Model Generator

Model Generator is the first of the Creative Intelligence nodes. There is no prompt box and no model menu: you describe the person through a form of studied options, and the server writes the words for a hidden image engine. What you own is the specification, and the node turns it into a photoreal model with the same identity in every shot.

Every section except gender and age can also switch to a reference picture: your own wardrobe, hair, posture, background or lighting, taken from an image rather than a menu.

1. 1

   Add Model Generator

   Find it in the Creative Intelligence group and add it to the canvas.
2. 2

   Describe the person

   Set gender, age, body type, heritage, skin tone, hair color and hair style. Auto leaves a choice to the engine.
3. 3

   Set the shot

   Choose posture, framing from full body to beauty close-up, expression, wardrobe, background and lighting.
4. 4

   Pick the angles

   Tick the views you want: Front, 3/4 Left, 3/4 Right, Profile Left, Profile Right, Back. Each is one generation.
5. 5

   Choose the output

   A portrait ratio, 3:4, 2:3 or 9:16, a quality of Low, Medium or High, and 2K or 4K.
6. 6

   Press Run

   The button quotes the price of every angle together. One Image Preview appears per angle.

Identity

Gender, age range, body type, heritage, skin tone, hair color, hair style. Body type and hair style offer different choices for each gender.

Pose and frame

Posture, from standing to dynamic, and framing, from full body down to a beauty close-up. Expression sets the face.

Wardrobe

A basic outfit by default, or a reference picture of the garments you want on the model.

Background and lighting

Studio white, grey or black, outdoor urban or nature, a minimal interior, or a reference. Lighting lists only what the chosen background can physically produce: a studio never offers daylight.

Switch any of body type, heritage, skin tone, hair color, hair style, posture, expression, wardrobe, background or lighting to Reference image. The node opens a port for that section and hands you a wired, renamed Image Input to drop the picture into.

Each reference is read for its own feature and nothing else: a wardrobe reference gives the garments, not the person wearing them; a hair style reference gives the cut, not the color. Several sections can take references at once, and the hidden prompt binds each one by its slot, so the model knows which picture is which.

Each ticked angle is one generation, and the price on the button is their sum. The first angle in the list is the identity hero: it is made first, and every other angle references it so the face, the build and the outfit stay the same from view to view.

Run Overwrite replaces the previews from the last run; a normal run adds new ones. The output quality and size apply to every angle.

Wire a Batch Run straight into Model Generator and a Batch group appears on the node with one choice: Each item fills. Pick the section each picture should land in, wardrobe by default. The run then repeats for every picture in the set, with the same person wearing each garment in turn.