A detailed image-to-image prompt for GPT Image 2 designed to generate a realistic social media snapshot of an anime character on a Japanese commuter train, emphasizing character consistency and authentic environment.
Use the character from the reference image as the base design to draw "a single frame of a short video accidentally taken by a regular person on a train, showing an anime character living a normal life in the real world." Accurately maintain the atmosphere of the character as an anime illustration. Preserve the impression of the face, hairstyle, hair color, eyes, ears, body proportions, silhouette, and facial expression. Never break the character's identity. Do not turn the character into a photorealistic human. Maintain high-quality anime illustration expression for the character part. The setting is the interior of a typical Japanese commuter train. Realistic, photorealistic background depiction. Naturally depict advertisements, windows, handrails, straps, seats, doors, and guide displays. Do not make the space look like a clean advertising photo. Give it a sense of everyday life and use typical of real public transportation. The composition is a vertical short video. A screenshot of an SNS video taken with a smartphone. The photographer is sitting near the opposite seat. The character is sitting on the long seat opposite. A natural angle seen slightly from the side, not directly from the front. The character is looking at a smartphone. Not looking at the camera. Not posing. Not aware of the photographer. A bag, paper bag, or shopping bag that the character might have is placed at the character's feet. An atmosphere like coming home from work or shopping. They can be wearing earphones. They can have a tired or sleepy expression. They are just normally in transit. Surrounding passengers exist naturally. No one is reacting specially to the character. Depict the surrounding world as a mysterious everyday life where people are "used to the existence of anime characters." Add a modern atmosphere similar to TikTok, Instagram Reels, or YouTube Shorts to the screen. Not a LIVE broadcast screen. Naturally display the number of likes, comment section, share button, etc. Do not fix the comment content; let the AI generate it naturally. Comments are mainly in Japanese. Important: The comment section should not treat the character like a celebrity. No flaming or news treatment. It should have a natural temperature, like seeing an urban legend or a rare wild animal. Emphasize textures typical of smartphone photography: light HDR processing, fine sensor noise, slight JPEG compression feel, natural exposure, slight chromatic aberration, autofocus feel, high ISO noise, and fine shakes typical of handheld shooting. Reality like a frame cut from a short video. Prohibit movie poster style. Prohibit cinematic direction. Prohibit dramatic lighting. Prohibit excessive depth of field effects. Do not overload with information. Prohibit advertising photo style. Avoid AI-specific glossy rendering. Prioritize the strange reality of "a mysterious video that accidentally flowed through social media," "for some reason no one is surprised," and "only the person themselves is living normally."
