Video Generation
Before
After
Was
Now — and why it changed
Last updated
Video generation was tucked inside the Audio & Video submenu, alongside Text To Speech and Text To Music. This grouping made sense architecturally, but it meant video — one of the most used tools — required two clicks from the top of the nav.
Video now has its own first-class spot in the left rail. The workspace follows the same unified approach as Image: model cards let you pick the generator (each with different strengths), task chips set the mode (Text→Video, Animate an image, Start & end frame), and settings live in the right panel.
Video lived inside a sub-menu, two clicks deep. The interface was split between a left settings panel and a black canvas with collapse/expand arrows.
Video is now a top-level destination. Model cards show each model's specialty (Refs · audio, Elements, Start · end frame, Cinematic) so you can make an informed choice. The prompt lives at the bottom — describe what you want and hit Generate.
Generated videos render inline with a full playback control bar. You can rate the output, download it, or immediately feed it into another prompt to iterate — all without leaving the page.
Audio & Video → Video Generation (2 clicks deep)
Left rail → Video (1 click from anywhere)
Model dropdown: Veo 3.1 Fast
Model cards: Seedance 2.0, Kling V3 4K, Veo 3.1, Runway Gen 4.5 — with labelled strengths
Reference Images 0/3 upload area
+ Reference files button; task chips set the mode automatically
No. of Generations 1–4, Duration 4/6/8s, Quality 720P–4K, Dimension 16:9/9:16
All identical in the right panel — same values, same options
Magic Prompt, Keywords
Magic Prompt in prompt bar; Keywords under Advanced
Last updated