How to make a ranking video that scores each entry

Last updated 2026-08-23 · 6 min read

What is the short version?

A ranking video is six clips, six numbers and six scores, and the whole thing lives or dies on whether the numbers land on the beat and can be read at a glance. Both of those are checked against real frames, not assumed.

1. Go to Vizard Agent and upload your clips.
2. Say what is being ranked and scored.
3. Say the length, the shape and how hard the edit should hit.

What do you need before you start?

Your clips and the premise. Vizard Agent reviews the footage, researches the subject and designs the overlays itself, so what has to come from you is what the ranking is actually about, because that decides both the order and the tone of the scores.

• The clips. Six is a good number for thirty seconds.
• The premise. What is being ranked and on what basis.
• The scores. Or let Vizard Agent assign them from the research.
• The length. Thirty seconds is the working size.
• The energy. Aggressive, deadpan, tongue-in-cheek.

What do you type into Vizard Agent?

Say what is being ranked and how hard it should hit. Vizard Agent uses the premise to research the entries and the energy to pick the music, the cut rhythm and the overlay style, and those two together shape almost everything else.

Prompt

Variants worth knowing:

• A hook card. A "hear me out" opening earns the first three seconds.
• Your own scores. Give them and Vizard Agent uses yours.
• An ending card. A freeze with a closing line finishes it cleanly.

What does Vizard Agent actually do?

Here is the order Vizard Agent worked in on a real ranking edit. The last two steps are the ones worth copying: it watched the cut back and then went in and gave one entry more time because its score missed the drop.

1. Probes all six clips and extracts sample frames from each.
2. Researches the subject so the ranking has a basis beyond the footage.
3. Looks at three frames from every clip — opening, middle and later — to know what each one shows.
4. Measures the audio in each clip and searches for aggressive tracks.
5. Generates a beat at the tempo the edit needs and listens to both candidate tracks for their drops.
6. Generates the slam, whoosh and grind hits and checks the effects are long enough to use.
7. Designs the ranking overlays, then rebuilds them once the correct font paths are found.
8. Checks each overlay individually — the hook card, each rank number, each score, the ending card.
9. Renders the edit, reviews every score on the shot it lands over, checks the mix waveform, then gives one entry more time and punches the opening after watching the cut back.

Step 9 is the difference between a ranking edit that feels tight and one that feels nearly right. A score that appears just after the drop instead of on it reads as sloppy, and only watching it back at speed catches that.

What does the result look like?

From the run this page is written from, probed on the delivered file: 1080x1920, H.264, 24fps, 30.46 seconds, AAC audio. Vertical, thirty seconds, six entries with rank numbers and scores, a hook card, an ending freeze and a generated beat under it.

Thirty seconds for six entries is five seconds each, which is exactly enough. The format works because the viewer is always about to see the next number, and stretching it any further loses the momentum that makes it watchable.

When does this not work well?

A ranking is an opinion presented as a measurement, and the format's charm and its risk are the same thing. Vizard Agent researches the subject to give the order a basis, and the argument is still yours to defend.

• Scores invite argument. That is often the point, and it is worth deciding whether you want it.
• Research is not authority. Vizard Agent grounds the ranking; it does not make it correct.
• Six entries needs six good clips. A weak entry drags the whole edit.
• Small text on a busy shot vanishes. Vizard Agent checks each overlay on its own shot for this reason.
• The pace is unforgiving. Five seconds per entry leaves no room for a slow clip.

How do you fix a result that came back wrong?

Name the entry. Vizard Agent keeps the frame reviews, the beat grid, every overlay and the mix measurements, so reordering the ranking or changing a score is a re-render against work it already did rather than a rebuild.

• "Swap three and four." Reordered, with the overlays re-rendered to match.
• "That score is too generous." Changed and checked on its shot.
• "Number one needs more time." Re-timed, the way this run fixed entry three.

How does Vizard Agent compare to doing it yourself?

By hand this is trimming six clips to five seconds each, building twelve overlays in a design tool, nudging them until they roughly hit the beat, and then finding two of your scores are unreadable against the footage.

Common questions

Should the best one come first or last? Last, almost always. The format works because the viewer is waiting for number one.

How many entries work best? Five or six for thirty seconds. More than that and each one gets too little time.

Does it decide the scores? It can, from the research, or you can supply them.

Will the numbers be readable? Vizard Agent checks every overlay on the shot it appears over rather than in isolation.

Does Vizard Agent research the subject? Yes. It looks the entries up so the ranking has a basis beyond whichever clip happens to look best.

Does it need music with a drop? It helps a great deal, and Vizard Agent generates a beat at the right tempo if nothing suits.

Can I use my own footage only? Yes. That is what this run did — six uploaded clips and nothing else.

Does this work for other subjects? Yes. The format transfers to anything comparable — cafés, guitars, cities, cars — because the method is the same: research the entries, look at every clip, and land the numbers on the beat.

Can it add sound effects? Yes — slams, whooshes and grinds on the transitions. Vizard Agent checks each one is long enough to use before building the edit around it.

Does Vizard Agent watch the finished cut? Yes. On this run it watched the render back for sync, pacing and mix, then went in and re-timed one entry that was landing off the beat.