《京城的雪·落不到江南》完整生图与视频提示词

===== image-01 | 01 · 携莲入庭 | 首次生图原文 =====
Use case: identity-preserve.
生成一张独立的电影分镜写实照片，必须是16:9横版，目标分辨率1536×864，无黑边，不是拼图，不含文字、编号、字幕、标志或水印。系列《白莲暖庭·归房换装》，暖柔自然侧光，深巧克力棕的旧木结构、象牙白和克制橄榄绿，真实细微肤质、发丝和布纹，柔亮高光有层次，阴影保留木纹，避免整图橙色或磨皮。用户原图定义同一名明确成年的约28岁女性，保持相同五官、体型、深棕黑发的松挽低髻和两侧碎发，神情自然。
图1是本次扩展编辑目标，图2辅助核对同一个人的脸和白衣织纹。将图1原有竖图向左右自然扩展为16:9横幅，完整保留人物头到脚、原表情朝向、低髻后白纱、白莲花束、象牙白立领长袖长上衣与白色宽松长裤、白鞋和原石阶。她此时未穿黑色丝袜，也未换旗袍。人物位于中央略偏左，左右补全同一处雕花木门、廊柱、台阶与树叶，维持原图光向和建筑透视。不得裁掉头脚、不得改变人物动作或添加别人，不改变成片里的原白衣裤装。

===== image-02 | 02 · 指间抚莲 | 首次生图原文 =====
Use case: identity-preserve.
生成一张独立的电影分镜写实照片，必须是16:9横版，目标分辨率1536×864，无黑边，不是拼图，不含文字、编号、字幕、标志或水印。系列《白莲暖庭·归房换装》，暖柔自然侧光，深巧克力棕的旧木结构、象牙白和克制橄榄绿，真实细微肤质、发丝和布纹，柔亮高光有层次，阴影保留木纹，避免整图橙色或磨皮。用户原图定义同一名明确成年的约28岁女性，保持相同五官、体型、深棕黑发的松挽低髻和两侧碎发，神情自然。
图1是本次扩展编辑目标（窗边插莲原图），图2辅助核对人物原白衣裤装。把图1自然扩成16:9横幅，保留她低头整理白莲的动作与原面容，手握连续花茎，白莲落入原有灰白斑点陶瓶，原木桌、浅碗与花瓣位置维持逻辑，新增横向空间展现雕木窗和后方旧木家具。原有象牙白高领盘扣长袖长上衣搭配白长裤，尚未穿黑色丝袜，不得变旗袍或黑腿。人物居左中，花瓶居右中，枝叶少量虚化框边。保持花瓣衣物层次，画内无可读题字。

===== image-03 | 03 · 理齐袜口 | 最终修订原文 =====
Create the NEXT shot of this same ordinary dressing sequence, a single cinematic photorealistic full-bleed 16:9 horizontal frame. Reference1 is the accepted preceding shot04 and governs adult woman's face, low bun, opaque ivory long closed robe, room, chair, stool, light and left foot on stool. Reference2 demonstrates the next action only: her hands now smooth the SAME black hosiery upward to just BELOW the knee on that same raised left leg. Reference3 is identity/natural skin only, never triptych. Make the framing visibly tighter than shot04: compose a medium seated portrait from full head to the entire raised foot and stool top, with face, both hands, covered thigh and knee, black-covered calf/heel/toes all in frame. Do NOT crop the top of her head, hands or foot; her upper body remains a meaningful focus. She occupies about 75% of frame height, move camera closer slightly on same side, retain some window and part of green qipao on rack at right as context. The robe securely covers chest, lap and thighs. Folded hosiery edge held by her fingers at just below knee, sheer black weave fully covers heel and closed toes, natural joint anatomy. No garter, no lingerie exposure, no erotic posing. Neutral focused face with natural pores, fine stray hair, minimal makeup, no oily sheen or plastic skin. Lighting exactly like reference1: warm-neutral softened side window daylight, dim brown wood interior, subdued highlights, detailed whites and greens, no orange or blue cast. No text, annotations, split panels or watermark.

