Best Photo Requirements for Tiny Desk AI

The photo you upload is the identity reference for one selected Tiny Desk AI performer position. A good reference does not need to look like the final video. It needs to be clear enough for the workflow to understand the person or pet you are assigning to Lead, Sax, Keys, Back left, or Back right.

Use one subject per selected position

The cleanest setup is one person or one pet in each uploaded image. If you are making a group result, upload separate references for each position instead of one crowded group photo. Separate inputs make the mapping explicit: one file belongs to Lead, another belongs to Sax, and so on.

If you only want to replace one performer, upload one photo. You do not need to prepare five images just because the reference contains five people. Empty positions remain unchanged in the fixed video, which is useful for testing one identity before committing to a larger cast.

Keep the face or pet features readable

Choose a sharp image where the eyes, nose, mouth, jawline, hairline, or pet face are easy to see. Avoid strong motion blur, tiny faces at the edge of a group photo, sunglasses that hide the eyes, and filters that change the natural shape of the subject. A neutral or natural expression is often easier to read than a heavily distorted meme face.

Pets work when the animal’s face is visible and the reference is not mostly fur, darkness, or a distant full-body shot. For a dog or cat meme, a front-facing head-and-shoulders photo is usually a stronger starting point than a blurry action photograph.

Prefer even light and a useful crop

Soft, even light helps keep important features visible. Avoid a hard shadow across one side of the face, colored lighting that hides skin or fur detail, and bright windows behind the subject. The background does not need to be blank, but clutter should not overpower the subject.

Crop out distractions while keeping the complete face and hairline or pet head visible. Do not crop through the eyes, ears, or mouth. A portrait crop is appropriate for this workflow because the fixed template supplies the stage composition, clothing direction, and performance context.

Check the file before uploading

The current creator accepts JPG, PNG, and WebP images, with a maximum file size shown in the upload control. Use the original clear file instead of a compressed screenshot from a messaging app. If your image is too large, resize a copy while keeping the face readable. Do not keep re-saving a JPEG until fine details disappear.

After upload, check the position label and thumbnail. The selected reference should be visually distinct from the sample state. If it is mapped to the wrong performer, remove it and upload again before submitting. The creator is designed to let you correct the cast before a generation request starts.

Permission matters as much as image quality

Only upload a photo and likeness you are allowed to use. A profile picture found through search is not automatically available for transformation or publication. Get permission for adult friends, use your own pet photos, and do not upload minors, public figures, celebrity likenesses, or protected promotional images. If the final video is shared, label it as AI-generated when required and do not imply an official endorsement.

When you are unsure, start with one authorized photo and one performer position. Read the full photo guide, then open the Tiny Desk AI creator when your input is ready.