1枚の画像を2分割し、左側にはリアル、右側にはその似顔絵を生成したいと思います。
画像生成AIプロンプト作成用My Gemを使ってプロンプトを作成します。
テーマを与えると、アンケートシートが返ってきたのでそれに回答しました。
いただいた回答：
被写体 (Subject): 女性、20代
動作と関係性: 正面向き、微笑んでいる
設定 (Setting): 自然（背景）、昼
スタイルと媒体: 左は超リアル、右は水彩
構図とカメラ: バストアップ
照明と色: （未回答：昼の自然光を想定）
品質と詳細: （未回答：超リアルのための詳細さと、水彩のテクスチャを想定）
Nano Banana 2 Lite
/imagine prompt: A vertically divided single image. The left half shows an ultra-realistic, highly detailed photograph of a smiling Japanese woman in her 20s, facing forward, from the bust up. The background on the left is a soft, sunlit natural setting with blurred green foliage (bokeh), shot with a 35mm lens, high resolution, photorealistic skin texture, natural daylight. The right half shows a delicate watercolor illustration of the same woman with the same smile and pose, created with wet-on-wet watercolor techniques on textured paper, using natural hues of green, blue, and warm tones, with visible paint drips and soft edges, maintaining the natural setting background in an artistic style. The lighting is soft natural daylight, consistent across both panels. The two halves create a harmonious comparison of realism and art, side by side. –ar 16:9 –style raw
右側のスタイルだけ変更して、次々に生成します。
/imagine prompt: A vertically divided single image, featuring the same 20-year-old Japanese woman facing forward, smiling, from the bust up, in a natural daylight setting. THE LEFT HALF shows an ultra-realistic, highly detailed photograph of the woman, capturing fine skin texture, individual hairs, and natural eye glint. She wears a simple casual top. The background is a soft, sunlit natural forest with diffused bokeh (blurred greens and light), shot with a 35mm lens, high resolution, photorealistic. THE RIGHT HALF shows a bold, graphic Pop Art style illustration of the same woman. This panel features highly saturated, vibrant, flat colors with black outlines. Her face is rendered with distinct blocks of color and halftone dot patterns. The smiling expression is captured with dynamic lines. The background in this half is simplified into stylized, repeating graphic patterns of leaves and sunbursts in strong, clashing colors (like electric blue, bright yellow, hot pink, and emerald green), reminiscent of Lichtenstein and Warhol, with a clean, vector-like quality. The natural daylight from the left panel is reinterpreted into bold graphic highlights in the right panel. The composition is balanced, with a sharp vertical line dividing the realistic world on the left from the vibrant, graphic art on the right. –ar 16:9 –style raw –niji 6
/imagine prompt: A vertically divided single image, featuring the same Japanese woman in her 20s, facing forward and smiling. The composition is a bust-up shot. THE LEFT HALF shows an ultra-realistic, highly detailed photograph of the woman, with photorealistic skin texture, natural eye glint, and individual hairs visible. She has a natural smile and wears a simple casual top. The background on the left is a soft, sunlit natural setting with blurred green foliage (bokeh), shot with a 35mm lens, high resolution, capturing natural daylight and realistic colors. THE RIGHT HALF shows a professional, clean anime-style illustration of the same woman with the identical smile and pose. This panel features clean, defined black outlines and flat, yet vibrant anime colors. Her eyes have classic large, expressive anime highlights. The background on the right is a stylized, clean anime interpretation of the same natural setting, with simple shapes and cheerful, saturated colors (clean blues, soft greens), without heavy texture. The lighting is soft and natural across both panels, reinterpreted to suit each medium. The line between the two panels is sharp and precise. –ar 16:9 –style raw –v 6.0
