Choose photos for your Smalltown Boy AI video
Photo formats, file limits, framing tips and role selection for people and pets.
Smalltown Boy AI needs two separate photos, with one clear person or pet in each. A supplies the opening reaction; B supplies the dancer behind. You can swap the roles before generating.
File requirements
- Formats: JPG, PNG, WebP, BMP or AVIF. HEIC/HEIF works only when your browser can decode it; otherwise export as JPG or PNG.
- File size: up to 20 MB per photo.
- Dimensions: at least 240 pixels on each edge.
- Shape: the longer edge must be no more than 8 times the shorter edge.
- Very large images may need resizing if your browser cannot process them. Follow the upload message before continuing.
These are upload requirements. A file can meet them and still be blurry or poorly framed.
People: keep the face and shoulders visible
Choose a sharp, evenly lit portrait with the face unobstructed. Keep useful hair, shoulder and clothing detail. Avoid group photos, sunglasses, hands over the face and heavy beauty filters. All identifiable people must be 18 or older and agree to this AI transformation.
Useful framing: one person with a clear face and visible shoulders. Change the crop: several people, a tiny distant face, or a frame showing only the eyes.
Pets: keep distinctive markings visible
Use one animal per image. Keep its face, ears, fur markings and useful body detail in the frame. Avoid dark silhouettes and photos where the animal is a small part of the background.
See the fictional cat and dog input images for the starting point of our demonstration. Real pet results can vary.
Before you generate
Check both photos and the A/B order. Photos guide appearance; the template supplies the dance. A different pose does not request new choreography. Use different subjects for the two roles; matching one subject across both roles is not guaranteed.
Use original photos when available. Enlarging a tiny screenshot cannot restore missing detail, and choosing a higher output resolution does not guarantee a better likeness. Watch the completed video for changes to faces, hands or paws before sharing.