AI video generation

The b-roll you wish you'd filmed: an establishing shot, a mood-setting cutaway, a visual for a concept nobody can point a camera at. Inwista generates short video clips from a text description, right inside the editor — open the Videos panel in the left sidebar and switch to the AI generate tab.

Describing the clip

Generation runs on Veo 3.1, a quality-first video model — fixed engine, no choices to agonize over. Your prompt does the heavy lifting: describe the shot like a director would, with camera movement, subject, atmosphere and light. "A cinematic drone shot flying over a misty pine forest at sunrise" gives the model a shot list; "a forest" gives it a shrug.

Four settings frame the result: Aspect ratio (Landscape 16:9 or Portrait 9:16 — match your project), Resolution (720p or 1080p), Duration (4, 6 or 8 seconds), and Audio — Veo can generate matching sound for the scene, or deliver the clip silent if you're scoring it yourself.

Cost and generation time

Video is the most credit-intensive generator in Inwista — the exact price is shown on the button before you commit, and it scales with the duration you pick: an 8-second clip costs twice a 4-second one. Choose the shortest duration that serves the edit; b-roll cutaways rarely need more than a few seconds.

Generation isn't instant — producing video takes noticeably longer than images, so expect a wait while the clip renders. The panel keeps you posted on progress.

Review before it costs you twice

The finished clip plays back in a review modal before anything touches your project: watch it, then either Add to project or Re-generate — the re-generate button shows its credit price right on it, so there's never a surprise. As with images, a sharper prompt beats repeated re-rolls: if the motion or framing is wrong, say so in the prompt rather than spinning again with the same words.

Into the project

Add to project places the clip on your timeline at the playhead position — cue up the exact moment first, and the cutaway starts precisely there. From then on it behaves like any other video item: trim it with the handles, split it at the playhead, layer it on its own track over your main footage (see Working with the timeline).

A note on privacy

Your prompt and the generated clip are processed on servers solely to create your video — they are never used to train AI models. What you generate is yours, stored with your workspace media like anything else you upload.