friends

Best-Friends AI Rap Video Prompt

The best-friends variant turns the standard duo render into a bromance hype video: matching jackets, a high-five on the drop, shoulders bumping on the beat, and bars that roast each other just enough to be funny. It's the template for reunions, birthdays, fantasy-league wins, and "we've been friends since" posts.

When to use this template

Use it when the two performers are friends and the clip should feel like an inside joke the whole feed is allowed to hear. The core duo template is neutral; this one dials energy up and warmth sideways — bigger gestures, synchronized moments, playful competition between the two performers.

Hook & verse structure

Friendship bars work best as roast-and-toast: the left performer opens with a light jab ("still late to everything"), the right performer fires back and then both land a togetherness line. Write it with the lyrics template using tone playful — one jab each, one shared punchline, hooks that name the friendship's milestone ("fifteen years, zero arrests" energy).

Copy-ready main prompt

Create an AI rap video duo performance with best-friends energy from two separate portrait references.
Left performer = photo 1, right performer = photo 2, fixed sides for the whole clip.
Stage: original booth-style set, warm orange key light, one hanging microphone between them, dark backdrop with soft rim light.
Wardrobe: matching varsity-style jackets over each performer's own look, same color scheme on both.
Action: high-energy best-friend performance — traded bars with sharp mouth movement, synchronized head nods on the beat, one high-five on the beat drop, shoulders bumping once mid-clip, big grins throughout.
Choreography rule: the high-five happens exactly once, at the midpoint; hands meet in the center of frame and return to sides immediately.
Camera: locked medium two-shot, subtle push-in during the final two seconds.
Audio: original playful rap cadence, roast-and-toast call-and-response — write NEW original lyrics; never quote commercial songs.
Duration: 8-12 seconds, ending on a shared confident pose for the loop hold.
Template focus: Best-Friends AI Rap Video Prompt.

Beat & camera breakdown

1. 0-2s — Hype open. Both performers already nodding, grins on. Left spits the opening jab.
2. 2-5s — Jab and counter. Right performer's answer lands with a shoulder bump — the friendship's physical signature.
3. 5-7s — The high-five. On the beat drop, hands meet center frame. One take, one clap, back to position. More than one reads as chaos.
4. 7-12s — Shared toast. Both performers lean into the mic for the final together-line, then hit the matching pose for the loop hold.

Negative prompt

identity swap, merged faces, third person appearing, extra arms during high-five, deformed hands, extra fingers, blurry faces, aggressive expressions, weapons, alcohol, on-screen lyrics or captions, watermarks, brand logos, celebrity lookalike, NSFW, scene changes, camera shake

Photo prep for this template

• Photos where each friend is already smiling — the model amplifies the baseline expression; neutral inputs read as stern in a hype clip.
• Similar heights read best in the two-shot; if heights differ a lot, say "both performers standing at similar depth" to keep framing stable.
• The high-five needs visible hands: crop both photos to include arms where possible (see the prep checklist).

Render steps (any model)

1. Upload both references in slot order and paste the main prompt; add the roast-and-toast script lines from the lyrics template into the Audio block.
2. Render once with the full choreography. If the high-five produces extra arms, cut it from the prompt (keep just shoulder bumps) before touching anything else — handshake-class motion is the most failure-prone gesture.
3. Check the midpoint frame: one clean clap, four hands total. Then check the ending pose matches the opening for the loop.

Common failure modes

• Extra arms during the high-five → the gesture fired twice; enforce "exactly once, at the midpoint" or drop to shoulder bumps.
• Reads aggressive instead of playful → the jab lines were too sharp or the faces too neutral; soften the script and pick smiling inputs.
• Matching jackets invented differently → "matching varsity-style jackets, same color scheme" must name the color; vague matching drifts per performer.

FAQ

Can I do a trio of friends?

Render two pairs and cut them — three simultaneous identities is beyond what most image-to-video models hold stable for eight seconds.

The roast lines feel risky. How sharp is too sharp?

Roast the situation, never the person: "always losing his keys" lands, anything about appearance or relationships doesn't. You know your friend; render what they'd repost about themselves.