THE RAP DUO FIELD GUIDE
How to make an AI rap video from two photos
A practical guide to photo choice, topics, credits and downloading your rap duet.
Updated October 23, 2026
Choose your duo
Start with two portraits of people who have agreed to participate. Good lighting, a visible face and a little space around the head work best. Match the general photo quality if possible. You do not need a photograph of both people together.
Pick the right scene
Orange Booth is warm and playful. Luxury Lobby gives the duo a cinematic setting. Studio Booth focuses on the performance. Street Cypher brings outdoor energy. All scenes use the same short two-person video format. Available scenes are shown in the generator.
Keep the topic simple
One specific idea is better than several competing instructions. Try “Best friends since day one” or “Celebrating our first small win.” Avoid long lyrics and exact word-count requests. The video model creates new music and words, so pronunciations and lyrics can vary.
Set the length and frame for your idea
Standard makes a short clip of up to 15 seconds; Seedance 2.5 offers lengths up to 30 seconds where that model is available. Pick a vertical 9:16 frame for a phone post, square 1:1 for a compact share, or 16:9 for a wide frame. The creator shows the available resolutions and updates the credit quote as you change settings. A longer clip gives the duo more room to perform but uses more credits.
Understand the credit step
One standard 15-second 720p video reserves 45 credits. Packs have no expiry; subscription credits expire at the end of their paid period. Credits closest to expiry are used first. A confirmed failed render returns the reserved credits. If a task is being checked, wait for that task rather than starting the same request again.
If a face or lyric comes out differently
Watch the whole result before sharing. The photo guides the character, but it cannot lock every expression or mouth movement; a topic guides the music without guaranteeing exact words. For another attempt, change one input at a time—for example, use a clearer portrait or a shorter topic—so you can tell what helped. A creative variation in a delivered clip is different from a confirmed render failure.
Download and share thoughtfully
My videos contains active and completed requests. Play your result, check the faces and sound, then download the MP4. Rap Duo does not publish your result automatically. Tell viewers it is AI-generated when the context could otherwise mislead them.