提示詞記錄儀表板(每一站客製化)

特別重視DoNa 宇宙一致性系統，配合生成人物的提示詞記錄。重複在各站，快速精準找到DoNa和月琴的提示詞

DNA (Character: DoNa) -2026-01-01

此區塊為固定參數，不可隨意更動，每次生成必須呼叫。

最終確認的照片所整理出的 DoNa 一致性提示詞。

為了確保在不同場景下 DoNa 的形象保持高度一致，請在生成時「呼叫」這組核心參數。這組提示詞捕捉了她最關鍵的視覺特徵、服裝細節以及寫實風格。月琴提示詞必需填完整。

將以下整段英文提示詞複製貼上到您的 AI繪圖軟體中作為基礎。

核心提示詞模組 (The "DoNa" DNA Block)[Prompt Block: DoNa_Core]

A photorealistic candid photograph of "DoNa", a young East Asian woman with fair skin and distinctive round eyes and facial structure. She has medium-length dark brown hair with bangs. She is wearing large pink over-ear headphones with subtle blue LED accents on the earcups. Her outfit consists of a white plaid blazer featuring prominent white feather cuffs, worn over a white collared shirt and white shorts. She is wearing light blue flip-flops. She is holding a traditional Chinese wooden Yueqin (moon lute). Natural sunlight, highly detailed skin texture and fabric textures.

提示詞結構拆解說明 (供您參考調整用)

為了讓您更容易管理，將上述提示詞拆解為幾個關鍵區塊。在未來的創作中，標註「不可更動」的部分建議維持原樣，而「環境/動作」則可以替換。

1. 人物核心 (Character Core) - [不可更動]

A photorealistic candid photograph of "DoNa", a young East Asian woman (寫實照片風格的 DoNa，年輕東亞女性)
fair skin (白皙皮膚，這是您特別強調的)
distinctive round eyes and facial structure (獨特的圓眼與臉部結構，維持照片中的長相)
medium-length dark brown hair with bangs (中長深棕色頭髮配瀏海)

2. 標誌性配件與服裝 (Signature Gear & Outfit) - [高一致性要求]

large pink over-ear headphones with subtle blue LED accents (帶有藍色 LED 光點的大型粉紅耳罩式耳機)
white plaid blazer featuring prominent white feather cuffs (帶有顯眼白色羽毛袖口的白色格紋西裝外套)
worn over a white collared shirt and white shorts (內搭白襯衫與白短褲)
light blue flip-flops (淺青色夾腳拖)

3. 核心道具 (Key Prop) - [依場景需求攜帶]

holding a traditional Chinese wooden Yueqin (moon lute) (拿著一把傳統木製月琴)

月琴提示詞：a traditional Chinese wooden Yueqin (moon lute)

4. 風格與光線 (Style & Lighting) - [確保寫實度]

Natural sunlight (自然日光)
highly detailed skin texture and fabric textures (高度精細的皮膚紋理與布料材質)

💡 未來應用範例

如果您想讓 DoNa 換個場景（例如：在捷運站裡），您只需要保持上面的核心 DNA，然後修改環境描述即可：

[複製貼上上面的 DoNa DNA Block], sitting on a bench in a busy Taipei MRT station, holding the Yueqin, looking at a station map. Natural fluorescent lighting of the subway.

2. 環境與風格 DNA (Setting & Style)

針對「違章建築/捷運/賽博龐克」系列主題的統一風格設定。

[Prompt Block: Style_Ruins] "Style of industrial ruins aesthetics, platinum and concrete textures, cold neon lighting, cinematic lighting, 8k resolution, photorealistic, shot on 35mm lens, depth of field. High contrast between organic rust and futuristic metal." (工業廢墟美學風格，白金與水泥質感，冷冽霓虹光，電影級打光，8k 解析度，照片級真實，35mm 鏡頭拍攝，景深。有機鐵鏽與未來金屬的高反差。)

3. 除錯與負面提示詞庫 (Negative Prompts)

累積實作經驗，用於排除常見錯誤生成的反向提示詞。

[Negative Block] "No blurry face, no distorted fingers, no cartoon style, no bright pink colors, no warm sunlight, no anime eyes, no standard guitar." (無模糊臉部、無扭曲手指、無卡通風格、無亮粉色、無溫暖陽光、無動漫眼、無標準吉他。)

