Vizard Agent

How to get a square version of clips you have already approved

Last updated 2026-09-04 · 7 min read

Ask Vizard Agent for the approved clips in another shape. It aligns each delivered cut against the original recording's audio to recover the exact in and out points you signed off, then re-cuts those moments from the full-resolution master — so the new version is a fresh cut, not an upscaled crop.

What is the short version?

Once a set of clips has been approved, a second aspect ratio looks like a five-minute job and is not. Cropping the delivered file loses resolution and framing; re-cutting from scratch risks landing on different moments than the ones everyone agreed to.

  1. Go to Vizard Agent and say which approved clips you mean.
  2. Name the shape you now need — square, wide, vertical.
  3. Say to cut it from the master, not from the delivery.

What do you need before you start?

The approved renders and the original recording. Vizard Agent uses the first to work out where the cuts are and the second to actually make them, which is why having both matters more than having a written edit list.

What do you type into Vizard Agent?

Make the source explicit. "Give me square versions" is ambiguous in a way that matters — it could reasonably mean cropping what you already have, and the difference between the two readings is the difference between a sharp deliverable and a soft one.

Prompt

Cut square (1:1) versions of the five approved clips. Take the picture from the 4K master at the same edit points, not from the delivered renders. Keep the approved audio.

Variants worth knowing:

What does Vizard Agent actually do?

Here is the order Vizard Agent worked in on a real set of approved clips re-cut to a new shape. Step four is the interesting one — it does not need your edit list, because the delivered files contain the answer.

  1. Checks what of the project is still on disk.
  2. Locates the project files and checks what is still in the cloud.
  3. Fetches the master and the approved renders.
  4. Recovers the exact approved edit points by aligning the audio.
  5. Reads the price table before committing to the renders.
  6. Cuts the new picture from the full-resolution master.
  7. Assembles each file with the approved audio.
  8. Checks the new framing on real frames.
  9. Checks lip sync against the approved renders.
  10. Compares the new frames with the approved versions side by side.
  11. Uploads the finished set.

Step four is the part worth understanding. The approved render carries the same audio as the master, just shorter — so lining the two waveforms up tells Vizard Agent precisely which second of the original each clip began and ended on, to the frame. No edit list, no notes, no memory of the original session required.

Step nine is the check that catches the failure this method can produce. An alignment that is off by a few frames looks fine on a contact sheet and wrong the moment somebody speaks, so the lip sync is verified against the version everybody already approved.

What does the result look like?

The same moments, in a new shape, cut from the original resolution. Because the picture comes from the master rather than from a delivered file, the square version is a genuine crop of a large frame instead of an enlargement of a small one, and there is room to reframe rather than just centre.

The audio is the approved mix, unchanged. That is deliberate: re-mixing is the fastest way to turn a format conversion into a new approval round.

When does this not work well?

The whole method depends on the delivered clip and the master sharing an audio track, because that shared track is what Vizard Agent measures against. Where the link is broken the edit points have to come from somewhere else, and usually that means you supply them by hand.

How do you fix a result that came back wrong?

Say which clip and what you see. Vizard Agent keeps the recovered edit points, the master and the approved reference renders, so a reframing or an alignment fix rebuilds one file without disturbing the other four or reopening the edit.

How does Vizard Agent compare to doing it yourself?

By hand, this depends entirely on whether you still have the project open. If you do, it is fiddly but fine. If the session is gone — and by the time someone asks for a square version, it usually is — you are scrubbing a long master trying to find six moments you approved a fortnight ago.

By hand Vizard Agent
Finding the approved points From memory or an old project Recovered by audio alignment
Picture source Whatever is to hand The full-resolution master
Quality Crop of a delivery Fresh cut of the original
Verifying it matches Play both and hope Lip sync checked against the approved cut
Doing five of them Five times the work One pass

Common questions

Do I need the original project? No. Vizard Agent recovers the edit points from the delivered files.

What if I only have the delivered clips? Then it crops those. Vizard Agent will tell you the quality cost.

Can it do more than one new ratio? Yes. Vizard Agent renders each shape from the same recovered points.

Will the framing be right? It reframes rather than centring. Vizard Agent measures where the subject sits.

Does the audio change? No, by default. Vizard Agent carries the approved mix across untouched.

How accurate is the alignment? Frame-accurate, and Vizard Agent verifies it against the approved render.

Can it keep the burnt-in captions? It re-burns them at the new size, which is better than scaling them.

What if my master is a different edit? Say so. Vizard Agent aligns against whatever file the clips came from.

Does this cost as much as the original job? Much less. Vizard Agent is re-cutting, not re-deciding.

Can it also make a vertical version? Yes. Any shape, from the same recovered points.

What about clips with no talking? Music and room tone align too. Silence is the hard case.

Will it re-choose better moments? Not unless you ask. The point is that the moments are already approved.

Can it deliver a wide master as well? Yes. Vizard Agent can render the full set of shapes in one pass.

Why not just crop what I have? Because a square crop of a vertical delivery throws away most of the frame, and there is a full-resolution original sitting right there.