Storyboard a Multi-Scene AI Video Ad Without Reshoots

To storyboard a multi-scene AI video ad without reshoots on Flovaly, build the whole sequence as one Workflows graph: one Talking Actor node per scene, each fed its own script line, all of them running against the same automatically-resolved character. Hit Run once and every scene generates together in a single job — no reselecting or re-uploading a face between scenes. What Workflows doesn't yet do is stitch those finished clips into one continuous file for you; that part still means downloading each scene and cutting them together yourself.

Why this is normally a manual, repetitive job

A multi-scene ad — a hook, a value-proposition beat, a call to action, say — is really three or more separate generations. Done one at a time in the Studio, that means opening a fresh generation, picking your character again, writing that scene's line, running it, then repeating the whole process for the next scene. Nothing about that is hard, but it's three or four trips through the same setup for what is conceptually one ad. Workflows exists for exactly this repeated-pipeline case.

Build the sequence once: one Talking Actor node per scene

Open Workflows from the sidebar, start a new workflow, and add one Talking Actor node for every beat in your ad — three nodes for a hook / value / CTA structure, more if your script has more scenes. Add a Text Input node next to each one and wire its output into that scene's Scripthandle, with that scene's specific line. Each Talking Actor node is independent in the graph, so there's no limit on how many scenes you lay out before running.

You don't need to attach a character to any of them. Every Talking Actor node currently resolves its face automatically — your most recently created saved character, or your account's first premade character if you haven't made one yet — the same reused-reference mechanism our character consistency guide covers for the Studio. The canvas has no per-node character picker yet, which in practice means it does the consistency work for you: every Talking Actor node in a workflow you build today lands on the same face without you doing anything extra, and there's currently no way to deliberately mix two different saved characters across scenes in one graph.

What stays consistent between scenes, and what doesn't

The face and voice base stay fixed scene to scene. The camera framing does too — every Talking Actor node renders the same upper-body, speaking-to-camera setup in a vertical 9:16 frame, regardless of what each scene's script says. There's no separate scene-setting or background prompt field on this node type; the only thing that changes clip to clip is the script text and the lip-synced delivery of it. That's a real constraint worth knowing before you plan a storyboard around different locations or camera angles per scene — today, a Talking Actor-only storyboard is a sequence of spoken beats on one consistent setup, not a sequence of different-looking shots.

Adding a scene that isn't a talking beat

For a cutaway that doesn't need your character's face — a product close-up, an establishing shot, a text-on-screen beat — add a plain Videonode instead and give it its own prompt. It renders independently from a text description and doesn't carry any character reference: the Video node type only exposes a single prompt input in the current canvas, so there's no face-consistency mechanism to lean on here the way there is for Talking Actor scenes. Use it for shots where showing a specific, recurring face isn't the point.

Running it, and the part that's still manual

The action bar's Run button executes the whole graph in one submission — every node lights up as it processes, and all your finished scene clips land in the gallery tagged to that one workflow run, instead of you tracking down three or four separate Studio generations afterwards.

What it doesn't do yet is combine those clips into one finished ad file. The node picker's Tools category lists Resize, Concat, and Extract frame, but none of them run yet — each one currently stamps a "Coming soon" result rather than actually processing anything. Today, finishing the ad means downloading your scene clips from the gallery in order and cutting them together in whatever editor you already use. Building the scenes with consistent, correctly-ordered output at least removes the reshoot problem; final assembly is still a manual step for now.

What a multi-scene workflow actually costs

Talking Actor nodes are a flat 8 credits each, regardless of script length. A three-scene hook / value / CTA ad is 24 credits, about $4.64 on the $29 Starter plan (150 credits, roughly $0.193 a credit). Add a Video-node cutaway at its default length and you're paying 1 credit per second of rendered output on top — 5 credits for a 5-second cutaway, bringing a four-scene storyboard to 29 credits, about $5.61. Upload and Text Input nodes cost nothing; only the nodes that actually generate media are billed. The full breakdown of every generation type across Flovaly is in our AI video ad cost guide.

Where this fits next to batch-testing and script writing

This is a different axis from batch-testing ad hooks, which varies one opening line across many separate clips to find a winner. Storyboarding in Workflows instead fixes several distinct lines into one ordered sequence you build once and rerun whenever the script changes. If you haven't written the scenes yet, our guide to UGC ad scripts that convert covers the hook-problem-proof-CTA structure that maps cleanly onto a three- or four-node storyboard.

Start with the Workflows canvas overview or building your first workflow if you haven't opened the canvas before, or go straight to the $1 trial and lay out your first multi-scene sequence.

FAQ