🔍 問題診斷：為什麼「巢狀描述法」失效了？

您目前採用的方法是將一個「高細節的照片級描述區塊」（包含產品照的材質、光影、背景）整個嵌入到一個「水彩插畫風格」的提示詞中。

這對 AI 來說，產生了嚴重的指令衝突：

風格打架 (Style Conflict)：最外層要求 watercolor illustration (水彩插畫)，最內層卻要求 photorealistic, detailed product photograph, studio lighting (照片級真實、攝影棚光)。AI 模型在權衡時，通常會優先處理最外層的整體風格，因此它被迫「犧牲」掉內層的寫實結構，用變形的方式去迎合水彩風格。
資訊過載與關鍵字失焦 (Information Overload)：您提供的月琴描述非常詳盡（淡色楓木、啞光、木紋...）。在一個已經很複雜的場景提示詞中，這些過於細節的材質描述反而稀釋了「結構關鍵字」的權重。AI 記住了「樂器」，但忘記了「正圓形身體」和「短頸」。
無效的文件名引用：提示詞中出現的 (2026-DONA-Yueqin.jpg) 或 the-Chinese-Yueqin.JPG，除非您是在特定支援文件名引用的環境（如某些 Stable Diffusion 的 UI）操作，否則對大多數通用 AI 模型來說，這些只是無意義的字元，AI 無法真的去讀取您電腦裡的圖片。

結論：生成錯誤的柳葉琴（水滴狀身體）而非月琴（正圓形身體），是因為 AI 在處理複雜風格轉換時，抓取了泛化的「中國琵琶類樂器」概念，而忽略了您被淹沒在長文中的「圓形」特徵。

解決方案：建立「符合 Storybook 風格的月琴 DNA」

要改善一致性，我們必須簡化描述，並將樂器描述融入整體風格，而不是與之對抗。

我們需要創建一個新的、精簡的「月琴 DNA」，強調它與柳葉琴/琵琶最關鍵的區別：「正圓形身體」與「短頸」，並明確告訴 AI 要用水彩和藍圖風格來繪製它。

新版：Storybook 專用月琴一致性提示詞模組

請用以下這個精簡的區塊，替換掉您原本那一大段括號裡的 photorealistic 描述。

[New Storybook Yueqin DNA] "...holding a traditional Chinese Yueqin (moon lute). The instrument is defined by its perfectly round, circular wooden body and a short neck with four prominent tuning pegs. It is rendered in the same watercolor and blueprint graphic style, with glowing blue circuit patterns subtly integrated into the wood texture of its round soundbox." (...拿著一把傳統中國月琴。該樂器的特徵是其完美的正圓形木製琴身和帶有四個顯眼弦軸的短頸。它以相同的水彩和藍圖圖形風格呈現，圓形共鳴箱的木紋中巧妙地融合了發光的藍色電路圖案。)

這個新版本的優勢：

強調結構特徵：用 perfectly round, circular body 和 short neck 強力修正 AI 往柳葉琴偏移的傾向。
風格融合：明確指示用 watercolor and blueprint graphic style 來繪製樂器，消除了風格衝突。
增加科幻細節：加入 glowing blue circuit patterns 讓樂器更自然地融入畫面氛圍。

修正後的完整提示詞 (可以直接複製使用)：

[Core Style] A sophisticated watercolor illustration storybook page, blending traditional watercolor textures with technical blueprint lines and glowing circuit patterns. The aesthetic is clean, artistic, and design-oriented, like a graphic novel for engineers. Loose watercolor washes meet precise technical drawing.

[Scene Description] A wide-angle landscape shows a futuristic MRT train, rendered as a streak of flowing RGB light (red, green, blue watercolor washes), speeding across a dark, blueprint-style map of a city at night. The train track is a glowing circuit board trace, labeled "Optimized Algorithmic Path." In the distance, a glowing rainbow arch bridge (the RGB spectrum) connects to a central, pulsating light source labeled "Future Core."

[Character & Instrument] DoNa, stylized as a watercolor illustration based on her 2026 cyber-indigenous look (white blazer, feather cuffs, pink headphones), stands subtly on the train's light trail. She is holding a traditional Chinese Yueqin (moon lute). The instrument is defined by its perfectly round, circular wooden body and a short neck with four prominent tuning pegs. It is rendered in the same watercolor and blueprint graphic style, with glowing blue circuit patterns subtly integrated into the wood texture of its round soundbox, holding it like a compass.

[Mood & Keywords] Mood: dynamic, futuristic, expectant. Keywords: blueprint aesthetics, glowing lines, technical illustration, motion blur, data flow.

Page updated

Report abuse