A recent photo, face to camera, in soft daylight gives the closest AI video of you. Here is what to pick and what to avoid.
The face photo
- Face to camera, head straight, eyes visible.
- Soft, even light. By a window in daylight is ideal. Avoid hard sun and dark rooms.
- Nothing in the way. No sunglasses, hats low on the brow, hands or hair across the face.
- No filters. Beauty and face filters change your features and the video copies them.
- Recent. Your current hair and glasses, as people know you.
The full-body photo
Use one when the video shows your whole body: dances, outfits, sports.
- Standing, arms by your sides, feet visible.
- Plain background if you can.
- Clothes that show your shape. Very long coats hide how you move.
Pets
- Eye level with the pet, in daylight.
- The whole animal in frame for anything that runs or dances.
- Their real markings visible: the white paw, the patch on the ear.
Products and cars
- Clean, on a plain background, label facing the camera.
- Cars from a three-quarter angle, in even light, no reflections of you in the paint.
Save yourself as a character
In Rift, add two or three of these photos once and save them as a character. Then say “me” in any line to Spark and it uses the same face each time, so a series looks like one person.
Quick check before you Make
Ask: could a friend recognise me from this photo at a glance? If yes, it’s a good start.