用 Nano Banana 2 或 ChatGPT 圖片生成，自己在家就能把照片轉成似顏繪與多種插畫風格。下面整理了 15 種風格的現成 prompt，複製、連同照片貼上就能用。
這兩款是目前把「照片轉插畫」做得最好、又最容易上手的選擇，台灣都能用。挑一個順手的開始即可。
Google 最新推出的進階圖片生成模型，是廣受好評的 Nano Banana 系列再進化，比前一代 Nano Banana Pro 更強，人物相似度與細節表現更上一層樓。
OpenAI 內建在 ChatGPT 裡的圖片生成功能（一般俗稱 GPT Image，最新為 ChatGPT Images 2.0），會先理解你的指令再生成。
其他像 Midjourney、即夢、Adobe Firefly 等工具也能做，prompt 概念相通，下面的範例都適用。
完全不用會畫畫，照著做就行。重點在第二步的 prompt 與第三步的「多生幾張再挑」。
挑一個你順手的 AI 生圖工具（如 ChatGPT、Gemini 等），上傳一張想轉成插畫的照片。原圖畫質越清楚，生成結果越穩定。
從下方風格庫複製想要的 prompt，連同照片一起貼給 AI，再補上人數、背景、氛圍等描述。第一版不滿意很正常，逐句調整是常態。
同一組 prompt 多跑幾次，每次結果都會略有不同。一次生成 3～4 張，再挑出五官最像、構圖最好的版本，比死磕一張更有效率。
選定滿意的成品後，用工具的「高解析度／放大／2K」功能輸出，下載解析度足夠的檔案，就能進一步印製成實體作品收藏或送禮。
想讓成果一眼就認得出是誰，關鍵在 prompt 的「特徵鎖定」。除了描述風格，請主動寫出本人的辨識重點，例如：「保留原照片的臉型、五官比例、髮型與髮色、膚色與眼鏡，維持人物神韻」。臉部是最容易被 AI「改造」的地方，描述得越具體、越貼近原貌，結果就越像本人。若第一版臉變了，可在下一輪補上「臉部更貼近上傳照片」並重新生成，通常幾輪內就能調到滿意。
點「複製 Prompt」把整段指令複製起來，連同你的照片一起貼給 AI 即可。中文小提示說明可以怎麼微調，挑你喜歡的風格開始玩。
請用蠟筆畫風格把整張圖創作成一幅圖畫。簡化細節、使其看起來像是 10 歲孩子畫的。不要使用原來圖片的顏色。讓整體看起來是在一張白紙上作畫的感覺，非常可愛的感覺，可以加一些可愛的元素比如花朵、糖果、星星、雲朵等看起來很童真。
生成一張「我」的插圖、日系插畫風、角色特徵鮮明、情緒自然表情、半身構圖、動態姿態、服裝細節精緻、手繪塗鴉風、潑墨筆觸、隨性線條、粉彩與墨色混合、漫畫草稿質感、白色簡約背景、周圍符號化元素點綴、氛圍感強、高細節、高品質。
請觀察我上傳的照片內容，保留原本照片的主體、構圖、光線與氛圍，不要大幅改變原圖，只在照片上加入自然、有意義的手繪風註解。
【目標風格】
請將照片加工成「手繪註解風、文青手帳風、IG 限時動態風」圖片。
整體要像日系生活雜誌、旅行手帳、照片日記上的隨手筆記。
【畫面內容】
請根據照片自行判斷主要物件。
可能包含：美食、飲料、人物、穿搭、飯店房間、旅行風景、街景、建築、天空、樹木、水面、桌面小物、展覽空間等。
【描寫規則】
・使用白色細手繪線條
・線條要像一筆畫，隨性、自然、略帶不均勻感
・可以沿著物件外圍加上簡單描邊
・可以使用箭頭、虛線、圈圈做視線引導
・可以加入少量星星、愛心、小表情、陽光、波浪線等裝飾
・不要讓畫面太滿，請保留留白與呼吸感
【文字規則】
・文字請使用繁體中文
・文字要像手寫字，偏日系、可愛、自然
・句子要短，像自言自語的小碎念
・語氣偏日記感、生活感、情緒感
・不要使用太正式、太商業、太像廣告的文字
・每句文字盡量控制在 6～14 個中文字內
・整張圖的文字不要太多，約 4～7 句即可
【註解生成規則】
・飲料：描述味道、溫度、心情，例如「冰冰的剛剛好」「微甜很舒服」
・食物：描述口感、香氣、好吃程度，例如「外酥內軟」「今天最喜歡這個」
・人物：描述穿搭、姿勢、光線、氛圍，例如「今日份鬆弛感」「這個角度好喜歡」
・房間：描述空間、床、光線、行李、色調，例如「一進來就放鬆」「剛到就想躺下」
・風景：描述天空、陽光、空氣感、城市節奏，例如「今天的天空很舒服」「好適合慢慢走」
・街景：描述散步感、日常感、安靜感，例如「小小街角」「簡單，剛剛好」
・整體：加上一句總結，例如「想把這一刻存起來」「今天有點幸福」
【請避免】
・不要做成商業 Banner
・不要做成正式海報
・不要過度排版
・不要太整齊
・不要讓文字遮住主體
・不要大幅改變原本照片
・不要讓人物臉部變形
・不要出現錯字或簡體字
・不要加入過多元素
【完成效果】
請生成一張自然、有質感、帶一點慵懶感的手繪註解風圖片。
整體要像「隨手寫在照片上的生活筆記」。
4K 高解析。俯視角，女主全身優雅側躺在花海中。背景是白玫瑰、粉色花朵、滿天星交錯的立體花海。穿白色長洋裝，裙襬自然鋪散。柔和自然光，高級婚紗雜誌質感。保留上傳照片的五官特徵，自然精緻化但不改變辨識度，不要尖下巴、V型臉、網紅臉、混血臉，保留細微毛孔、皮膚紋理，不過度磨皮。加入藝術花卉線描裝飾能提升畫面層次（金色線描 + 發光邊線）
Transform the attached photo into a Japanese-style 似顔絵 caricature portrait: a cute, friendly likeness with clean confident outlines, flat cel-style coloring, and gently exaggerated yet flattering proportions (slightly larger head, big sparkling eyes, a warm smile). Faithfully preserve this exact person's identifying features, hairstyle, hair color, skin tone, glasses, and outfit so they are instantly recognizable. Use a simple solid pastel or soft-gradient background with no clutter, keep the framing as a head-and-shoulders bust shot. Render in crisp high resolution, portrait orientation, suitable for a profile avatar or sticker.
