input, the button appears only after you add an image. The button will also display the number of credits required for the operation.
Once the process is complete, the generated asset will be automatically added to your project.
Each generation consumes credits from your balance. For details about credits, see How do credits work?
Tips for better results
You can generate different types of assets for your projects. For example, create video scenes and animations, images ranging from realistic portraits to illustrations and logos, or audio such as music and sound effects.
To help the model better interpret your request, follow these guidelines:
- Be clear and specific: Describe the subject, setting, action, and other important details. Instead of "a car", try "a vintage red convertible driving on a coastal road during golden hour".
- Describe the style: Mention the aesthetic you want, such as "photorealistic", "3D render", "oil painting", or "cinematic lighting".
- Consider the language: Language support varies between AI models, but many models understand English best. Keep your prompt clear and simple. If the result doesn't match your request, try rephrasing it in English or using another model.
- Use references: If supported by the selected model, upload a reference image to help guide the style, character, or composition.
- Edit an image: To modify your own image, add it as a reference in the image generator, then describe what you want to change. This option is available only for models that support reference images.
AI-generated results may not always match your prompt exactly, as each model has its own capabilities and limitations. If you don't get the desired result, try another model.