Stages & capture checklist

Movement, not dance

Walking in, turning, sitting down, gesturing — the ordinary motion a video needs between the interesting parts. Full body in frame, nothing crossing in front of you, phone propped up and still. The dance and action clips are asked for separately.

Close-up of your face

Your features are learned almost entirely from these. Without them every generated face is an approximation of you rather than you.

Head and shoulders

The framing most generations actually use. It also teaches your neck, jawline and hairline, which close-ups crop away.

Head to toe

Your proportions, posture and height come from these, and they are what makes full-length and video work look like you standing there.

Facing the camera

The reference angle. Sharp, front-facing shots are also what the video pipeline needs as its identity reference.

Turned three-quarters

Turned about halfway between front-on and side-on. This is where the shape of your face lives, and a set without it renders flat.

Neutral expression

The baseline face. Everything else is learned as a departure from it, so a set that is all smiles cannot produce a resting face.

Natural daylight

Daylight, or a window, shows your real skin tone and needs no correcting. This is the most valuable light you can shoot in.

Outfit changes

The single biggest cause of a model that only ever produces one look. Changing clothes is what separates your face from your clothes.

Waist up

Bridges the gap between face and full body. Skipping it tends to produce a convincing face on a body that is not quite yours.

Side profile

Your nose, chin and brow in outline. A handful is enough, and without any the model guesses your profile from the front view.

Smiling

Teaches how your face actually moves — eyes, cheeks and mouth together. A smile pasted onto a neutral face is the classic tell.

Indoor light

Ordinary room and lamp light. Mixing it with daylight stops the model treating one colour of light as part of your face.

Different places

Shoot in more than one room. If the same wall is behind every photo, the model learns that wall as part of you and drags it into generations.

Ordinary clothes

What you actually wear — jeans, a jumper, a work outfit. This is the baseline the model dresses you in when nothing else is asked for, and a set that skips it can only ever produce a photoshoot.

Swimwearconsent-gated

Shoulders, waist, hips and legs together, which clothes hide and the model otherwise invents. Only asked because you granted swimwear, and only for as long as that stays true.

Lingerieconsent-gated

A different kind of light and a different fit to swimwear, so one does not teach the other. Only asked because you granted lingerie, and only for as long as that stays true.

Serious or intense

Widens the emotional range available later. Not needed for a first model, and worth a lot for editorial work.

Low light

Evening, bar and night light, where most requests end up set. Only useful if still in focus — dim is fine, blurry is not.

Talking to camera

Under ten seconds, looking at the lens, speaking normally. Gives your natural expressions in motion — the thing stills can never show.

Dancing, front on

Twenty to thirty seconds, facing the camera, whole body in frame, phone propped still. Your videos are animated from the movement in these, so this is the clip that makes the motion yours rather than a stranger's. One long take beats five short ones.

Dancing with turns

Twenty to thirty seconds again, but turn through side-on and away and back. Pose is read per frame, so a set that never turns can only ever animate you facing forwards.

Walking and running

Walk toward the camera, past it and away, then again at a jog and a skip — one take, twenty to thirty seconds, whole body in frame. Walk cycles are the motion videos lean on most, and without yours the model borrows a stranger's stride.

Turning on the spot

Turn on the spot both ways, then walk away and look back over your shoulder — one take, twenty to thirty seconds. Pose is read per frame, so calm turns are how the model learns you have a far side outside of dancing.

Sitting down and standing up

Sit down on a chair, stand back up, crouch, kneel — each a couple of times in one slow twenty-to-thirty-second take. Sitting down is the most requested ordinary motion there is, and rushed transitions blur into guesswork.

Arms and hands

Cross your arms, flex, point, wave, fix your hair, hold a phone to your ear — one long take, holding each for a beat. Hands are the hardest thing to animate, and footage of yours doing ordinary things is what keeps them yours.

Standing idle

Just stand there for twenty to thirty seconds — breathing, shifting your weight, touching your hair, looking around. This is the loop a video plays whenever nothing else is happening, which is most of the time, so it earns its take.

Slow and controlled

Half speed — hips, hands, hair, weight shifting. Fast movement blurs and the pose track guesses; slow footage is where the detail that reads as you actually survives.

Everyday gestures

Yawn, stretch, shrug, facepalm, laugh and clap — one long take, holding each for a beat. Ruder gestures are entirely your call: anything you would rather not do is simply skipped, never required, and the clip counts either way.

Athletic moves

Jump, a few jumping jacks, squats, shadowbox a punch or two, a kick — one take at half effort. Controlled beats explosive: fast movement blurs and the pose track guesses, so this is range for the model, not a workout.