跳到主要內容
金瓶梅宇宙JIN PING MEI UNIVERSE

角色檔案名冊其十

孫雪娥

雪娥、四娘、孫娥

要角 · 西門慶的第四房妾室,善理湯水並主管廚房,長期捲在妻妾、僕婦與來旺的衝突之中。

孫雪娥三視圖角色設定(AI 生成,非真人照片)
三視圖角色設定・AI 擬真攝影風(非真人照片)・以原典描寫為依據
性別
年齡
中年(推斷)
身份
西門慶第四房妾室,原為房裡人,主管廚房與僕婦
性格
強硬(推斷) / 直率(推斷) / 記仇(推斷) / 畏禍(推斷)

外貌

身量偏矮,體態結實,帶有常年料理廚務的俐落感;衣著較家中其他妻妾樸實,袖口收束便於做事(推斷)。

性情

平日務實幹練,受刺激時嗓門高、反應急,敢當面掌摑扭打;事到追責關頭又會迅速轉為驚惶求情(推斷)。

動機

守住自己在內宅與廚房中的位置,並在長期受挑撥、羞辱與責打後尋求報復與自保(推斷)。

人物弧光

由掌理廚務、參與內宅日常的妾室,逐步陷入與潘金蓮、宋惠蓮及來旺的衝突;後與來旺私逃,輾轉被賣,終因懼怕案發而自縊。

人物關係

原文依據

以下逐字引自《金瓶梅詞話》(萬曆本)原典,是這份角色側寫的文本根據。

視覺版本

同一個角色的其他視覺路線:擬真攝影風選角母版,以及早期的動畫風三視圖。以下全部為 AI 生成影像,並非真人照片。

孫雪娥選角母版(AI 擬真攝影風,非真人照片)
選角母版・AI 擬真攝影風
孫雪娥動畫風三視圖
原典研究・動畫風三視圖

AI 選角檔案

虛擬劇組的工作底稿:把原典證據翻譯成給 AI 的設定指令,再由指令生成上方的角色視覺與聲音方向。完整方法見影像工作室

形象設定指令(給圖像模型)
風格

Live-action late-Ming historical wardrobe-test photography, real adult Chinese actor, natural skin and textile detail, neutral warm-gray studio background

設定描述(中文)

真人古裝定裝照:四十至四十五歲的成年中國女性,成熟漂亮、身量偏矮、體態結實、方圓臉、警醒略狹長眼、平直眉與堅定嘴角;神情能幹、受壓而帶戒備。穿暗赭交領短襖、深青長裙、炭灰圍腰與黑布鞋,髮髻牢固、只配一支樸素深色玉簪;有妾室基本體面但不是富貴正室或老僕。

設定描述(英文原稿)
Create a late-Ming live-action wardrobe-test photograph of another adult character as one real adult Chinese actor, visibly a mature woman around forty to forty-five. This is a new casting from text: no face identity image is authorized. Give her a unique face that does not resemble any other woman in the reference set. She must be attractive in a grounded, mature, practical way—not youthful glamour, not aristocratic polish, and not an elderly servant. She has a compact, sturdy build, slightly shorter stature, stable shoulders, capable arms and hands, and the physical credibility of a woman who has long supervised a large household kitchen and its women servants. Her face is round-square and naturally beautiful, with alert slightly narrow eyes, level practical brows, a firm mouth, and subtle age-appropriate lines. Her expression combines guarded competence, compressed resentment, pressure, and worry without becoming grotesque, theatrical, or defeated.

Her social role must read clearly: she is the fourth concubine of a wealthy late-Ming merchant household and also the practical manager of kitchen and servant-women affairs. Preserve modest household-mistress dignity, but keep her visibly below the primary wife and richer concubines. Dress her in historically plausible late-Ming everyday working clothes: a muted russet cross-collar jacket in substantial plain-woven silk-cotton or ramie, sleeves neatly secured at the wrists for work, a deep desaturated teal long skirt with believable weight and folds, a clean charcoal waist apron tied simply over the skirt, and plain black cloth shoes. Her hair is arranged in a neat, secure married-woman period bun with one modest dark-jade hairpin. Use restrained period makeup only. No luxurious brocade, no gold-heavy embroidery, no phoenix crown, no elaborate jewelry, no servant uniform, no rags, and no modern fashion.

Show exactly one person in a single coherent full-body portrait, standing front-facing with a slight three-quarter turn and a broad, grounded practical stance. Both hands must hang naturally and remain completely visible; both feet and shoes must be fully visible on the studio floor. No tray, cookware, food, furniture, set dressing, or other prop. Frame head to toe with comfortable margin above the hair and below the shoes; the figure should occupy about eighty to eighty-five percent of the image height. Use an eye-level, distortion-controlled full-frame 70–85 mm portrait-lens look, natural perspective, and no wide-angle elongation.

Use the approved another adult character image only as a photographic Art Lock: match its disciplined live-action wardrobe-test finish, neutral warm-gray seamless studio, softly modeled large-source key light, restrained fill, gentle separation, grounded contact shadow, realistic exposure, and natural skin, hair, and cloth behavior. Do not copy or blend his face, eyebrows, eyes, nose, mouth, facial hair, hairstyle, hat, body, pose, garments, dark teal-black palette, masculine traits, rank, or personality into this woman. Render real skin microtexture, pores, fine facial hair, individual hair strands, age-appropriate under-eye and hand detail, natural tonal transitions, and physically credible textile weave, seams, creases, thickness, and gravity. Keep the result photographic and believable rather than digitally perfected.