/imagine prompt: A vertically divided single image, creating a side-by-side comparison. THE LEFT HALF shows an ultra-realistic, highly detailed photograph of [INSERT SUBJECT HERE, e.g., a Japanese woman in her 20s with long black hair], facing forward and smiling, bust-up shot. The background on the left is a soft, sunlit natural setting with blurred green foliage and realistic bokeh, captured with a 35mm lens, photorealistic skin texture, natural daylight. THE RIGHT HALF shows a hand-drawn marker illustration (Copic marker style) of the exact same person with the identical smile, pose, and framing. This right panel features visible marker stroke textures, vibrant alcohol-ink blending, soft paper grain, and expressive ink line art. The background on the right is a stylized marker sketch of the same natural setting with loose artistic coloring. The dividing line between the photograph and the marker drawing is sharp and clean. –ar 16:9 –style raw –v 6.0
/imagine prompt: A vertically divided single image, creating a side-by-side comparison. THE LEFT HALF shows an ultra-realistic, highly detailed photograph of a Japanese woman in her 20s, facing forward and smiling, bust-up shot. The background on the left is a soft, sunlit natural setting with blurred green foliage and realistic bokeh, captured with a 35mm lens, photorealistic skin texture, natural daylight. THE RIGHT HALF shows a detailed hand-drawn graphite pencil sketch of the exact same woman with the identical smile and pose. This right panel features visible pencil hatching, delicate cross-hatching shading, fine line work, graphite texture, and soft paper grain. The background on the right is a subtle, stylized pencil sketch of the same natural setting. The dividing line between the photograph and the pencil drawing is sharp and clean. –ar 16:9 –style raw –v 6.0
/imagine prompt: A vertically divided single image, creating a strict side-by-side comparison. THE LEFT HALF shows an ultra-realistic, highly detailed photograph of a Japanese woman in her 20s, facing forward and smiling, bust-up shot. The background on the left is a soft, sunlit natural setting with blurred green foliage and realistic bokeh, captured with a 35mm lens, photorealistic skin texture, natural daylight. THE RIGHT HALF shows a dynamic, bold American Comic Book style illustration of the exact same woman with the identical smile, pose, and framing. This right panel features heavy black ink outlines, dramatic cel-shading, visible halftone dot patterns for texture, and highly saturated colors. The background on the right is a stylized, graphic art interpretation of the natural setting with dynamic lines. The line between the two panels is sharp and clean. –ar 16:9 –style raw –v 6.0
/imagine prompt: A vertically divided single image, creating a strict side-by-side comparison. THE LEFT HALF shows an ultra-realistic, highly detailed photograph of a Japanese woman in her 20s, facing forward and smiling, bust-up shot. The background on the left is a soft, sunlit natural setting with blurred green foliage and realistic bokeh, captured with a 35mm lens, photorealistic skin texture, natural daylight. THE RIGHT HALF shows a traditional Japanese Ukiyo-e woodblock print style illustration of the exact same woman with the identical smile and pose. This right panel features delicate, flowing black ink outlines (sumi-e style), flat color fields (benizuri-e style) using traditional pigments like indigo and vermilion, and a textured washi paper finish. The background on the right is a stylized, classical Ukiyo-e interpretation of the natural setting with decorative clouds and patterns. The line between the two panels is sharp and precise. –ar 16:9 –style raw –v 6.0
似顔絵
普通なら、リアル写真をアップロードして、「似顔絵+スタイル指定」で似顔絵を作成する使い方をすると思います。しかし、最近、Googleの規約変更が何かと話題となっているように、これからは個人情報を特定される恐れがあるものは、安易にAI界で使わない方が良いようです。新しい規約では、マルチモーダル化されたAIが、ユーザーのアップロードした画像、動画、音声、ファイルなどを学習に使用する可能性があるとされているようです。
似顔絵のスタイル指定簡単編
プロンプトの「THE RIGHT HALF」以下の赤マーカー部分がスタイル指定になっています。shows以下の部分だけ使えば、任意の1枚の画像からお好みのスタイルの似顔絵を作成できます。
似顔絵のスタイル指定応用編
プロンプトのshows以下の部分を生成AIにコピペします。そして、「AI画像生成ツールのプロンプト（呪文）として最適化・アレンジしてください。」とすれば、複数の詳細なプロンプトの提案をしてくれます。それを使えば、簡単編より高度な似顔絵を作成できます。
※注意点としては、アメコミ風、サイバーパンク風などにするとテキストが入るので、あらかじめテキスト除外指定が必要です。
コメント