===== image-04 | 04 · 盘扣凝眸 | 最终修订原文 =====
Photorealistic cinematic still, single full-bleed 16:9 landscape image, never a collage. Continue the same classical adult woman, dark loose low bun, natural approximately 170 cm body proportions. Skin with delicate pores and real lip texture, restrained makeup, no oily shine, no plastic smoothing, no exaggerated eyelashes. Preserve warm-neutral ivory, pine green and dark brown old wood palette. Slightly quieter daylight than the source: reduce the brightness of the room and golden glare subtly, retain soft natural window light on her face and legible details. Same daytime, no teal night, no orange filter, no text, borders, arrows or watermark. Edit the FIRST reference. Same waist-up composition in dressing room, same dark green short-sleeve jacquard qipao and hanging ivory robe, same furniture. The top Chinese frog closure is ALREADY FASTENED. Her fingertips rest lightly near the fastened collar after finishing, a small pause rather than still struggling to fasten. Her face lifts slightly and eyes turn toward the off-frame doorway at screen LEFT, with a restrained expectation; lips relaxed neutral, no smile, no tears, no melodrama. Hands anatomically clear and lightly separated. Natural facial texture based on second reference, identity only, no triptych. Leave no second green dress on the hanger. The lotus vase stays in another room, don't add flowers here.

===== image-05 | 05 · 镜前照影 | 最终修订原文 =====
Photorealistic cinematic still, single full-bleed 16:9 landscape image, never a collage. Continue the same classical adult woman, dark loose low bun, natural approximately 170 cm body proportions. Skin with delicate pores and real lip texture, restrained makeup, no oily shine, no plastic smoothing, no exaggerated eyelashes. Preserve warm-neutral ivory, pine green and dark brown old wood palette. Slightly quieter daylight than the source: reduce the brightness of the room and golden glare subtly, retain soft natural window light on her face and legible details. Same daytime, no teal night, no orange filter, no text, borders, arrows or watermark. Edit the FIRST reference as locally and faithfully as possible. Preserve exact full-body 170cm proportions, same mirror, pine-green short-sleeved below-knee qipao, black sheer footed hosiery on both legs, black low-heel pumps, pose and room. Change expression from smile into relaxed unsmiling lips and a very slight inward hesitation, her eyes looking toward her mirror reflection at screen LEFT. A quiet pause after getting dressed. No frown, tears or exaggerated sadness. One hand lightly rests near waist, other hangs relaxed as source. Make mirror reflection match her exact pose, bun, garment and lighting with credible mirror geometry. Skin should match second identity reference's natural pores and minimal makeup, not a shiny beauty render. Single full-body frame, no triptych.

===== image-06 | 06 · 花前归静 | 最终修订原文 =====
Create a new ending storyboard image continuing these references, single full-bleed 16:9 landscape cinematic photograph, not collage. Image1 is the exact flower-arranging room to revisit: dark carved wooden lattice windows, old wood table, round-bellied ivory ceramic vase with small dark speckles and flared short neck, white lotus blossoms/bud and stems arranged in it, shallow brown petal bowl, small wooden boxes. Keep this identifiable room and vase arrangement. Image2 governs SAME adult woman's identity, dark low bun, natural 170cm proportions and NEW outfit: muted pine-green short-sleeved jacquard qipao with closed mandarin collar and frog closures, black sheer footed hosiery and black low-heeled pumps. Image3 reinforces the real face, pores, lip texture, minimal makeup, no oily highlights. Do NOT reproduce three-view layout. She has returned to the flower room after changing. Compose a new medium-wide quiet ending: she stands on the left half beside the flower table, shown head to approximately below knees, body turned partly toward an OPEN wooden window at screen right. We clearly see her natural three-quarter/profile face and the direction of her gaze OUT toward an EMPTY sunlit courtyard beyond the window; no other people. One relaxed hand rests naturally on the table beside the same speckled lotus vase; other arm relaxed. Her lips are neutral without smile, expression composed with subtle unanswered expectation, eyes open, no tears or melodrama. Leave visual breathing room through the open window to suggest waiting. The lotus in foreground ties to image1, yet does not block her face; flowers softly illuminated, skin face softly side-lit. Warm-neutral ivory and muted green, quiet darker brown interior, soft late-day light matching preceding image2, same daytime, restrained photographic contrast and realistic wood/skin/fabric, no blue night, no orange filter. The ivory tunic has been changed OFF: she wears ONLY the green qipao. No bed, garment rack, mirror or extra dress in this flower room. No letter, subtitle, logo, watermark or border. Balanced atmosphere, ordinary quiet storytelling.

