GPT Image-2 - Text to Image

プロンプト

Use the attached reference image as the character reference. Analyze and extract the person’s main visual features: hairstyle, hair length, hair color, bangs, eye shape, eye color, facial impression, outfit, accessories, body proportions, and overall vibe. Keep the person recognizable, but redraw them as an intentionally charming messy doodle collage character sheet. ultra messy doodle collage, intentionally bad but cute drawing, chaotic sketchbook page, MS Paint mouse drawing aesthetic, rough scribbles, ugly-cute, charming amateur doodle, white background, low quality hand-drawn feeling, uneven anatomy, childish lineart, goofy proportions, awkward but adorable composition, pixel-ish roughness, messy coloring outside the lines, loose sketch strokes, ballpoint pen texture, colored pen sketch, notebook doodle vibe multiple drawings of the same person across the canvas, character reference sheet, expression sheet, pose sheet, sticker sheet, doodle collage layout, many small variations of the same character, large main portrait, full body standing pose, sitting pose, pointing at viewer, peace sign pose, close-up face, sleepy face, crying face, angry face, smug face, embarrassed face, chibi face, tiny mascot-like face, looking down at camera, dynamic perspective convert the reference person into a cute anime-style doodle character, keep their distinctive features and fashion identity, soft blush, expressive eyes, simple cute face, rounded cheeks, slightly exaggerated cute proportions, small hands, simple legs, loose clothing folds, adorable imperfect sketch feeling scribbled handwritten text random doodles around the character: stars, hearts, arrows, speech bubbles, question marks, exclamation marks, barcode, tiny notes, emoji faces, “owo”, “???”, abstract scribbles, small mascot animal, messy symbols, color swatches color palette based on the reference image, limited to 35 main colors, rough marker coloring, sketchy hatching, visible pen strokes, white empty background with scattered doodles, cute chaotic energy, charming messy collage, playful personality

image

GPT Image-2 - Text to Image

GPT Image-2image:gpt-image-2-text-to-image画像生成
2026年5月23日 20:58

設定

プロバイダー
GPT Image-2
モデル
image:gpt-image-2-text-to-image
作成時間
2026年5月23日 20:58
クレジット
3
モード
画像生成
アスペクト比
auto
解像度
1K

この作品をもとに生成する

下のジェネレーターはこの記録で使われたモデルを自動選択するので、すぐに試せます。

画像を生成
送信したタスクはここに表示されます。