GPT Image Prompt
Same real person from the active LoRA identity reference image, preserving her exact recognizable facial geometry and full upper-body identity: same face shape, eye spacing and eye structure, nose, lips, jaw, cheek structure, natural skin tone, dark tied-up hair identity with loose wispy bangs and black bow, same shoulder-to-head relationship, shoulder width, upper-body proportions, torso presence, natural posture behavior, and realistic physical presence. Do not replace her with a similar-looking woman, do not beautify her into a generic attractive subject, and do not alter her body geometry, shoulder width, torso balance, waist thickness, or natural physical scale.
She is standing in a crowded train carriage, body turned slightly sideways in a physically believable public-transit stance. One arm is raised above the frame holding an overhead support, with the raised arm naturally cropped by the top edge of the vertical phone photo. Her other hand holds a smartphone near her chest with realistic grip pressure, believable finger spacing, natural wrist angle, and the phone physically connected to her hand rather than floating. Her black shoulder bag strap crosses her torso, pressing lightly into the cream sleeveless blouse. The blouse has small front buttons and vertical pleat details, with fabric draping naturally over her body, small wrinkles forming under the strap and around the torso from posture and movement. She wears a high black waistband or skirt, visible enough to preserve her natural upper-body proportions without fashion-model reshaping.
Her face is caught in a quiet real-life moment, not posed: eyes lowered toward the phone with relaxed eyelids, softly closed lips, calm absorbed expression, private and slightly distracted, with living human presence rather than blank stiffness. Preserve the source-prompt human moment while keeping the active LoRA facial structure intact; the expression should not reshape her into a different person, should not become seductive, dramatic, commercial, or influencer-like, and should not copy any dead-eyed or stiff reference-photo expression.
The crowded train environment feels ordinary, practical, and physically real: nearby passengers partially visible behind and beside her, metal poles, overhead train handles, carriage walls, signage fragments, reflections, small background clutter, and imperfect visual overlap. Background passengers and poles softly recede behind her with believable environmental scale and smartphone perspective. Feet and lower body may be partially cropped by the close vertical framing, but the visible body must remain grounded, weight-aware, and consistent with the active LoRA identity.
All expression, gesture, outfit, pose, camera behavior, lighting, and realism details remain subordinate to the active LoRA identity reference. The final image must still be unmistakably the exact same LoRA subject in face and body presence, not a similar-looking substitute, not a beautified replacement, not a slimmer or more model-like version, and not a generic attractive interpretation of the prompt.
Casual vertical iPhone-style photo from close distance, handheld and slightly asymmetric, as if taken by someone standing nearby in the train. Mild wide-angle smartphone perspective is allowed, including a slightly enlarged raised foreground arm and imperfect crop balance, but the lens must not slim her waist, elongate her body, narrow her shoulders, alter torso proportions, or change her recognizable physical presence. Bright practical train lighting from overhead carriage lights, mild smartphone HDR, small-sensor depth behavior, uneven sharpness between face, blouse fabric, hand, phone, hair, and background. Natural hair flyaways, subtle skin texture, realistic smartphone compression softness, slight reconstruction inconsistency, mild exposure imbalance, practical public-transit clutter, ordinary social-photo realism.
Avoid cinematic polish, studio lighting, fashion editorial posing, influencer beauty, luxury train aesthetic, perfect symmetry, over-clean HDR, excessive bokeh, DSLR portrait look, plastic skin, over-sharpened details, mannequin expression, blank stare, seductive gaze, commercial smile, AI-perfect hands, fused fingers, floating phone, disconnected bag strap, impossible arm position, unrealistic train geometry, staged portrait feeling, and any body or face beautification that reduces LoRA identity accuracy.
How to use: Copy and paste this prompt into ChatGPT, Gemini, Nano Banana, or your favorite AI image tool. If it mentions LoRA, reference image, uploaded photo, or identity reference, upload your own photo first. No reference photo? Replace the character with a clear age, gender, and nationality, such as “25-year-old Japanese woman” or “30-year-old American man”.