===== image-identity | 女主三视图｜选定身份参考 | 人物参考生成原文 =====
Create a photographic facial identity reference sheet of the SAME adult heroine shown in the three supplied references. Image 2 (the closest face with white veil) is the main source of her facial anatomy and recognizable identity; image 1 and image 3 support continuity with the existing heroine. This is a new three-angle FACE sheet, not full-body scenes and not a new woman.
Output a single high-detail 16:9 LANDSCAPE image, ideally 3840 x 2160, with THREE equally sized portrait panels side by side. Each panel uses the same plain warm-gray beige background, same scale of head, same eye level, same exposure and same soft light. Head-and-shoulders crops from complete crown/hair to the upper chest, with comfortable headroom, face large enough to see real skin texture. No labels, no captions, no frame graphics.
LEFT PANEL: precise frontal face, 0-degree yaw, upright head, eyes looking straight into the lens, natural closed relaxed lips.
MIDDLE PANEL: turn the same head and shoulders 45 degrees toward the viewer's right, eyes following the face direction, clear three-quarter view with both eyes visible in perspective.
RIGHT PANEL: a true strict 90-degree profile toward the viewer's right, upright head, only the near-side eye visible; the far eye is occluded. The forehead, nose bridge, nose tip, natural lips and rounded chin form a clear coherent profile. This must be a distinctly different angle from the middle panel.
Identity: the same adult woman about 28 years old, quiet classical Chinese elegance conveyed by calm posture and period-inspired grooming. Preserve the referenced oval face, naturally shaped eyebrows and almond eyes, authentic eye spacing, nose shape, mouth and jaw. Real human anatomy and subtle natural asymmetry, without enlarged doll eyes, exaggerated V jaw, tiny pinched nose, plumped lips or a frozen beauty-ad expression. Same dark low loose bun and same hairline in all three panels, a few natural wisps at the temples but hair swept clear of the eyes and facial contour. The low bun is visibly plausible in the profile view. No veil, flowers, earrings or hair ornaments. The same simple ivory mandarin-collar top in all three panels, visible only at shoulders and upper chest.
MOST IMPORTANT — natural skin and restrained photographic realism: realistically varied skin at the appropriate portrait scale, fine pores most visible around the nose and inner cheeks, very fine facial vellus hair, faint under-eye creases, believable soft eyelid texture, subtle natural warm/cool skin tone variation and normal slight redness near the nostrils. Healthy clean skin; do not add freckles, acne or heavy rough grain. Skin is primarily matte with delicate broad highlights, not oily, sweaty, wet, glossy, waxy or plastic. Minimal invisible makeup, natural lashes, unpainted individual brow hairs, soft naturally colored lips with fine lip lines and no shiny gloss. Preserve age-appropriate tiny details instead of digitally airbrushing them away.
Lighting: large soft daylight source just above camera height and slightly to one side, gentle fill, enough modeling to show facial anatomy without theatrical shadows or strong backlight. Warm-neutral balanced white balance and realistic restrained color; no orange/sepia filter, no bloom veil. Crisp natural eyes, softly resolved skin microtexture, all facial planes suitably in focus, a neutral portrait-lens perspective with no wide-angle distortion. The result should resemble a careful, understated, minimally retouched photographic casting reference session rather than a 3D render, beauty filter or glamour illustration.
Keep one consistent face across all three panels; change only the angle. No textual elements or watermark.

