XAZINGA

SERRE 아이스크림 포트레이트 프롬프트 사용자가 보내는 얼굴과 헤어스타일 이미지를 반드시 참조하여 동일 인물 기반으로 생성할 것. 얼굴 구조, 눈 분위기, 코 형태, 턱선, 헤어 길이와 흐름은 사용자 이미지를 우선 기준으로 유지. 자연스럽고 현실적인 한국인 여성 분위기. 과한 AI 미인형 느낌 금지. 아이돌 느낌 금지. 인플루언서 메이크업 느낌 금지. 표정은 은은하게 장난기 있는 표정. 입꼬리가 자연스럽게 살짝 올라가 있음. 입은 아주 살짝 벌어진 상태. 혀를 살짝 옆 볼 방향으로 내민 모습. “낼름” 느낌이 아니라, 조용히 장난치는 듯한 악동 느낌의 표정. 눈빛은 부드럽고 맑지만, 약간의 mischievous mood 포함. 과한 섹시 표정 금지. 의상은 시원한 느낌의 연보라 튜브탑. 화이트 + 연보라 체크 패턴. 여름 무드. 부드러운 원단 질감. 어깨와 쇄골 라인이 자연스럽게 드러남. 메이크업은 깨끗하고 투명한 피부 표현 기반. 양 볼에는 아주 은은하게 반짝이는 스프링클 글리터 메이크업. 과한 펄 금지. 빛에만 은근히 반짝이는 수준. 아이스크림 콘은 한 손으로 들고 있음. 콘의 SERRE 로고 윤곽과 형태 유지. 아이스크림 위에는 작은 리본 토핑 장식이 뿌려진듯 올라가 있음. 핑크1층 민트2층 계열 아이스크림. 아이스크림을 든 손목에는 오로라 느낌의 투명 구슬 비즈 팔찌 착용. 머리에는 아이스크림위에 뿌려진 토핑 같은 리본만 남기기. 과한 헤어 액세서리 금지. 배경은 몽환적인 실버-오로라 무드. 커다란 반짝이는 은색 구체 조형물 2~3개가 공중에 부유하는 모습. 홀로그램 반사, 오로라빛 반짝임, 유리 구체 같은 재질감. 전체 무드는: dreamy, sparkling, soft pastel fantasy, SERRE aesthetic, clean glossy studio portrait, quiet playful energy. 조명은 부드러운 화이트 + 오로라 반사광. 깨끗하고 맑은 하이라이트. 광택감은 있지만 플라스틱 피부 금지. 실제 피부결은 자연스럽게 유지. 고해상도, 실사 기반, DSLR portrait photography, soft bokeh, gentle glow, 4:5 vertical composition.

iphone photograph, A natural photo of a stylish Korean woman in her 20s sitting on a wooden bench in front of a bakery. She is wearing a white knit sweater. The scene features warm natural light, a European street cafe atmosphere, film-style photography, soft vintage tones, and a warm, lovely mood.

surreal fashion portrait of the woman duplicated in multiple dynamic poses and scales, playful chaotic composition. model with long braided pigtails decorated with colorful fabric ribbons and flowers, straight bangs, glossy makeup with glitter and rhinestones around the eyes. wearing a yellow patterned crop top, pleated dark mini skirt, chain belt, cowboy-style boots. expressive gestures, energetic movement, wide-angle lens distortion, white studio background, high-fashion editorial aesthetic, sharp lighting, vivid colors, high detail.

