How to add reaction text to a finished edit without changing the edit
Tell Vizard Agent the cut is finished and only the text is being added. It works from the approved edit rather than rebuilding it, finds the beats across the full runtime that are worth reacting to, and spaces the lines so each one lands on something that actually happened instead of arriving at a regular interval.
What is the short version?
Reaction text is what makes a long silent edit watchable, and adding it is usually treated as a reason to re-edit. It is not. The cut you approved stays exactly as it is and a text layer goes on top of it.
- Go to Vizard Agent with the finished cut.
- Say not to recut, reorder or shorten anything.
- Say roughly how many lines and what they should react to.
What do you need before you start?
The approved cut and a sense of density. A number is the most useful thing you can give — "fifteen to twenty across six minutes" sets a rhythm, where "add some reactions" produces either three or sixty depending on how the material reads.
- The finished cut. As approved.
- Roughly how many lines. A range is fine.
- What they react to. Wins, near misses, mistakes, progress.
- The tone. Short and blunt, or wry.
- Where they sit. Usually upper or lower third.
What do you type into Vizard Agent?
Give the count and the kinds of moment. Both matter: the count sets the pace, and naming the moment types stops every line landing on the same sort of beat, which is what makes reaction text feel automatic.
Prompt
Variants worth knowing:
- "Do not recut or reorder." Protects the approved cut.
- "Fifteen to twenty." Sets the density.
- "Risky, successful, progress." Varies what they land on.
What does Vizard Agent actually do?
Here is the order Vizard Agent worked in on a real long gameplay edit that needed reactions added after approval. Notice that finding the beats is a pass over the finished cut rather than over the original footage.
- Takes the approved cut as the fixed input.
- Locates the progression beats across its full runtime.
- Identifies the moments worth reacting to — wins, near misses, mistakes.
- Spreads the lines so they are not clustered in one section.
- Writes short reactions matched to what each moment is.
- Places each line at the beat rather than at an interval.
- Checks the text against the picture at each placement.
- Confirms nothing else moved — cuts, order, length, music.
- Renders the text layer over the approved cut.
- Reviews the finished video for readability and spacing.
Step two is why this works from the cut rather than the source. The beats that matter in the finished edit are the ones that survived it, and going back to the original footage finds moments that were deliberately cut out.
Step seven is worth asking for by name. A line that reads perfectly in a list can sit across a scoreboard, a face or a subtitle once it is on the picture, so each placement is checked against the frame it actually lands on rather than against the timecode alone.
Step four is what stops it feeling lumpy. Left to the material, reaction moments cluster wherever the action is densest, and a six-minute video with fourteen lines in its first two minutes reads as a video that ran out of ideas.
What does the result look like?
The same cut you approved, at exactly the same length, with short lines appearing over it at the moments that earn them — spread by Vizard Agent across the whole runtime rather than bunched into the opening minutes, and each one written about the thing that has just happened on screen.
Nothing in the edit moved. Only the text is new.
When does this not work well?
Reaction text needs actual things to react to. A long, steady, uneventful stretch will either get lines from Vizard Agent that comment on nothing in particular or get none at all, and both of those are conspicuous when the density everywhere else is high.
- Long uneventful stretches. Nothing to react to.
- Commentary already present. The text competes with it.
- Very fast cutting. No room for a line to be read.
- Dense captions already. Two text layers fight.
- A cut that needs re-editing. Text will not fix pacing.
How do you fix a result that came back wrong?
Point at the particular line and say what is wrong with it. Vizard Agent keeps the beat list it built and every placement it chose, so changing the wording, moving a single line or thinning out a crowded section is an edit to the text layer alone and touches nothing underneath it.
- "Too many in the first half." Redistributed across the runtime.
- "That one is on nothing." Moved to the nearest real beat.
- "The tone is off." Rewritten without moving the placements.
- "It covers the action." Repositioned in the frame.
How does Vizard Agent compare to doing it yourself?
By hand this means scrubbing the finished video looking for moments, which is slow and tends to find the loud ones. The count is also rarely planned, so the density drifts — heavy where the editor was paying attention and thin where they were tiring.
| By hand | Vizard Agent | |
|---|---|---|
| The cut | Often re-opened | Left exactly as approved |
| Finding the beats | Scrubbing | A pass over the finished cut |
| Density | Drifts | Planned to a count |
| What they land on | The loud moments | Several kinds of beat |
| Checking | Played through | Text checked against each frame |
Common questions
Will the edit change? No. Vizard Agent adds a layer over the approved cut.
How many lines should I ask for? Give a range. Vizard Agent spreads them across the runtime.
Can I write them myself? Yes. Vizard Agent will place them at the right beats.
Where do they sit in frame? Wherever you say. Vizard Agent checks they do not cover the action.
Can it match a style I like? Yes. Send an example and Vizard Agent matches the treatment.
Will the music change? No. Vizard Agent leaves the audio untouched.
What if a section has nothing? Vizard Agent leaves it alone rather than inventing a reaction.
Can it use emoji? Yes, if that is the tone you want Vizard Agent to write in.
Will they overlap the captions? No. Vizard Agent checks both layers against each other.
Can I approve the wording first? Yes. Ask Vizard Agent for the list before it renders.
Does the length change? No. The text sits inside the existing runtime.
Can it add sound effects too? Yes. Vizard Agent can place them on the same beats.
What about a second pass later? Yes. For Vizard Agent the approved cut stays the input.
Why work from the cut rather than the footage? Because the original source still contains every moment you already decided to take out, and writing a reaction to one of those quietly argues for putting it back in.