Deliver a 1024x1536 vertical 2:3 live-action character costume master on a neutral warm-gray studio background. No text, captions, labels, collage, border, logo, watermark, signature, extra people, duplicated body parts, cropped hands, cropped feet, missing fingers, extra fingers, anatomy errors, floating garments, or harsh cast shadows. Exclude anime, manga, illustration, painting, ink outline, line art, cel shading, game concept art, 3D render, CGI character, digital avatar, doll-like skin, plastic skin, beauty-filter skin, over-sharpening, fantasy, wuxia, xianxia, Qing-dynasty styling, imperial styling, royal-primary-wife styling, glamorous courtesan styling, modern influencer makeup, sexualized pose, revealing clothing, exaggerated hourglass body, frailty, hunching, gray-haired elderly coding, and old-servant coding.
排除條件(negative prompt)
illustration, anime, cartoon, comic, 3d render, CGI, child, teenager, modern clothing, fantasy armor, plastic skin, beauty filter, cropped head, cropped feet, malformed hands, extra fingers, extra limbs, text, label, logo, watermark
三視圖指令
Create one native LANDSCAPE live-action wardrobe turnaround sheet from this another adult character interim master. Preserve the exact same unique adult Chinese actress identity, apparent age about 40–45, round-square face, alert narrow eyes, level brows, firm mouth, compact sturdy build, capable shoulders and arms, guarded competent expression, secure married-woman bun with one dark-jade hairpin, muted russet cross-collar work jacket, beige inner cuffs, deep desaturated teal skirt, charcoal waist apron, plain black shoes, realistic skin, hair, and work-cloth. Do not youthify, age into an elderly servant, slim, glamourize, or redesign her.

OUTPUT CONTRACT: one real PNG exactly 1536×1024 pixels. Exactly three separate head-to-toe full-body views in equal 512px conceptual columns: LEFT strict straight front; CENTER strict 90-degree LEFT side profile, nose and toes toward viewer's left, not three-quarter; RIGHT strict back, fully facing away with no over-shoulder face. Same actor, scale, hairpin, apron, costume and baseline across all views. Keep at least 64px clear background at outer canvas edges, 32px from figures to column boundaries, 40px above hairpin and below shoes. No hairpin, sleeve, hand, apron, skirt or shoe may touch/cross an edge or neighboring column. Hands relaxed and readable; profile arms slightly offset while body remains strict profile.

Neutral continuous warm-gray studio background, natural live-action wardrobe photography, realistic pores, age-appropriate lines, hair strands, plain textile weave, folds, contact shadows and lens perspective. No text, labels, lines, logo, watermark, border, kitchen prop, furniture, extra person, crop, padding, blurred/mirrored extension, rich jewelry, anime, illustration, CGI, plastic skin, extra or missing fingers/limbs. Generate natively on the 1536×1024 landscape canvas; do not crop or stretch the portrait source.
聲音設定(給語音模型)

音色:略粗而乾實的中年女中音,胸腔共鳴明顯,氣聲很少(推斷);音高:中低音域,激動時迅速抬高(推斷);語速:平時中速俐落,爭執時加快並帶短促重音(推斷);口音:帶北方市井色彩的口音,咬字直接清楚(推斷);情緒:預設為戒備而務實,底層壓著不平與焦慮(推斷);類比:像一位長年管著大宅廚務、受了委屈便會立刻高聲回擊的中年婦人(推斷)

聲音描述(中文)

成年中年女性的女中低音(推斷),音色乾實、穩厚且略粗,胸腔共鳴清楚,幾乎沒有氣聲(推斷)。音高位於中低區,受壓時會猛然升高;語速俐落務實,節奏帶短促重音,具有北方市井口音色彩,咬字直接(推斷)。平時能量中等,衝突時投射有力,預設情緒是戒備能幹,底層壓著怨氣與焦慮(推斷)。請設計穩定一致的說話音色,而非演繹任何特定臺詞。

聲音描述(英文原稿)
An adult middle-aged female contralto voice (inferred), dry, sturdy, and slightly coarse, with clear chest resonance and very little breathiness (inferred). Medium-low pitch that rises sharply under pressure, brisk practical tempo, clipped rhythmic stresses, and a northern urban vernacular flavor with direct articulation (inferred). Moderate energy at rest, forceful projection during conflict, and a default emotional color of guarded competence with suppressed resentment and anxiety (inferred). Design a consistent speaking instrument, not a performance of any specific line.

定妝設計圖

開拍前的完整定裝一頁看完:大頭特寫、三面全身,加上隨身道具的細節。以本頁核准的身份形象為基準生成——同一位孫雪娥、同一張臉,換一種說明書式的呈現。圖為 AI 生成,並非真人照片。

身穿棕色上衣配深藍長裙、繫黑色圍裙的婦人定妝照(AI 生成,非真人照片)
定妝設計圖・AI 生成

角色寫真

定裝照管的是「像不像」,寫真管的是「有沒有戲」——同一位孫雪娥,離開設定表、走進光線裡的樣子。全部角色的寫真收在寫真館。圖為 AI 生成,並非真人照片。

孫雪娥站在竹影牆前,身穿墨黑與深綠交領衣裙,眉頭微蹙(AI 生成,非真人照片)
竹影