Prompt: Transform the input character into a vertical side-by-side "art vs artist" image, split down the middle by a thin clean seam. The SAME character appears on both halves — drawn, then embodied.

LEFT — the art: the input character as a flat 2D anime illustration (clean lineart, cel-shading, saturated anime palette), faithful to the input — exact face, eye color and shape, hairstyle, hair color, bangs, outfit, and signature accessories. Keep it a flat drawing; do NOT photo-realize. Chest-up, facing forward or three-quarter.

RIGHT — the artist: a real photoreal person cosplaying that same character, taking a MIRROR SELFIE with a smartphone raised in one hand that FULLY covers their face (back of the phone toward the viewer) — no real face visible at all. Accurate cosplay: real hair matching the character's exact color and style (wig or dyed, keeping twin buns as twin buns, twintails as twintails, including streaks/gradients), and a faithful real-fabric version of the outfit with its signature accessories. Mirror the art half's pose and chest-up framing so the two sides line up across the seam.

Right-half photo: a candid smartphone mirror selfie — natural indoor light, true-to-life color, slight handheld softness; a cozy artist's bedroom (mirror, desk, posters) by default.

Layout: two equal vertical halves, thin neutral divider, the same framing scale so the character aligns across the seam. Optional tiny clean "ART" / "IRL" labels only if they stay crisp.

Avoid: photo-realizing the left (art) half (it stays flat 2D); ANY visible real face on the right half (the phone must fully hide it); cosplay hair or outfit not matching the character; a floating phone with no hand; warped or extra fingers; garbled labels; mismatched characters between halves; watermarks; landscape orientation. In bed, wearing gray cargo pants, black fishnet top, fit girl, athletic, at the gym, gym mirror, Model: GPT Image 2 v1