What's new?
FFmpeg video processing support: FFmpeg is now integrated across the app enabling advanced video manipulation capabilities within workflows. This unlocks more complex video processing operations that were not previously possible through the platform.
FPS controls for video creation and upscaling: You can now set a target frame rate when creating videos and when upscaling existing footage, giving you more precise control over output quality and compatibility with downstream delivery pipelines.
Keep audio option for video editing: A new toggle lets you preserve the original audio track when editing videos with prompts, so you do not have to re-add audio after making visual edits.
Social media size presets: Instagram, Facebook, LinkedIn, and other platform-optimized size presets are now available directly in image composition workflows, removing the need to manually enter dimensions for common formats.
Multi-workspace subscription management: Parent workspaces can now allocate creative units to child workspaces and transfer credits between them, with support for auto-transfers from invoices and manual allocation controls. Built for teams operating across multiple workspaces under a single subscription.
Video timeline blueprint node: A new video timeline node type has been registered in Blueprints as the foundation for upcoming video editing workflows. Not yet active in production runs.
AI-assisted forge execution via MCP: AI agents can now directly execute image, video, 3D, and audio generation through MCP tools, with built-in cost estimation and session tracking before and during generation runs. Agents can also check Creative Unit balance prior to executing, so you avoid unexpected spend. Reach out to us to access this feature.
Bug fixes & improvements
Workflow and activity feed permissions: Members without workspace admin permissions can no longer see restricted navigation items in workflows and the activity feed.
LTX v2.3 video generation restored: Updated the model endpoint URL to resolve failures that were preventing LTX v2.3 video generation from completing.
SD3 image conversion fixed: Corrected the image field format from URL strings to binary data, resolving API compatibility issues with Stability AI models.
Workspace creation redirect fixed: Users are now properly navigated into their new workspace after creation, instead of being left on the creation page.
Form validation on style changes: Fields now revalidate correctly when switching styles, surfacing errors that were previously hidden until submission.
Video creation from object tiles: Right-clicking an object to create a video now correctly extracts the first frame instead of passing an invalid array, resolving failures in that workflow.
Drive sort toggle fixed: Corrected sort direction values so ascending and descending sort in Drive now behave as expected.