How to hit an exact runtime to the second
Give Vizard Agent the source and say the exact runtime you need. It treats the number as a constraint rather than a target — selecting the combination of moments that fills it, opening on the strongest hook, and checking the original for burnt-in captions before it reframes anything.
What is the short version?
"About a minute" and "exactly sixty seconds" are different jobs. The second one is a packing problem: the edit has to be built out of moments whose lengths add up, which is a different decision from picking the best moments and seeing what you get.
- Go to Vizard Agent and give it the source video or link.
- Say the exact runtime — "exactly 60 seconds".
- Say what the opening has to do.
What do you need before you start?
A source with more material in it than the target length, and a reason for the number. Vizard Agent handles the arithmetic on its own, but knowing why it has to be sixty seconds changes which idea gets sacrificed when something inevitably has to go.
- The source. A file or a link.
- The exact length. In seconds.
- Why that length. A platform cap, a slot, a spec.
- The opening. Strongest hook, or start at the start.
- The frame. Vertical, square or wide.
What do you type into Vizard Agent?
Put the number in capitals if it helps you feel better, but the important thing is the word "exactly". Without it, a runtime reads as guidance, and Vizard Agent will deliver a very good fifty-four-second cut that is no use to you at all.
Prompt
Variants worth knowing:
- "Exactly." Turns a target into a constraint.
- "Start on the hook." Changes what fills the first three seconds.
- "Within 59–60." A tolerance, if you have one.
What does Vizard Agent actually do?
Here is the order Vizard Agent worked in on a real sixty-second cut built from a long recorded talk. Step nine is the check that saves the reframe, and it deliberately happens before any cropping decision has been made rather than after the crop is already rendered.
- Downloads the source and probes its properties.
- Transcribes it with word-level timings.
- Loads the repurposing and talking-head playbooks.
- Extracts sample frames and inspects the framing.
- Lists the first segments with exact timestamps.
- Analyses the key segment points across the recording.
- Extracts frames across the first minute and looks at the shots.
- Tests a vertical crop and checks the composition.
- Checks for burnt-in captions across the whole timeline.
- Inspects the captions in the intro frames.
- Samples vertical crops of every candidate segment.
- Combines the candidate previews into one image and reviews them.
- Compares the exact transcripts of the two strongest candidates.
- Measures the speech loudness and pulls word timings.
- Checks whether music is already in the audio before adding any.
- Searches, downloads and measures a music bed.
- Generates animated captions and tests two styles on a frame.
- Builds a headline badge and tests the full composite.
- Writes a dynamic framing script and renders the minute.
- Measures the loudness and reviews frames across the whole edit.
- Re-mixes with proper normalisation and renders the master.
Step nine is the one worth stealing. Reframing a wide video to vertical crops the sides, and a source that already has captions burnt into the bottom of the frame will end up with them half cut off — which looks like a bug and is actually just an unasked question.
Step thirteen is how the packing problem gets settled. When two segments are both good and only one fits, the decision comes from comparing what they actually say rather than which one had more energy.
What does the result look like?
A cut that is the length you asked for, not close to it. It opens on the moment that earns the first three seconds, the segments inside it are complete thoughts rather than trimmed-to-fit fragments, and the framing follows the speaker rather than centring the frame.
The exact number is met by choosing which moments go in, not by speeding anything up or trimming the last sentence short. That distinction is what stops an exact-length cut sounding rushed at the end.
When does this not work well?
An exact runtime is a real constraint, and constraints cost something. When the source cannot fill the number with complete ideas, something has to give — and it is better to know which than to have it decided quietly.
- A source barely longer than the target. Nothing to choose from.
- Very long individual points. One idea may not fit at all.
- Frame-exact specifications. Say so; seconds are not frames.
- Hard-cut music. A track that must land needs its own timing.
- A number with no slack. One extra word and something else goes.
How do you fix a result that came back wrong?
Say what should have been in it. Vizard Agent keeps the transcript, the candidate list, the loudness measurements and the framing, so swapping a segment re-solves the arithmetic around your choice rather than re-cutting from the source again.
- "Include the bit about X." Forced in; something else gives way.
- "It feels rushed at the end." Rebalanced across the same length.
- "Make it 45 instead." Re-solved at the new number.
- "Captions are clipped." Re-cropped clear of the burnt-in text.
How does Vizard Agent compare to doing it yourself?
By hand, exact lengths are the fiddliest kind of edit: you build the cut you want, find it is sixty-seven seconds, and then spend longer shaving seven seconds off it than you spent making it. The last seven seconds are always the expensive ones.
| By hand | Vizard Agent | |
|---|---|---|
| Hitting the number | Trim until it fits | Segments chosen to fill it |
| What gets cut | The end, usually | Whichever idea is weakest |
| Burnt-in captions | Found after reframing | Checked before cropping |
| Choosing between two good bits | A feeling | Transcripts compared |
| Changing the number | Start again | Re-solved from the same candidates |
Common questions
Will it be exactly that length? Yes. Vizard Agent treats the number as the constraint.
Can I give a range instead? Yes. Say "59 to 60" and Vizard Agent works inside it.
Does it speed things up to fit? No, unless you ask. Vizard Agent chooses different moments instead.
Can it hit a frame count? Say so explicitly. Vizard Agent works to frames if you need that.
What if my source is too short? Vizard Agent tells you rather than padding it out.
Will the last sentence be complete? Yes. Complete thoughts are what it packs with.
Can it do several exact lengths at once? Yes. Vizard Agent can deliver 30, 60 and 90 from the same source.
Does it work from a link? Yes. Vizard Agent downloads and probes it first.
What about the burnt-in captions? Checked before reframing, so they are not cropped in half.
Can it add its own captions? Yes. Vizard Agent generates them from the cut's own word timings.
Will the music end cleanly? Yes. Vizard Agent fits the bed to the exact runtime.
Can I pick the opening moment? Yes. Name it and Vizard Agent packs the rest around it.
Does the loudness get checked? Yes, on the master, before delivery.
Why not just trim to length? Because trimming takes it out of the ending, and the ending is the part people remember.