Nomad Studio/ Documentation/ Create/ Portraits

Create · Series & batch

Portraits

Portraits describes a character once, then exposes a long list of dedicated dropdowns — gender, age, ethnicity, hair style and colour, framing, art style, lighting, camera angle, expression, background, mood — and has the LLM write N distinct portrait prompts (1–20, default 4) built from those fixed parameters.

The system prompt explicitly requires variety even when every parameter is held constant: it must differ portraits by micro-detail — gaze direction, hair movement, background texture, depth of field, colour grading — so a batch of "4 bust cinematic portraits" doesn't come back as four near-identical images.

Guided walkthrough

1. Step type: Workflow picker

   Choose engine

   Pick the imported engine, or auto-selected if you only have one.

2. Step type: Prompt input

   Character

   Free-text character description — appearance, personality, visual traits.

3. Step type: LLM transform — llm.studio_scenes

   Generate portraits

   Builds N portrait prompts from the character description plus every dropdown below. All but Amount, Gender, Age and Framing are optional.

   • Amount — 1–20, default 4.
   • Gender — default female.
   • Age — 14 ranges, default mid twenties.
   • Ethnicity — optional, default caucasian.
   • Framing — headshot to full body, default bust.
   • Style — 26 options (photorealistic, anime, cyberpunk, oil painting…), default cinematic.
   • Lighting — optional, default soft studio.
   • Angle — optional, default frontal.
   • Expression — optional, default neutral.
   • Hair style / colour — optional, default long wavy / dark brown.
   • Background — optional, default soft bokeh.
   • Mood — optional, default editorial.
   Show the exact system prompt

   You are a portrait photographer, lighting director, and AI image prompt specialist. Given a character description and portrait parameters, generate a series of distinct, visually rich portrait prompts — one per portrait.
   Rules:
   - Each portrait must be fully self-contained and incorporate the provided parameters as its foundation.
   - Vary across portraits: even with fixed parameters, differ in micro-details (hair movement, gaze direction, subtle background texture, clothing detail, depth of field, color grading).
   - Each prompt must include: subject features, hair, framing type, lighting setup, camera angle, expression, mood, visual style, background treatment, and lens/quality cues (e.g. 85mm f/1.4, 8K, film grain).
   - Write in the comma-separated descriptive style used by Stable Diffusion and Flux image generation models.
   - Do NOT number the portrait descriptions in the prompt text itself.
   Output ONLY a valid JSON array of exactly {n} strings. Each string is one complete portrait prompt.

4. Step type: Parameter group

   Parameters

   Seed, steps, CFG — no image-count field here, since the portrait count above already controls the batch size.

5. Step type: Action — generation.submit_queue

   Generate Portraits

   Queues all N portrait prompts.

6. Step type: Result view

   Result

   Save, Upscale, Send to Refine, Regenerate.

Examples