You can now generate video, work with your own media as inputs, use image and video generation through assistants or direct API calls, and control generation settings based on the capabilities of the model you choose.
The goal is simple: give your AI applications access to more than text without adding another generation stack to your architecture.
Video Generation Is Here
Backboard now supports video generation directly through the API.
You can generate videos from a prompt, use supported media as inputs, and work with video models from the same infrastructure you already use for your AI applications.
Assistants can also use video generation as a tool during a conversation.
Edit, Extend, and Upscale Video
Compatible video models can do more than generate from scratch.
Depending on the model, you can also:
Edit existing video
Extend generated video
Upscale video output
This makes it possible to build workflows around existing media rather than treating every generation as a blank canvas.
More Flexible Image Generation
You can now generate images directly from a prompt without creating an assistant or thread first, making it easier to add image generation to existing workflows.
You can also control output settings based on the capabilities of the model you choose.
Bring Your Own Media
Generation can also start with media you already have.
Depending on the model, you can use:
An image as a starting point for animation
First and last frames to guide a video
Image, video, and audio references
Supported inputs vary by model, and Backboard validates the configuration against what the selected model supports.
Control the Output
Different models support different generation controls.
Backboard lets you configure options such as:
Resolution
Aspect ratio
Duration
Image quality
Generated audio
Backboard validates these settings against the capabilities of the model you're using, so your application can work with different models without having to hard-code every model's limitations.
An Updated Model Library
We've also updated the Backboard Model Library with dedicated image and video cards.
You can now compare:
Supported inputs and outputs
Generation options
Model capabilities
Pricing
This makes it easier to choose the right model before you build a generation workflow around it.
Updated SDKs
Image and video generation is available in the latest Backboard SDKs:
Python: v1.5.17TypeScript: v1.5.17
Both SDKs support the new generation features, media inputs, and configuration options.
New Documentation
We've also added new documentation and examples covering:
Assistant-based image and video generation
Direct generation from a prompt
Supported media inputs
Generation configuration
Python and TypeScript SDK usage
Available Now
Image and video generation is now available through the Backboard API.
Explore the documentation → docs.backboard.io
Explore models → Backboard Model Library
The same infrastructure you use for text-based AI can now power media generation too.
SHARE