============ PART A: INPUT (EDIT THIS ONLY) ============ [RATIO: 3:4 / 4:5 / 1:1 / 9:16] [LAYOUT_VARIANT: pilih satu: 1 / 2 / 3] [SUBJECT_MODE: portrait / person with object / product / interior scene] [CAMERA_FEEL: keep original angle / high-angle wide lens (fisheye feel) / eye-level] [FACET_DENSITY: large facets (very low-poly) / medium / fine] [PALETTE: keep colors from the photo / warmer / cooler] [MATERIAL: matte / matte-to-semi-gloss toy-like] [INSET_PHOTO: yes / no] [CARD_TITLE: maksimal 1 kata, or none] [CARD_MODEL_NAME: maksimal 3 kata, or none] [CARD_RATIO_LABEL: contoh 3:4, or none] [CARD_BODY: maksimal 40 kata, or none] PHOTO TO ATTACH: SOURCE_PHOTO (one clear, sharp photo; the subject fully visible). I confirm that it is a photo I took, or that I have permission to use it and to edit the people in it (adult only). DATA RULES: 1. Do NOT say who the people are. No names, signatures, logos, or text other than the fields above. 2. If a field is "none" or "no", leave that element out completely. 3. Before creating the image, tell me in two short lines what you understood (layout, subject mode, camera feel, palette) and wait for my "OK". LAYOUT_VARIANT options: 1 = Full-bleed low-poly render with INSET_PHOTO (small rounded white-outlined frame, top-left) and a frosted-glass prompt card at the bottom. 2 = Full-bleed low-poly render only, no inset, no card (text only if provided, small, at the bottom). 3 = Split: top 50% SOURCE_PHOTO unchanged, bottom 50% low-poly render, with the card in the middle. ============ PART B: STYLE (DO NOT CHANGE) ============ TASK: Transform SOURCE_PHOTO into a stylized low-poly 3D animated render, similar to Pixar/DreamWorks-style character art, ratio RATIO. STEP 0: If no photo is attached, stop and ask me to attach one. Do not invent a scene. SUBJECT LOCK (most important): - Keep the exact composition, perspective, pose, position, expression, clothing design, number of people or objects, and layout of SOURCE_PHOTO. Do not add, remove, or move anything. - People: keep each person recognizable and unchanged (face shape, hair, accessories, age, skin tone). Do not beautify, slim, or alter skin tone or expression. Do not identify them. - Product: keep exact shape, color, label, printed text, and logo recognizable; only the rendering style changes. - Face, eyes, hands, and any text on a product must stay clear and correct. LOOK: - Convert every element (subject, background, environment) into faceted low-polygon geometric shapes with visible flat-shaded triangular facets, density per FACET_DENSITY. - Preserve the original color palette and light direction of SOURCE_PHOTO (skin tones, clothing colors, background colors, light and shadow placement), only reinterpreted through faceted low-poly shading. PALETTE adjusts only if not "keep colors". - Soft cinematic depth of field: main subject sharp, background softly blurred. CAMERA_FEEL controls the lens look. - Warm, soft directional lighting with gentle ambient occlusion in the creases between facets. - MATERIAL finish: polished 3D animated film still, high-quality render, not flat cartoon, not photorealistic. Small metallic details (glasses frames, jewelry) keep a subtle shine. POST ELEMENTS (only if enabled): - INSET_PHOTO: SOURCE_PHOTO unchanged in a small rounded-corner frame with a thin white outline, top-left, about 25-30% of the width. - Prompt card: rounded-corner frosted glass panel (translucent, soft blur) across the lower part. Inside: CARD_TITLE in bold white sans-serif at top-left; CARD_MODEL_NAME in regular white below it; CARD_RATIO_LABEL aligned to the right on the same line; CARD_BODY in small white italic sans-serif, justified, below. - Spell every word exactly as given. If text cannot render correctly, make it smaller rather than invent letters. Text must not cover the face or the product. QUALITY CHECK: (1) composition, pose, and subject identical to the photo, (2) face/product recognizable and unchanged, (3) facets clearly visible across all elements, (4) colors and light direction match the photo, (5) subject sharp, background softly blurred, (6) text spelled exactly, small, readable. AVOID: changed identity, pose, or expression; added or removed elements; melted hands or face; smooth photorealistic skin; flat 2D cartoon look; neon or oversaturated colors; harsh black outlines; extra text, watermark, logo, or signature; messy facets that destroy the face or product details.
Salin prompt di atas.
Buka ChatGPT. Kalau punya foto produk atau foto model, lampirkan dulu.
Ganti bagian yang berwarna kuning dengan data usahamu, lalu kirim.
Pilih rasio 4:5 saat membuat gambar agar hasilnya sesuai.









