How to assemble a video from a shot script and numbered files

Last updated 2026-08-16 · 6 min read

What is the short version?

A shot script and a folder of numbered files is how professional edits actually arrive, and the assembly is mechanical, tedious and easy to get wrong — particularly when there are two takes of shot seven and nobody wrote down which one was good.

1. Go to Vizard Agent and upload every clip and audio file.
2. Paste the shot script with its shot numbers and directions.
3. Say the final shape, the length and what music you want under it.

What do you need before you start?

The files and the script. Vizard Agent works out which file is which shot from the filenames, so the numbering does not have to be tidy — but the script does have to say what each shot is, because that is what decides between two takes of the same moment.

• All the clips. Numbered filenames are enough. Duplicates are fine and expected.
• The shot script. Shot numbers, what happens, and any lighting or framing notes.
• The dialogue, if any. Vizard Agent transcribes what is in the clips and works to the script's lines.
• The output shape. Vertical, wide, square. Say it once and everything is normalised to it.
• What you want added. Music, ambience, sound effects. These are generated rather than sourced from the clips.

What do you type into Vizard Agent?

Paste the script as it is. Vizard Agent handles the file matching itself and does not need a mapping table from you, so the useful extra sentence is the one telling it what to do when two files clearly cover the same shot.

Prompt

Variants worth knowing:

• Name your preference. "Prefer the take with dialogue" resolves duplicates without you watching them.
• Music brief in the script. Per-shot audio notes in the script are read and followed.
• Assemble only. Say so if you want a straight assembly with no generated music or effects.

What does Vizard Agent actually do?

Here is the order Vizard Agent worked in on a real script assembly. The first half of the run is inventory: it probes, samples and looks at everything you sent as one labelled sheet before it makes a single decision about the order or which take to use.

1. Lists the workspace and downloads every video and audio file.
2. Probes all their durations.
3. Extracts sample frames from every clip.
4. Composes one labelled mega contact sheet of all of them together.
5. Builds grouped sheets and reviews them in batches.
6. Compares the duplicate takes and gets transcript timings for the dialogue.
7. Analyses each clip's motion and beats.
8. Generates music and ambience, then checks the result and regenerates until the track suits the scene.
9. Measures the dialogue's loudness and trims the audio stems before normalising every clip to the output shape.

Step 4 is the one that makes the rest possible. Vizard Agent looked at every clip as a single labelled image, which is how it can hold an eleven-file folder in mind at once rather than opening them one at a time and forgetting the first.

What does the result look like?

From the run this page is written from, probed on the delivered file: 1080x1920, H.264, 24fps, 37.33 seconds, AAC audio. Vertical, under forty seconds, cut in the script's order, with dialogue trimmed, ambience and piano underneath, and every clip normalised to the same shape.

Ask for a wide version and Vizard Agent re-frames each shot rather than letterboxing the vertical cut.

The assembly matches the script's shot order exactly. Where the script asked for a slow move into a scene, that clip is held longer than the cutting rhythm around it, because the direction in the script is treated as an instruction rather than a description.

Expect tens of minutes rather than minutes. This category was not separately measured; comparable work runs a median of 28 to 38 minutes end to end, and across all projects the median cost by tier is Flash 47, Pro 55, Max 242, Ultra 263 credits.

When does this not work well?

The script is the contract, and Vizard Agent follows it closely. Where a script is ambiguous, or where the footage does not actually contain what the script describes, it has to choose — and its choice may not be the one you had in mind.

• A shot the script names but the files do not contain. Vizard Agent will say so rather than inventing it.
• Filenames that do not indicate order. Random names mean the matching is guesswork against the script's descriptions.
• Two takes with no stated preference. Vizard Agent picks the stronger one, and stronger is a judgement.
• Dialogue that differs from the script. The recording wins; say if you want the script's wording instead.
• A script written for a longer film. Forty seconds of footage will not carry a two-minute structure no matter how it is cut.

How do you fix a result that came back wrong?

Name the shot number. Vizard Agent keeps the contact sheets, the take comparisons, the transcript timings and the audio stems, so swapping a take or re-ordering two shots is a change to the assembly rather than a fresh pass over the files.

• "Use the other take of shot seven." A direct swap; nothing else moves.
• "Shot three should come before shot two." A reorder against the same stems.
• "The music is too loud under the dialogue." A mix change from the measurements it already took.

How does Vizard Agent compare to doing it yourself?

By hand this is importing eleven files, watching each one to work out what it is, renaming them, then assembling to the script. The assembly is the quick part. The identifying and the choosing between takes is where an afternoon goes.

Common questions

Do the filenames have to be tidy? No, as long as the number is in there somewhere. Vizard Agent reads it off the name and checks against the script.

What if I have two takes of a shot? Vizard Agent compares them and uses the better one. Say your preference if you have one.

Will it use the audio from my clips? Yes. It transcribes and trims the dialogue, and generates music and ambience around it.

Can it work from a script in any language? Yes. Vizard Agent reads the script as written and matches it to the footage.

Can I get another aspect ratio? Yes. Vizard Agent re-frames each shot for the new shape rather than adding bars.

What if a shot is missing? Vizard Agent tells you which script shot has no matching file instead of quietly skipping it.

Can it add music the script asks for? Yes. Per-shot audio notes are read as instructions, and Vizard Agent generates or sources what they describe.

How many files can it handle? A folder of a dozen is routine. The contact sheet is what keeps them all straight.