Reimagine the attached photo as a hand-drawn Studio Ghibli-esque anime illustration: soft hand-painted watercolor backgrounds, warm natural lighting, gentle rounded line art, and that nostalgic Miyazaki storybook atmosphere. Keep the person's real facial features, expression, hairstyle, and clothing recognizable, rendered with the simple expressive eyes, soft cheeks, and delicate shading typical of the style. Place them in a serene scene such as a sunlit meadow, a cozy countryside window, or a breezy hillside with fluffy painted clouds. Deliver in high resolution, portrait orientation, with a dreamy hand-illustrated finish.
Convert the person in the attached photo into a Pixar/Disney-style 3D animated character: smooth stylized 3D render with soft subsurface-scattering skin, large expressive glossy eyes, rounded friendly features, and cinematic studio lighting with gentle rim light. Preserve this person's true likeness, hair, skin tone, and clothing so the 3D character clearly resembles them, with a warm cheerful expression. Use a softly blurred, tastefully lit background with shallow depth of field, like a frame from an animated feature film. Output ultra high resolution, portrait orientation, with polished cartoon-realistic 3D quality.
Turn the attached photo into a modern Japanese anime illustration with clean cel-shading: sharp confident line art, bright saturated flat colors with crisp shadow blocks, glossy detailed anime eyes, and a polished contemporary look like a current TV anime key visual. Keep the person's facial structure, hairstyle, hair color, and outfit faithful and recognizable while stylizing them into anime form. Add subtle highlights in the hair and a clean, vibrant background such as a soft bokeh sky, school rooftop, or city street at golden hour. Render in high resolution, portrait orientation, with sharp professional anime finish.
Paint the attached photo as a delicate hand-painted watercolor portrait: soft translucent washes, visible paper texture, gentle color bleeds and blooms, loose expressive brushwork around the edges, and a light airy palette. Preserve the person's real likeness, expression, hairstyle, and clothing, rendered with soft painterly edges rather than hard outlines so it feels authentically hand-made. Keep the background as loose abstract watercolor washes or a few suggestive botanical strokes, leaving plenty of clean white paper space. Deliver in high resolution, portrait orientation, with a fine-art watercolor feel.
Illustrate the attached photo in a soft Korean healing illustration style: minimal clean shapes, muted pastel palette, gentle flat coloring with very subtle soft shading, simple dot-and-line facial features, and a calm cozy mood. Keep the person recognizable through their hairstyle, hair color, skin tone, and outfit, simplified into a cute soft character with rosy cheeks and a serene little expression. Use a clean minimalist background in a single soft pastel tone with at most one small decorative element (a plant, mug, or tiny heart). Render in high resolution, portrait orientation, with a tidy modern Korean-illustration aesthetic.
