10% off monthly plans with code NODE10, until August 31
Three photos of you are enough to end up pressed against a rock while a cyclops hunts for you. The Camera node shoots the still and the video through the same lens.
Real workflow, real results. Pan and zoom freely.
Use this workflowThis workflow is the real template: the reference sheet carries the face, the Camera node feeds the image AND the video. Move around the canvas and open the nodes.
At a glance
Putting yourself in an epic, a film noir or a science fiction scene asks two things ordinary generators handle badly together: your face has to stay yours, and the shot has to look like cinema rather than a posed portrait. This workflow separates the two and wires them.
The face lives in a three-angle Reference sheet, called with @ in the prompt: the model receives your actual photos and places you in the set without transforming you. The staging lives in a Camera node, with its focal length, aperture, angle and movement. That node feeds both the still and the video, so the two are shot through the same lens, which a prompt retyped by hand never guarantees.
Seedream 5 Pro composes the shot from the sheet, and Kling V3 animates it over ten seconds with sound. That choice is deliberate: Seedance models refuse starting images that may contain a real person, which rules out any photoreal shot with a face. The guide on keeping a character consistent explains how to shoot your sheet, and directing the camera covers the settings.
Build your sheet
Three photos: front, profile, three quarters. Even light, neutral background, neutral expression. That sheet is what puts you in the shot, not a description.
Set the camera
The Camera node carries focal length, aperture, angle and movement. A low angle 35 mm at f/2 does not tell the same story as an 85 mm at eye level.
Describe the scene, not your face
The set, the light, what threatens off screen. Your face is already handled by the sheet: the prompt is there to build the world around you.
Animate ten seconds
One single take, a slow move, one simple action. Ten seconds is the length that holds without the face drifting.
The same canvas, tuned for different needs: each variation is two or three lines of prompt away.
Three photos of you in the Reference card, an epic setting in the prompt, and there you are in the shot. The Camera node does the rest: long lens, slight low angle, late-day light. Framing, not the setting, is what separates a montage from a poster.
A chest-up shot in square format is enough. Ask for strong side light and a simple background: at thirty pixels wide, only the contrast on the face reads. See also the professional headshot for the sober version.
The birthday person as a knight, an astronaut or a manga heroine, printed as a poster. Be upfront that the likeness depends on the photos provided: three well-lit angles beat ten selfies shot from the same spot.
The costume is described in the prompt, not sewn. Describe the material rather than a licensed character: weathered plate armour gives a better render, and avoids generating a copy of protected work.
The image becomes a ten-second video with Kling V3: the camera pulls back, the cape moves, the eyes lift. At five seconds it costs half as much and still carries a channel intro.
The Reference card needs different angles, not volume. Front, profile, three-quarter, in natural light: that is what keeps the face stable when the scene changes.
Three lines of scenery and nothing about the camera gives a flat image. The reverse works better: one sentence for the setting, precise framing in the Camera node.
Costume, set, light and pose all changed at once makes it impossible to know what helped and what broke. One change at a time, keeping both versions side by side on the canvas.
Yes, as long as the sheet has several angles and even light. It is the number of viewpoints, not the quality of the camera, that makes the likeness.
Yes, the sheet and the Camera node stay in place: only the set and the action change. Going from an ancient cave to a spaceship corridor takes rewriting two prompts.
Only use photos of yourself or of people who have given you their consent. Staging someone else's face without their agreement is not an acceptable use.
Three are enough if they show three angles. Add a fourth if you wear glasses or have a very distinctive haircut, and describe in the card what must never change.
No. They are attached to the generation as references, for the duration of the call, and stay in your library. Nothing is reused for training.
Yes, with two Reference cards mentioned in the same prompt. Expect more attempts: models handle two simultaneous faces less well than one, and wide shots hold up better than close-ups.
The image is five credits, the ten-second Kling V3 video 303, so about three euros for the whole canvas. Drop the video to five seconds and the shot costs 1.60 euros, shown on the button before you generate.
Put yourself in a film with AI
Use this workflowFree account, no card required. The workflow opens pre-filled in your canvas.
Seedream 5 Pro
Kling V3
Nano Banana