What Is Visuall and What Can It Do?

Visuall is a browser-based AI video and image generator built around a single canvas. Instead of writing one prompt and hoping for a finished clip, you describe a film, approve the still frames you like, and then animate each scene with a video model. It is aimed at creators who want shot-level control over AI video rather than a one-shot text-to-video box. You can start free with 35 credits and no credit card.

The core idea: approve the frame, then film it

Most AI video tools collapse everything into one generation step. Visuall splits it in two:

1. Keyframes first. A text-to-image model (the site shows FLUX.2 and FLUX.3 in its examples) renders still frames for each scene.
2. Animate second. Once you approve a frame, a video model turns it into motion.

The practical benefit is that you catch a bad composition while it is still a cheap image, not after you have paid for a video render. The site's own framing is "Approve the frame. Then film it."

Build it box by box

Visuall describes every video as a set of visible, editable steps on the canvas:

• Story — your description of the film.
• Scenes — the AI scene plan, broken into shots with durations (the example shows a 10-second film split into 3 scenes of 4s, 3s, and 3s).
• Keyframes — the approved stills, with model and aspect ratio noted.
• Animate — the video model applied per shot.
• Final video — the stitched output.

Because each step is a separate box, you can change one shot and keep the rest. The site states that only what you change is regenerated, and the rest is reused for free. That is the main structural difference from a single-prompt generator.

Models you can choose from

The site names several models across its examples and description:

Model availability and per-step costs are listed on the pricing page under "See what each step costs," so check there before committing credits to a long project.

A worked example from the site

Visuall's own behind-the-scenes demo is a 32-second "vlog from Rome in 44 BC" made by a person who does not exist. The described workflow:

1. Describe her once — one paragraph about a character named Mia; FLUX.3 draws a consistent face.
2. One line per scene — four locations in Rome, one per line; Edit Image places the same face in each.
3. Write what she says — Gemini Omni Flash turns each frame into an 8-second talking clip with sound.
4. Stitch the episode — Stitch joins the clips into one 32-second video with her voice.

That run used 11 boxes, 225 credits, and about two minutes of waiting. It is a useful reference point for scoping: a short multi-scene piece with dialogue is a few hundred credits, not a handful.

What it is good for

Based on the formats shown on the site, Visuall fits:

• Product films — including ones built from as few as three phone photos.
• Vertical social clips — short reels shot vertical for the feed.
• Short narrative films — multi-scene pieces with consistent characters and narration.
• Campaign work and quick viral clips — the site positions one canvas as handling both.

How to start

1. Go to visuall.ai and enter the canvas.
2. Start from a template or describe your film to the canvas agent.
3. Review the generated scene plan and keyframes; approve the frames you want.
4. Animate approved frames with your chosen video model.
5. Re-run only the shots you changed, then stitch the final video.

You begin with 35 free credits and no credit card required, which is enough to test the keyframe-then-animate loop on a short piece before deciding whether to buy more.