Convert the attached photo into a fine hand-drawn graphite pencil sketch on textured off-white drawing paper. Preserve the person's exact likeness, facial structure, hairstyle, expression and clothing so they remain fully recognizable. Render entirely in monochrome graphite: soft tonal shading built from delicate cross-hatching and fine pencil strokes, realistic light and shadow on the face, crisp darker outlines and lighter, looser construction lines toward the edges that fade into the bare paper. Keep the background minimal — a subtle vignette of sketch strokes or clean white. Produce a high-resolution, detailed portrait sketch.
Reimagine the attached photo as a warm, hand-drawn children's storybook illustration in a naive crayon / soft pastel style. Preserve the person's recognizable likeness, hairstyle, expression and clothing, but simplify them with gentle rounded shapes, slightly oversized friendly eyes and a cheerful storybook charm. Use visible waxy crayon and colored-pencil texture, warm cozy colors, soft uneven outlines and a little white paper showing through. Place them in a sweet, simple illustrated scene (a sunny field, cozy room or starry sky) with hand-drawn decorative details. Render as a high-resolution, heart-warming vertical picture-book page.
Turn the person in the attached photo into a cute chibi-style LINE messenger sticker. Preserve their recognizable features — same hairstyle, hair color, glasses, signature clothing colors and expression — but redraw in adorable super-deformed proportions with a big oversized head, tiny body, large sparkly eyes and a clean flat-color anime look. Use a bold even outline around the whole character, soft cel shading, and a cheerful exaggerated pose or emotion (waving, thumbs-up or happy). Output the character as a single sticker on a fully transparent (or plain white) background with a subtle thin white sticker border, high-resolution and centered.
Render the attached photo as a classical oil-painting portrait in the tradition of Renaissance and Baroque masters such as Rembrandt and Vermeer. Preserve the exact likeness, facial features, hairstyle, expression and clothing of the subject — they must remain clearly recognizable. Use rich visible oil brushwork, warm chiaroscuro lighting with a dramatic single light source, deep luminous skin tones, subtle glazing and fine canvas texture. Place them against a dark, softly graded museum-style background. Compose as a dignified, high-resolution vertical bust portrait with the gravitas of an old-master gallery painting.
Transform the attached photo into a cinematic cyberpunk neon portrait set in a futuristic night city. Preserve the person's exact likeness, facial features, hairstyle, expression and clothing, keeping them clearly recognizable. Light the face with glowing magenta-and-cyan neon rim light, vivid reflections and subtle holographic accents, against a dark rainy background of blurred neon signs and bokeh city lights. Add a moody high-contrast color grade, faint atmospheric haze and a few sci-fi tech details (glowing lines or HUD glints) without hiding the face. Render as a sharp, high-resolution, dramatic vertical portrait.
同樣一張照片，懂這幾點就能少走很多冤枉路、結果更像也更耐看。
五官清楚、無濃重陰影或側臉死角的照片，AI 最容易抓住特徵，成果也最像本人。
加上「保留原本臉部特徵、髮型與膚色」這類措辭，能大幅降低 AI 把人畫成另一個人的機率。
明確指定「單人／雙人／一貓一狗」「室內咖啡廳／海邊夕陽」等細節，避免 AI 自行加減人物或亂配背景。
不要只跑一次就定生死。多生幾張、橫向比較，挑出最像也最耐看的一張，品質落差往往很大。
生成或後製時加上高解析、放大、2K 等指令，像素夠多，放大印製才不會糊掉或出現鋸齒。
若照片含他人或公眾人物，請先確認已取得同意；部分 AI 工具對「真實人臉」也有使用規範，商用前務必確認授權與政策。
辛苦調好的 AI 插畫，留在手機相簿裡太可惜。phootime 以 11 色藝術微噴技術，把你的數位作品輸出成可以掛上牆、能傳家的實體無框畫與周邊商品——讓它從一張圖，變成生活裡真正的藝術品。
印成藝術微噴作品不管是 AI 生成的插畫，還是你早就準備好的檔案，都能直接在 phootime 印製成無框畫與商品。
我有插畫檔案了，要訂製商品本頁為教學分享，phootime 提供的是印製服務而非代繪服務。AI 生成結果會因模型、版本與 prompt 不同而有所差異，使用 AI 工具請遵守各平台當下的功能與政策，含真人肖像時請留意肖像權。