1. 추천 프롬프트 Generate four images arranged in a 2×2 grid layout. Consistency Rule (Critical) Use the attached reference image as a strong anchor to preserve subject consistency across all four frames. The same person must appear in every image with consistent overall facial proportions, body scale, hairstyle, and clothing. Do not change identity between frames. At the same time, do not replicate the reference image exactly or copy any identifiable facial details. The subject must remain non-identifiable and clearly not a real-person replica. Variation Rule (Critical) Each of the four frames must depict a different accidental micro-moment of the same subject within the same scene. Treat the images as if they were captured seconds apart, unintentionally, by a handheld smartphone. No frame may repeat the same pose, hand position, head angle, or motion blur pattern. Scene Nighttime urban sidewalk beside a building with white ceramic tile or plain concrete walls. Quiet city atmosphere with minimal readable background detail. All four frames share the same location, lighting conditions, and general camera distance. Pose and Motion (vary per frame) Across the four images, vary naturally: – the degree of head turn while looking back, – the height, angle, and distance of the raised hand blocking the camera, – the stage of the walking motion (mid-step differences), – how close the subject is to exiting the frame. Each frame must feel like a different instant of the same rushed walk, not a posed sequence. Camera and Motion Modern smartphone camera, handheld, rushed capture. Strong uncontrolled camera shake. Heavy directional motion blur affecting hand, hair, and face differently in each frame. Facial features are partially smeared and distorted by motion, faintly visible but not readable. Subtle ghosting and motion trails vary per image. Framing is imperfect and inconsistent across the four frames. Lighting On-camera smartphone flash fired mid-motion in all frames. Harsh flash highlights on skin and hand. Deep surrounding darkness. Uneven exposure with blown highlights and hard shadows typical of hurried night phone photography. Background The tiled or concrete wall remains consistent across all frames. Motion blur causes the wall to streak differently in each image. Street details dissolve into darkness with minimal legible information. Mood Candid, intrusive, shy, slightly playful. Feels private, raw, and unpolished. Each frame captures a fleeting, awkward reaction — never posed, never intentional. Image Quality Very high noise. Heavy blur and motion distortion. Raw smartphone aesthetic. No cinematic treatment. No editorial polish. Strict Exclusions anime, illustration, painting, stylized imagery studio or beauty lighting sharp focus or clean facial detail perfect anatomy or smooth skin fashion editorial or posed portrait cinematic color grading or film still look tripod or professional photography identity change between frames face copying or real-person replication ar=4:5

photorealistic, real human, natural skin texture, real photograph, raw photo, DSLR quality, 8k uhd, realistic lighting, pores, skin details, full character reference sheet, of a woman. wearing casual, 흰색반팔, 데님, 스티커즈. multiple views showing: front full body view, back full body view, face close-up front view, face close-up side profile view, face close-up 3/4 angle view, upper body detail shot, outfit detail shots. 16:9 aspect ratio, white clean background, professional studio soft lighting, fashion photography quality, high detail, consistent character across all views, professional character design sheet layout --- CHARACTER TRAITS (for text overlay reference): MBTI: not specified Voice: not specified --- NEGATIVE PROMPT: 3D render, CGI, game character, anime, illustration, digital art, plastic skin, doll-like, artificial, fake, cartoon, drawing, painting

Shibuya crossing filled with humans and yokai, chaotic urban Tokyo scene, neon lights and crowded street, a giant Buddha appearing above the city, radiating soft golden light, peaceful and compassionate presence calming the scene, darkness fading into light, cinematic wide shot, low angle, worm’s-eye view, looking up at the towering Buddha, dramatic perspective, atmospheric, highly detailed

업로드한 이미지의 메인 피사체를 분석해서 귀여운 레트로 픽셀 아이콘으로 재구성해줘. 1. 주제 판별 (사람/동물/음식/사물) 2. 핵심 실루엣과 색감 추출 3. 레트로 게임 스프라이트 스타일로 변환 픽셀: 크고 선명한 도트, 블러 없음, Aseprite 느낌. 색감: 원본 기반 레트로 팔레트, 흰색 배경. 구도: 중앙 1:1 정사각형, 주제 60~75%. 금지: 3D 렌더, 복잡한 배경, 음식에 얼굴.

Majestic snow leopard sitting on a mountain ridge at sunset, golden hour, photo-realistic wildlife photography, telephoto lens

Astronaut floating in deep space, Earth reflected in helmet visor, stars, nebula in background, hyper realistic