Image - 2026-09-26 07:49

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": "Elias Bergström", "gender": "Male", "pronouns": "he/him", "age": "34–34", "ethnicity": "Northern European / Swedish", "origin_country": "Sweden", "current_location": "Berlin, Germany", "height": "6′1″", "body_type": "Tall, lean, slightly lanky", "skin_tone": "Fair, slightly ruddy cheeks", "hair": "Dirty blond, medium length, parted loosely to one side; tends to push it back mid-conversation", "eyes": "Grey-blue", "distinguishing_features": "Rectangular wire-framed glasses; a faint scar on the right side of his chin from a soldering mishap in his teens", "immutable_identity_prompt": "Elias Bergström, Male, age 34–34, Northern European / Swedish, from Sweden, living in Berlin, Germany. Appearance: 6′1″, Tall, lean, slightly lanky, Fair, slightly ruddy cheeks, Dirty blond, medium length, parted loosely to one side; tends to push it back mid-conversation, Grey-blue, Rectangular wire-framed glasses; a faint scar on the right side of his chin from a soldering mishap in his teens. Life context: Software developer & AI video cost analyst, Single, Minimalist apartment with two dedicated monitors for benchmarking. Cycles everywhere. Cooks elaborate meals on Sundays as a counterbalance to screen time. Reads technical whitepapers for fun.. Personality: Rigorous, Deadpan, Introverted but expressive on camera, Allergic to hype, Quietly enthusiastic about data. Elias was a backend developer at a Stockholm startup before relocating to Berlin for a role at a generative AI research lab. Outside work, he builds side projects, and AI video tooling became one of them when he noticed nobody was publishing rigorous cost-per-output benchmarks. He approaches every reel the way he'd approach a pull request: define the inputs, measure the outputs, call out the edge cases. His channel is the driest on this list and somehow the most watchable because the numbers are always right. Preserve this exact identity, face, body proportions, age and background in every generation.", "biography_context": "Elias was a backend developer at a Stockholm startup before relocating to Berlin for a role at a generative AI research lab. Outside work, he builds side projects, and AI video tooling became one of them when he noticed nobody was publishing rigorous cost-per-output benchmarks. He approaches every reel the way he'd approach a pull request: define the inputs, measure the outputs, call out the edge cases. His channel is the driest on this list and somehow the most watchable because the numbers are always right." } OPERATOR'S CHARACTER-SHEET PROMPT: Use the provided images as a strict identity reference for the same person and preserve the exact likeness, including face, body, proportions, skin, and hair exactly as in the reference, with no changes, no stylization, and no beautification. Create a professional character turnaround sheet on a pure white background showing the same person in multiple views: front view full body, 3/4 front view full body, side profile full body, back view full body, and 3/4 back view full body, plus a close-up front portrait and a close-up 3/4 portrait. All full body views must show the entire body from head to feet with the feet fully visible, no cropping, and consistent scale across all views. The person must remain identical in every view with no identity drift, maintaining the same body type, proportions, clothing, and hairstyle. Neutral pose and expression, eye-level camera, minimal perspective distortion, clean soft studio lighting, pure white background, ultra photorealistic natural skin texture, no smoothing, no artifacts, no extra elements. 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.