Select AI Model
{ "positive": "Mid-shot fashion portrait of a young woman with the exact facial structure of the reference, long {argument name="hair color" default="blonde"} hair flowing down her back, and a small black bindi. She is wearing a delicate white ruffled spaghetti-strap crop top that reveals her waist and navel in a classy, natural way. She is standing with a slight curve in her posture, one hand resting on her hip, looking directly at the camera with a warm smile. Direct on-camera flash lighting, soft vintage aesthetic, highly detailed skin texture, realistic shadows, resolution, candid lifestyle photography.", "negative": "deformed, ugly, bad anatomy, disfigured, crossed eyes, morphing, extra limbs, blur, watermark, text, cartoon, illustration, painting, low quality, pixelated, oversaturated, bad hands, missing fingers", "parameters": { "aspect_ratio": "9:16", "steps": 30, "cfg_scale": 7.0, "sampler": "", "seed": -1, "style": "Photorealistic" } }, Dramatic low-angle hero perspective looking upward, captured on Sony A7R V paired with FE 85mm f/1.4 GM lens, High-key crisp editorial studio lighting with subtle catchlights in eyes, authentic skin micro-textures, true-to-life surface physics, 8K ultra-detailed editorial photography
Master Version
Prompt Architecture & Anatomy
🧠 Prompt Engineering & Structural Breakdown
- Core Intent: Achieve precise task completion with zero conversational filler and maximum actionable depth
- Role & Persona: Domain expert practitioner with specialized technical insight and pedagogical clarity
- Context & Assumptions: Clear input parameters, defined audience proficiency, and stated foundational assumptions
- Constraints & Boundary Rules: Strict negative constraints: no vague advice, no unsupported claims, adherence to requested format
- Output Structure & Format: Hierarchical markdown organization with bold callouts, ordered steps, and clean tabular data where applicable
- Verification & Quality Checks: Self-correcting validation step ensuring all prompt constraints and sub-questions are fully addressed
