How to sort a pile of detailing clips into matching pairs

Last updated 2026-08-30 · 7 min read

What is the short version?

A detailing job produces a dozen clips of the same car twice — once filthy, once finished. The value is in seeing each view before and after, which means every clip has to be matched to its partner and nothing else.

1. Go to Vizard Agent and upload every clip.
2. Give the pairing order using the numbers in your filenames.
3. Say you want pairs rather than a general montage.

What do you need before you start?

All the clips and a consistent naming scheme. Vizard Agent reads the frames to confirm what each clip actually shows, but your filenames carry the intended pairing and it treats them as the instruction rather than as a hint to be second-guessed.

• Every clip. Both states of each view.
• The filenames. Numbered so pairs are obvious.
• The order. Which view goes first.
• The framing. Whether shots must match exactly.
• The length. A few seconds per pair.

What do you type into Vizard Agent?

Spell the sequence out in full. Vizard Agent follows a stated pairing order exactly, and writing it out — dirty one, clean one, dirty two, clean two — removes every question about which of twelve near-identical clips belongs where.

Prompt

Variants worth knowing:

• Matched framing. Each pair cut at the same moment.
• A wipe between states. Rather than a hard cut.
• The worst view first. So the sequence improves.

What does Vizard Agent actually do?

Here is the order Vizard Agent worked in on a real set of twelve detailing clips. The awkward part is the middle: the filenames said front and rear, but working out left from right meant looking at the footage.

1. Checks the metadata of every uploaded clip, twice, to be certain.
2. Downloads the clips and prepares thumbnails to identify the shots.
3. Builds a thumbnail overview for analysis.
4. Reviews the thumbnails to identify the individual parts of the car.
5. Pulls frames from several points in each clip.
6. Builds preview sheets covering all twelve shots.
7. Reviews clips zero to five, then six to eleven.
8. Checks the orientation of the first two clips and confirms it visually.
9. Checks every clip's technical parameters, saves detailed frames for geometry, compares the rear-interior shots to establish left from right, and saves the first and last frame of each clip to determine which direction it faces.

Step 9's ending is the part that makes the pairs correct. A clip labelled "rear" could be either side of the car, and the only way to know is to compare frames — the alternative is a video where the left door's "after" is the right door's.

What does the result look like?

A sequence of before-and-after pairs, each view appearing twice in a row — the state it arrived in and the state it left in — in the order you specified, with the framing matched so the two halves of each pair line up.

The pairing is the whole product. A montage of clean-car shots proves nothing; the same view immediately before and after is the only arrangement that shows what the work actually did.

When does this not work well?

Pairing depends on both states having been filmed the same way, and detailing footage is almost always shot at speed between jobs. These are the cases where Vizard Agent finds a pair that does not actually match, and says so rather than assembling it anyway.

• Unmatched framing breaks the pair. A different angle is a different shot.
• Ambiguous filenames cost a round. Front, rear and side are not enough on their own.
• Missing partners cannot be paired. A view filmed once has nothing to match.
• Different lighting exaggerates. A darker "before" flatters the "after" dishonestly.
• Very short clips leave nothing to cut. Two seconds is barely a beat.

How do you fix a result that came back wrong?

Name the pair that came out wrong and say which two clips should have gone together instead. Vizard Agent keeps all the thumbnails, per-clip frames, orientation checks and direction analysis, so a pairing is corrected without reviewing everything.

• "Those two are different doors." Re-paired from the direction analysis.
• "The after shot is upside down." Corrected from the orientation check.
• "Start with the interior." Re-ordered from the same pairs.

How does Vizard Agent compare to doing it yourself?

By hand this means opening twelve near-identical clips of the same car and working out which is which, which is genuinely difficult when six of them are of a dirty vehicle from angles you filmed in a hurry.

Common questions

How should I name the files? Consistently, with a number per view and a word for the state. Vizard Agent reads that as the pairing.

What if I filmed one view twice from different angles? Then the pair does not match. Vizard Agent will flag it rather than pairing them anyway.

Does the framing need to be identical? Close is enough for Vizard Agent, and closer is better. A very different angle reads as a different subject.

Can it tell left from right? It works it out from the frames, since filenames rarely say. That is what the comparison is for.

How long should each pair be? A couple of seconds each side. Long enough to register, short enough to get through twelve.

Why not just show all the after shots? Because a clean car proves nothing on its own. The pair is the evidence, and without it you are asking people to take your word for the difference.

Can it add a wipe between the states? Yes. Vizard Agent uses a transition rather than a hard cut if you prefer.

Does this work for other trades? Yes. Vizard Agent handles any job with a before and an after state of the same view.

Should the worst view go first? Usually. Vizard Agent orders it so the sequence has somewhere to improve toward.

Should the finished video be one file or several? One usually works better. Vizard Agent assembles the pairs into a single sequence, because the cumulative effect of six views improving is the argument that a single pair cannot make alone.

How many pairs hold attention? Five or six. Vizard Agent orders them so the most dramatic view lands late, because a sequence of twelve near-identical improvements stops registering around the eighth.

Can it label each pair on screen? Yes. Vizard Agent adds the view's name from your filenames, which helps a viewer follow what they are looking at when six near-identical panels go past in half a minute.

Does Vizard Agent check the pairs? Yes. It confirms each pairing against the frames before assembling.