How can I generate custom video and image assets in Synthesia?

📌 This feature is available as an add-on to all plans and consumes credits. Enterprise plans can generate FLUX.2 images for free.

If you would like to upgrade your plan, check out our upgrade guide

Generate AI assets using text prompts

Open a video project and navigate to the Media tab at the top of the page.
Select the Create with AI tab.
Select the type of asset you would like to create: Video or Image.
Select Prompt to describe the asset yourself, or select Presets to start from a ready-made setup.
- To use a preset, choose one from the gallery and select Next to tailor it to your video.
Choose the AI model you want to use from the model dropdown:
- Video assets:
  - Gemini Omni (beta): 48 credits, whether or not audio is included.
  - Veo 3.1: 96 credits without audio, or 192 credits with audio.
  - Veo 3.1 Fast: 48 credits without audio, or 60 credits with audio.
- Image assets:
  - FLUX.2 (by Black Forest Labs): 1080p, 3 credits per image. Free for Enterprise plans.
  - Google Nano banana Pro: 1080p, 9 credits per image.
  - OpenAI GPT Image 2: 1080p, 9 credits per image.
Select an avatar or space already in your scene as a reference, so the generated asset matches the avatar or space in your scene.
Choose a Style to set the visual look of your asset:
- Select a Brand Kit to apply your brand colors and fonts.
- Or choose a Style, such as Painter 2D, Comics, or Illustrated, for a different artistic look.
In the prompt input box, describe the video asset you want.
- Press Tab to use the auto-prompt feature, which uses your video script as context to generate a prompt for you.
Set the output options before you generate:
- For a video, choose a clip duration, such as 4, 6, or 8 seconds.
  - The available durations depend on the model you select.
- Choose an aspect ratio: 1:1, 3:4, 4:3, or 16:9.
  - The available aspect ratios depend on the model you select.
  - The aspect-ratio picker appears in Prompt mode. Presets mode does not support custom ratios yet.
Press Enter to begin generating the asset with your selected options.
Once generated:
- Your asset appears in the media panel.
- You can add it to any scene in any video.
- Assets are stored at the user (not workspace) level.
- Use the three-dot menu to copy the asset ID or prompt.

✍️ Can't see the Gemini Omni model?

If you are on an Enterprise plan and do not see the Gemini Omni generative model, ask your workspace admin to confirm that the Gemini Omni feature setting is enabled under Workspace Settings → General → Feature Settings.

💬 FAQs

What is the difference between Presets and Prompt in Create with AI?

Presets let you start from a ready-made setup and tailor it to your video. Prompt lets you describe the asset in your own words, and it is the only mode that supports custom aspect ratios.

What is Veo 3.1?

Google’s latest text-to-video model that creates 8-second clips with motion and advanced effects.

What is the difference between a Brand Kit and a Style?

A Brand Kit applies your saved brand colors and fonts to a generated asset. A Style applies a visual art direction, such as Painter 2D, Comics, or Illustrated, giving the asset a distinct look beyond your brand colors. You can choose either one when you generate a video or image asset.

What is Gemini Omni?

Gemini Omni is a Google text-to-video model, currently in beta, that creates 8-second clips at 720p and supports audio.

What are FLUX.2, Nano banana 2, and GPT Image 2?

FLUX.2: A fast, cost-effective image generator—3 credits/image (or free for Enterprise).
Nano banana Pro: A high-end image model offering superior detail—9 credits/image.
OpenAI GPT Image 2: Generate detailed, photorealistic images from your prompts —9 credits/image.

Can I turn an image into a video?

Yes. Open the three-dot menu on an image in your media panel and select Convert to video.

How long does it take to generate a text to video asset?

Generation can take 3–5 minutes.

Can I reuse generated clips?

Yes, they’re saved in your media panel and tied to your user account. Future updates will allow you to add them to the shared workspace media library.

How many credits does a generative asset consume?

Each generative model consumes a different amount of credits. To learn more about credits check out the article: What are credits, and how do they work for Enterprise customers in Synthesia?

How do I enable or disable generative video assets in Synthesia?

Admins can turn this feature on or off for the entire organization or specific workspaces via Feature Settings under General settings.

📚 To learn more about how moderation works for AI-generated assets, check out Why was my AI-generated asset blocked?

To learn more about credit consumptions, check out our credits for self serve users and credits for Enterprise customers articles.

Note

Generative Video assets are artificial intelligence components created, trained, and deployed by third parties, rather than by Synthesia itself. When making these features available through the Services to customers, Synthesia takes steps to test and verify their performance and accuracy and to apply similar governance and moderation controls that otherwise apply to the native components of the Services. When possible, Synthesia will onboard the providers of these components as sub-processors. The goal is to give you a single, trusted environment for video creation without the burden of sourcing and integrating these models on your own.

Unlike Synthesia's native components, generative video assets are often general application models that can be prompted to create a wide variety of outputs. The content produced by generative video assets ultimately reflects the prompts that you provide, notwithstanding the governance and moderation controls. For that reason, customers must exercise good judgement should they elect to use, as they will be responsible in the event they prompt these components in a manner that violates the Acceptable Use Policy , for example, by generating and publishing an asset that infringes a third party's intellectual property rights.