How to make a great AI music video from your song

Open the AI music video generator and upload your MP3 or WAV, or paste a Suno or Spotify link. All four videos in this tutorial started from a Suno link: the song was finished before the video began, and Revid uses that recording as the soundtrack. Pick the format for where it will be posted: the four examples are 9:16 for TikTok, Reels and Shorts; choose 16:9 for a YouTube music video.

Then use Trim audio to choose the part the video covers. Top Floor Itch is a 3 minute 10 second song cut to the 70 seconds from 0:20 to 1:30, the hook and one verse, which is what a vertical video needs. The other three run the whole track, from 3:14 to 3:21.

Sync with lyrics or beats is the setting that shapes the edit. With Lyrics, Revid transcribes the vocal and changes scenes on the lines; with Beats, the cuts follow the rhythm and no lyrics are produced. All four examples sync to the lyrics.

• Singer lip sync: three of the four turn it on. Little by Little sets the slider to 35%, so about a third of the video shows its singer performing the vocal and the rest tells the story around her. Lip sync needs Lyrics sync and AI video.
• Lyric captions: on for Little by Little and See Me, Ivy, off for the other two. Turn them on when the words carry the story or the song is new to the listener; leave them off when the picture should do the talking.
• Sound wave: on for Top Floor Itch only, as a strip along the bottom of the frame, off for the other three.

The style picker sets the medium for every scene. The four examples use four different styles, and none of them mixes two:

The most viewed of the four, Er passero ferito, was made on Base, the cheapest AI video tier, in a style that forgives small motion glitches and suits a folk song. A clear style that fits the song does more than a more expensive model on a vague idea.

The music tool opens a field called Visual direction for the whole video. It is optional, and it is where three of these four videos were made: Revid uses it as the brief for every scene. Er passero ferito had none, and Revid wrote its scenes from the song alone, which works when the lyrics tell a simple story.

Top Floor Itch plans the video part by part, with a running gag that comes back each time the hook does:

Luxury rap parody with hyper-detailed CGI meerkats, dark neon visuals,
cinematic lighting and absurd animal humor.

OPEN          A giant desert penthouse rises from red dunes.
INTRO         The rapper meerkat stands on a rooftop in chains, a fur coat
              and sunglasses, overlooking his kingdom.
VERSE 1       Infinity pools, tiny bodyguards, a slow-motion mansion strut.
PRE-CHORUS    Neon hallways, intense close-ups. He keeps trying to scratch
              an itch he cannot reach.
CHORUS        Meerkats hanging from balconies, gold coins tossed to the crowd.
RUNNING GAG   Every time the hook comes back: he freezes, everyone tries
              to help, nothing works.
FINAL CHORUS  Fireworks, confetti, and he still cannot scratch it.
OUTRO         At sunset he finally starts scratching. Cut to black.

Little by Little does the opposite: one continuous story across the whole song, and a precise list of what must not appear. Saying what to avoid is as useful as saying what to show:

Do not make this lyric-by-lyric. Tell one continuous city story.

No microphone anywhere: no stage, no studio, no concert scene.
Only LUMA sings. One consistent face. The love interest never sings.
Background people can walk, eat, laugh or dance lightly, but never lip-sync.

Jakarta: a street cafe, plastic tables, scooters, glass towers,
sudden tropical rain, neon on wet pavement, the night city glow.
Mood: catchy, fun, city-romantic. Not sleepy, not dark heartbreak.

Avoid stickers, random text, chaotic edits, distorted faces and hands,
and over-literal lyric visuals.

See Me, Ivy writes its cast like a casting sheet, so the family looks the same in every scene: Ivy, nine, blonde; Mum, 36, blonde; Dad, 40, brown hair; an older brother of 13 and a younger one of 7. Then the setting (a large living room, Ivy making animal masks out of card and glue while everyone ignores her) and one line for the tone: fun, with a dark edge of reality.

• Base: 15 credits per 5 seconds, 540 for a 3 minute song. Er passero ferito, the most viewed of the four, was made on Base.
• Pro: 50 credits per 5 seconds, 1,800 for a 3 minute song. Sharper motion and detail; Little by Little uses it.
• Ultra: 100 credits per 5 seconds, 3,600 for a 3 minute song. The best model available, for hero videos; Top Floor Itch and See Me, Ivy use it.

On the Growth plan ($39 a month for 2,000 credits), that is three full songs on Base, one 3 minute song on Pro, or a 70 second Ultra cut like Top Floor Itch (1,400 credits) with room for a second try.

The video opens in the Revid editor. Watch it with the track: replace a scene that misses the line it plays under, move a cut that interrupts a phrase, and correct names or words in the captions. Keep text away from the edges for phones. Check the last shot as carefully as the first: a held final frame suits a sustained note better than an abrupt cut.

Export it as an MP4, or publish and schedule it to TikTok, YouTube and Instagram from Revid. For a 16:9 YouTube video, cut a vertical teaser of the chorus for Shorts and TikTok.