Image - 2026-10-01 16:25
There are no identity images. Create one believable person strictly from the written identity specification and keep that exact same person in every view without identity drift. IDENTITY SPECIFICATION — ALL FIELDS ARE BINDING: { "name": "Aisha Mwangi", "gender": "female", "pronouns": "she/her", "age": "24–26", "ethnicity": "Kenyan (Kikuyu)", "origin_country": "Kenya", "current_location": "Nairobi, Kenya", "height": "5'6\"", "body_type": "Slender, leggy", "skin_tone": "Rich medium-dark brown", "hair": "Short natural TWA (teeny weeny afro), dark black; sometimes adds a thin headband", "eyes": "Warm brown", "distinguishing_features": "Long graceful neck; small round scar on right cheek from childhood", "immutable_identity_prompt": "Aisha Mwangi, female, age 24–26, Kenyan (Kikuyu), from Kenya, living in Nairobi, Kenya. Appearance: 5'6\", Slender, leggy, Rich medium-dark brown, Short natural TWA (teeny weeny afro), dark black; sometimes adds a thin headband, Warm brown, Long graceful neck; small round scar on right cheek from childhood. Life context: Freelance Digital Media Producer, Single, Matatu commutes, rooftop café work sessions, Swahili rap playlist, weekend hiking. Personality: Sharp, Idealistic, Patient teacher, Privately competitive. Aisha studied communication and digital media at the University of Nairobi and started freelancing immediately after graduation. Her early AI video projects for local NGOs revealed one crushing problem: AI characters looked completely different from scene to scene. She developed a systematic prompt-engineering approach to maintain character consistency and started sharing it for free — which quickly turned into a following. Preserve this exact identity, face, body proportions, age and background in every generation.", "biography_context": "Aisha studied communication and digital media at the University of Nairobi and started freelancing immediately after graduation. Her early AI video projects for local NGOs revealed one crushing problem: AI characters looked completely different from scene to scene. She developed a systematic prompt-engineering approach to maintain character consistency and started sharing it for free — which quickly turned into a following.", "wardrobe_and_visual_rules": "Светлый фон с лёгкой перспективной сеткой внизу кадра 9:16; рисованный 2D cel-shading персонаж крупно снизу; сверху — карточки и стикеры: мультяшные иконки с толстым тёмно-фиолетовым контуром, белой наклейной обводкой и жёлто-оранжево-фиолетовой палитрой (монеты, камера, бейдж-галочка, речевой пузырь с абстрактными символами, «поплывшее» лицо). Крупный чёрный текст верхним регистром, один фиолетовый акцент #6D4AFF.", "forbidden": "фотореализм, реалистичная кожа и фото-текстуры, тени под персонажем, фон за персонажем, текст и логотипы на персонаже, посторонние люди и предметы, обрезанные части тела, лишние или деформированные пальцы, кисти больше лица, имитация реального интерфейса сайта, чужие персонажи из референсов" } OPERATOR'S CHARACTER-SHEET PROMPT: Create a character identity sheet for the ONE fictional adult described in the account's IDENTITY SPECIFICATION. IDENTITY Use the account's name, age, face, hair, skin, build and distinctive features. If identity references are attached, preserve that account's exact likeness. Do not copy a presenter from the direction's example videos or project archive. Across accounts the people must be distinct; within this sheet the person must be identical. STYLE AND WARDROBE Follow the direction's wardrobe_and_visual_rules. Keep one consistent outfit, hair, proportions, lighting and rendering style across all views. If no style is specified, use a polished stylized 3D editorial presenter with natural anatomy. Avoid celebrity likenesses, brand marks and watermarks. SHEET LAYOUT Use a clean white 16:9 canvas with clearly separated panels: one front portrait, one three-quarter portrait, and five waist-up presenter poses. Keep the same head size and eye line across the five poses, leave space around the head and hands, and show every hand completely. Never overlap panels. FIVE POSES 1. Open palm: calm explanation, friendly expression. 2. Hand extended to the side: presenting an example. 3. Stop palm: a clear warning, focused expression. 4. Hand near chin: thoughtful analysis. 5. Both palms open: explanation or conclusion. CONTINUITY Only the gesture and expression change between poses. Preserve facial geometry, clothing, colors and body proportions. No extra people, extra limbs, props or decorative backgrounds. No captions or lettering. This sheet is the identity reference; separate animation sprites are generated from it later. The source rule above has priority: when no images are attached, interpret any mention of "provided images" as the written identity specification instead. CONSISTENCY REQUIREMENT: This is a casting/identity master asset for future video generation. All views must depict one exact person, with identical facial geometry, apparent age, body proportions, skin, hair and distinctive features.
Free to start · Generate videos and images with AI in seconds