Make an AI rap duo with someone you know. Add one photo for the lead and one for their partner, keep the familiar routine or choose an original verse, and download the finished clip.
Two portraits · one performance
Moves
Sound
Add two photos and tick the permission box to generate.
Sample clip for illustration — not generated on this site. Muted so it can play on its own.
These two photos are the inputs for the example video above.Two portraits · one performance
An AI rap duo brings two subjects from separate pictures into the same performance. You do not need a joint photoshoot, a green screen, or footage of either person dancing. The first photo casts the person on the left; the second casts the person on the right. The default setup uses a shared microphone and an orange studio, with one lead delivering the vocal and the partner responding with gestures. Watch the reference above before uploading: it shows the kind of movement you are asking the model to recreate, rather than a guaranteed result for your own faces.
Concept illustration · not a generated video result
Bring two friends into the same frame without coordinating a recording session. This preset keeps the orange stage and reference-track routine. Start with equally clear portraits; matching image quality matters more than matching backgrounds. The illustration suggests a scene, not a tested output.
Concept illustration · not a generated video result
Create a playful shared performance for a partner using two individual portraits. The preset preserves the lead-and-partner setup instead of inventing an anniversary song. For a personal topic, switch to original rap only when your chosen subjects meet the model requirements.
Concept illustration · not a generated video result
This preset switches to original rap and fills in a birthday topic, with both performers trading lines. Use fictional or stylized subjects accepted by Seedance. Review the generated words and faces before sharing; an AI-written verse can need another attempt to suit the occasion.
Upload two separate JPG, PNG, or WebP images, each up to 10 MB. Use one clearly visible subject per picture. Put the lead in the left slot and the supporting performer in the right slot. Their photos can come from different places.
Try this step in the generatorUpload your photo references
Enlarge interface screenshotLeave the orange stage and built-in motion selected to recreate the reference. Open Video settings to review the model, length, quality, and framing. An original-rap model also lets you choose a topic and whether both performers trade lines.
Try this step in the generatorChoose the video settings
Enlarge interface screenshotSign in, confirm permission to use your inputs, and check the credit total on the button. The preview shows progress while the model works. When the result is ready, play it, turn sound on, and save the MP4. My videos keeps your job history.
Try this step in the generatorConfirm and create
Enlarge interface screenshotFor a recognizable remake, start with the built-in motion and Wan 3.0. This is the default for real portraits. Reference-track mode follows supplied audio when an audio source is selected; typing a topic does not rewrite that recording. For an original performance, select a Seedance model in Video settings. That reveals the topic, language, and Both rap controls. Seedance can reject realistic human faces, so use permitted fictional or stylized subjects for that route. The model may change words or timing: it is an original rap generator, not a precise lyric editor.
Back to this generatorA front-facing or gently turned face in even light gives the model a clearer identity reference. Avoid sunglasses, heavy blur, a hand over the mouth, and tiny faces in a group shot. Keep enough of the shoulders visible to establish clothing. If one picture is a sharp close-up and the other is a distant silhouette, the two performers may have very different levels of detail. Uploading a better portrait is usually a more useful first change than increasing output resolution.
In reference-track mode, the left person leads and the right person supports. It does not turn a single vocal recording into two independent singers. Both rap belongs to the original-rap workflow, where the model creates new voices and lines. If the roles are reversed, swap the input photos before a new render. If you want different music, use the audio controls rather than writing a song title into a topic field. Only upload recordings you have permission to use.
Landscape framing gives both performers more room around the shared microphone. A vertical canvas is convenient for a phone feed, but the model must adapt a wider two-person routine into that space. Inspect both faces, hands, and the partner near the edge. The generator supports 5, 8, 10, 12, and 15 seconds; a short lower-resolution attempt lets you assess the cast before spending more credits on a longer or sharper version. A successful but imperfect render still uses credits.
The generator shows a quote for the current settings, and the server recalculates it before submitting a job. Credit packs are one-time purchases. Watching a reference clip or choosing settings is not a personalized free render. Credits are deducted when a generation starts; a technical failure returns them automatically. A successful but imperfect result is charged. Read the pricing page for current packs and the terms for purchase rules.
One-time credit packs. A 12-second 480p AI rap video uses 11 credits, and packs never expire.
Single
$9.99
one-time
$10.99 / 12s at 480p
Duo Pack
$29
one-time
$6.38 / 12s at 480p
Party Pack
$59
one-time
$3.25 / 12s at 480p
Crew Pack
$99
one-time
$3.11 / 12s at 480p
Studio Pack
$199
one-time
$2.74 / 12s at 480p
Add one clear portrait to each photo slot. The left slot casts the lead and the right casts their partner. Keep the built-in routine for a recognizable remake, or choose an original-rap model for a new verse. Check the credits, confirm your permissions, and generate right here in the browser.
Prepare two separate crops first, with one person clearly visible in each, then upload them to the left and right slots. This studio currently takes two separate images; it does not automatically separate the people in a group photo. Keep the face and some shoulders in both crops.
Yes, in original-rap mode. Open Video settings, select a Seedance model, and enable Both rap so the pair can alternate lines. Seedance can reject realistic human faces, so use permitted stylized subjects for that mode. The default Wan routine is a lead on the left with a supporting partner on the right.
For your recording, keep Wan and choose Upload audio: MP3 or WAV, 2–15 seconds, up to 10 MB. For an AI-written verse, choose an original-rap model and enter a topic of up to 120 characters. You can select English, Spanish, Portuguese, French, or Chinese. A topic is not a verbatim lyric script.
Watching the reference and setting up your duo are free. A personalized render uses the credits shown on the generate button. Credit packs are one-time purchases and their credits do not expire; no subscription is required for this workflow. Changing the model, length, or quality updates your quote.
Use two sharp, evenly lit portraits with one visible face each. Avoid sunglasses, tiny faces, or heavy filters. Start with a short 480p clip to check the cast. Available lengths are 5, 8, 10, 12, and 15 seconds. Choose 9:16 for a phone feed or a wider frame to give both performers room.
Try a clearer portrait if a face drifts or the hands look odd. Higher resolution cannot restore details missing from the photo. Technical generation failures return credits automatically. A successfully rendered clip can still have visual flaws, and a new attempt is charged separately. The reference preview is a guide, not an exact-result guarantee.
A completed MP4 appears in the preview beside the generator with playback, sound, and download controls. You can also return to My videos after signing in, including when a job finished while the page was closed. Review both faces and the soundtrack before sharing, and use photos and music you have permission to publish.
Inputs and completed videos are stored privately; they are not automatically added to a public gallery. The initial preview and external posts are references, while scenario illustrations describe ideas. AI may change faces, hands, timing, or sound. Use only photos, footage, and recordings you have permission to use. This independent tool is not affiliated with TikTok, CapCut, the original performers, or template creators. Download a local copy of a result you want to keep, and review it before publishing.
Start from the main studio: bring two portraits together, explore three stages, and choose a reference track or an original rap.
Explore this videoRecreate the orange-booth routine with two portraits, one hanging microphone, and a lead-and-hype performance.
Explore this video