===== image-identity-skin | 女主三视图｜自然肤质修订 | 人物肤质修订原文 =====
Refine the supplied three-panel facial reference sheet. Preserve the SAME adult woman's exact recognizable facial identity, the same three-panel 16:9 layout, equal head scale, low bun, ivory mandarin-collar clothing and warm gray background. Keep her attractive classical quiet appearance.
Make these precise corrections:
ANGLE: In the RIGHT panel, rotate the face to a truly strict 90-degree side profile, pointing to the viewer's right. Only ONE eye and ONE eyebrow can be visible. The far eye, far eyelashes and far brow must be entirely hidden behind the bridge of the nose. Show a clean continuous forehead–nose–lips–chin silhouette. Maintain the same believable nasal projection and jaw anatomy as the other panels. The middle stays a clear 45-degree three-quarter angle and the left stays straight-on.
REALISM IN ALL THREE PANELS: Gently remove the cosmetic beauty-filter finish without changing facial shape. Replace uniform long curled lashes and the dark eyeliner effect with delicate irregular natural eyelashes and a very subtle bare lash line. Preserve fine unretouched skin variation: small pores at nose and inner cheeks, subtle normal unevenness of skin color, delicate lower-eyelid creases, faint peach fuzz, individual brow hairs, tiny natural surface variation on the forehead. Healthy real skin, no artificial freckle pattern, no acne added and no gritty overlay. Lips are relaxed, lightly closed, softly pigmented with actual fine lip lines, without gloss or plumping. Keep skin mostly matte, soft natural light, mild broad highlights rather than shining hot spots; no waxy smoothness, no greasy skin and no exaggerated pore sharpening. Maintain subtle human facial asymmetry and natural restrained color. Preserve documentary portrait realism with enough facial depth and skin detail, not a cosmetics advertisement or a 3D render.
This remains a facial continuity reference of one adult woman photographed at three angles. Do not add text, labels, watermarks or decorative graphics. Return a single high-detail 16:9 image.

===== video-a | A 段 · 携莲入庭 | 资料包原文·H3 Ref2VA =====
subject_definitions:
<Subject 1> is the single adult woman shown from three angles in <Picture 1>. Preserve her facial proportions, dark eyes, natural skin texture, hairline and softly arranged dark low bun. Her standing proportions are naturally consistent with a height of approximately 170 cm. The three panels show one identity; only one physical woman appears in the video. Use this image for identity only; her clothes and surroundings come from the shot references.
<Picture 2> is the composition and wardrobe reference for [Shot 1]: the woman holds white lotus flowers on stone steps outside a carved wooden courtyard doorway, wearing an ivory long tunic, white trousers and white shoes.
<Picture 3> is the composition and prop reference for [Shot 2]: the same woman, still in her ivory clothes, touches a white lotus beside a wooden window and table. Preserve the speckled ivory vase, two open lotus blossoms and one bud.
<Picture 4> is the composition, wardrobe and action-state reference for [Shot 3]: the same woman sits in a securely closed ivory robe, with one foot supported on a wooden stool and black sheer hosiery already reaching just below that knee. A pine-green qipao hangs at screen right.

summary:
[reference generation] <Subject 1> carries lotus flowers into the courtyard, tends the same flowers at the window, then quietly prepares her clothing. Follow <Picture 2>, <Picture 3> and <Picture 4> in that order. Use restrained gestures and deliberate cuts to connect the three locations and clothing states within ten seconds.

retention_analysis:
<Subject 1> (appears in [Shot 1], [Shot 2], [Shot 3]): fully_preserved - maintain the same face, hairline, low bun and natural body proportions.
<Picture 2> ([Shot 1] composition anchor): fully_preserved - retain the doorway, stone steps, ivory outfit, lotus bundle and initial standing arrangement.
<Picture 3> ([Shot 2] composition anchor): fully_preserved - retain the flower room, ivory outfit, vase, lotus arrangement and initial hand contact.
<Picture 4> ([Shot 3] composition anchor): fully_preserved - retain the bedroom arrangement, closed robe, supported foot, existing hosiery height and hanging green qipao.

