Photos for Solo and Duo AI Rap Videos

A photo can pass a file check and still be a difficult reference for an AI performance. Resolution, subject visibility and the number of people are separate questions. This guide explains the inputs used by the current AI Rap Meme workflow preview.

The file requirements

Use JPEG, PNG or WebP. Each file must be no larger than 8 MB, and both its width and height must be at least 512 pixels. Animated files are not accepted. A larger file is not automatically a better photo: a sharp, well-framed image matters more than file size alone.

The server decodes and re-encodes accepted images, removes metadata and checks the applicable limits. This does not prove consent, identify every person or measure how well a model will preserve a face. You still need to choose appropriate references and have permission to use them.

For a solo performer

Choose one adult with a visible face. A front view or a slight three-quarter angle is a useful starting point. Keep the mouth in view and leave some space above the head and beside the shoulders.

Avoid a hand over the lips, a large microphone hiding the lower face or sunglasses that obscure most recognizable features. Heavy motion blur and extreme beauty filters can make it hard to compare the eventual performer with the reference.

The solo template page explains how one clear topic and a simple performance outline fit this input. Do not prepare an elaborate collage of different poses for the single-photo slot.

For two performers

The duo workflow requires two separate images, one adult in each. Keep the order deliberate: photo one identifies A, photo two identifies B. Both are identity references in the same scene, rather than a before-and-after transition.

Try to give both people a similarly clear face and enough head-and-shoulder detail. If one photo is sharp and close while the other face is a few pixels in a crowd, the two subjects are not equally usable references. Similar lighting helps comparison, but matching outfits are not required.

A group photo is not a substitute for the two upload slots. Multi-person extraction from a single image is outside the current workflow.

Pets and original characters

For a pet, show one animal’s eyes, nose, muzzle and distinctive markings. Avoid several animals overlapping or a toy covering the mouth area. Human-shaped dance instructions may not suit an animal body.

For an original character, use one clean view with a recognizable silhouette and consistent colors. Character sheets with many poses, large captions and speech bubbles introduce several possible subjects. Real pet and character performance quality has not yet been validated here.

A custom stage is its own reference

Choose the custom-stage option if you want to supply an environment image. It is optional and separate from the performer photos: one performer plus a stage uses two images; two performers plus a stage uses three.

Prefer an environment where the intended focus is obvious. People, signs or logos in a stage photo should not become extra performers or instructions. The stage preview is a composition aid, not evidence of a finished AI composite.

Before you press create

Confirm the template, photo order and vocal language. Use only material you can authorize, and only consenting adults for human performers. Refreshing or signing in can clear photos held in browser memory, so be ready to select them again.

Real generation and payment are currently closed. The free workflow demo returns a labeled placeholder without rap audio. Start at the template directory, or read the complete planning guide and privacy notice.