Generation rarely lands perfectly on the first try. A banner sits on a face, a line runs half a second long, the source is horizontal. The usual answer is "regenerate and pay again." Here you edit by hand, and re-assembly costs nothing.

Two studios on top of the result. Decorate lays down graphics (banners, titles, subtitles); Montage mixes clips and sound fragments on one timeline. Both open from the node and edit selected material without touching or re-paying for generation.

Decorate: the AI’s plan becomes yours

The Director proposes a decoration plan — where and when banners, titles, subtitles, and the outro appear. In the studio that plan stops being the AI’s plan and becomes the project’s document: you edit the text, the timing ("Start" and "Show"), the on-screen position, the font size, the gradient. Drifted off-brand? The "restore to the design system" button puts it back the way it was.

vinistudio.app
The Decorate studio: banners/titles/subtitles on a timeline over the video, a real-time preview, and an inspector on the right — text, timing, position, font size, gradient.
The Decorate studio: banners/titles/subtitles on a timeline over the video, a real-time preview, and an inspector on the right — text, timing, position, font size, gradient.
The preview can’t drift from the render — not because we carefully keep two pieces of code in sync, but because there is only one: what you see in the studio and what ships to the file are the same composition.

Montage: video and sound on one timeline

Video fragments share one timeline: each has its own in/out points, 0.25×–4× speed, fades, and framing. Sound fragments are placed by hand in time, while each video’s native audio can be kept, lowered, or muted. And the key thing — crop to 9:16, 1:1 or 16:9: a horizontal source becomes vertical — or the other way round.

vinistudio.app
A production screen from Montage with one fragment selected. Its trim, speed, fades, native audio, and cropping are visible on the right. In the current version, video and sound fragments share the same timeline.
A production screen from Montage with one fragment selected. Its trim, speed, fades, native audio, and cropping are visible on the right. In the current version, video and sound fragments share the same timeline.
Regenerate

Not right? Hit Run again — the whole thing, at the full price of generation. Even just to nudge the banner over by a centimeter.

Edit by hand

You move the banner, trim the clip, crop to the format you need — and re-assemble. Render without AI, $0: not a single paid model call.

The timeline is sacred: the ban is addressed to the machine. LLMs and assemblers don’t touch it — a human edits it, explicitly and visibly.— the studios’ principle
$0re-assembly after an edit — render without AI
0.25–4×clip speed in Montage
3 formatscrop to 9:16, 1:1 or 16:9 — either direction
video + soundmanual timing on one timeline

How it works — step by step

  1. The AI proposes. The Director sets the decoration plan and the generator produces the clip — as a starting point, not a verdict.
  2. You open a studio from the node. Decorate for graphics, Montage for the clip. A real-time preview in the video’s own format.
  3. You edit by hand. Text, timing, position, font size (Decorate); clip order, sound, trim, speed, fades, and cropping (Montage).
  4. You re-assemble for $0. The render runs without AI — not a single paid generation; the finished clip returns to the project.

Build your own hero in Vini Studio

A cloud AI video generator: from an idea in words to a finished clip with sound, 9:16 or 16:9. The project is in closed beta — I am looking for authors and partners.

Get started in the studio →Sign in with Telegram or Google. No account needed in advance — it is created for you.