detailed_description:
The target video is a ten-second, 16:9, realistic live-action sequence with exactly three shots. Warm-neutral daylight, ivory whites, restrained pine green and deep brown wood remain consistent. Natural pores, fine facial hair and lip texture stay visible. Movement has ordinary physical weight, softened wrists and relaxed shoulders. The entire video is silent, without speech, vocalization, ambient sound, effects or music. There are no captions or title cards.
[Shot 1] The opening follows <Picture 2>: <Subject 1> stands full length beneath the carved doorway, her hands gently gathering the lotus stems in front of her. Her elbows remain comfortably close to her body and her shoulders rest low. After a brief pause, she glances toward the next stone tread. From approximately 0.40 to 2.60 seconds, her leading foot settles onto that lower tread, her supporting knee softens, and her weight transfers before the trailing foot joins it. The hem follows with a small delayed sway; the flowers move with her body while remaining securely held. Her head stays poised over her shoulders. The camera pulls out with small amplitude at slow speed, preserving the whole figure and doorway. Allow the step to settle before the cut.
[Shot 2] At 00:03.000, the camera cuts to the window-side composition in <Picture 3>. The cut omits the short walk indoors and placing the stems into the vase; the lotus now stands in the same speckled vessel. <Subject 1> is still wearing the ivory tunic. Starting with the pictured fingertip contact, she lightly supports one petal, lets it yield a few millimeters, and releases it by approximately 5.80 seconds. Her other hand steadies the stems inside the vase. The moving wrist softens as her fingers withdraw, and her gaze follows them before returning to the flower. Her shoulders remain level. The camera pushes in with small amplitude at slow speed, keeping her face, hands and flowers together. Let the released petal settle.
[Shot 3] At 00:06.500, the camera cuts to <Picture 4>, connecting the withdrawing fingertips to hands arranging fabric. This deliberate time omission reveals that <Subject 1> has changed into the closed ivory robe. The camera holds a static medium shot including her face, both hands and the complete raised foot. The hosiery already covers the toes, heel and calf up to the pictured position below her knee. She levels its folded upper edge once, then smooths one small wrinkle immediately beneath it. Her torso inclines slightly from the hips with a relaxed neck; the raised foot keeps its weight on the stool. The robe continues covering her chest, lap and thighs, and the green qipao stays on its hanger. By approximately 9.60 seconds, her fingertips release the material and rise slightly before settling. End on this completed gesture, with wardrobe and room unchanged and no fade.

overall_soundscape:
N/A

non_diegetic_music:
N/A


===== video-b | B 段 · 照影归莲 | 资料包原文·H3 Ref2VA =====
subject_definitions:
<Subject 1> is the single adult woman shown from three angles in <Picture 1>. Preserve her facial proportions, dark eyes, natural skin texture, hairline and softly arranged dark low bun. Her standing proportions are naturally consistent with a height of approximately 170 cm. The three panels show one identity; only one physical woman appears in the video. Use this image for identity only; her clothes and surroundings come from the shot references.
<Picture 2> is the composition, wardrobe and initial gesture reference for [Shot 1]: the woman has changed into the pine-green short-sleeved qipao, its collar fastening already closed, with fingertips resting near it. The ivory robe now hangs behind her.
<Picture 3> is the full-length composition reference for [Shot 2]: the same woman wears the same green qipao, sheer black hosiery and black low-heeled pumps beside the tall carved mirror.
<Picture 4> is the composition and location reference for [Shot 3]: the same woman, still wearing the green qipao, stands beside the speckled ivory vase and white lotus in the original flower room, looking toward the empty courtyard.

summary:
[reference generation] <Subject 1> completes dressing, regards her reflection, then returns to the white lotus. Follow <Picture 2>, <Picture 3> and <Picture 4> in that order. This ten-second sequence follows a cut from her completed hosiery adjustment; opening fingertips at the fastened collar continue the quiet hand-led rhythm. Maintain composure and finish with her gaze returning to the flowers.

retention_analysis:
<Subject 1> (appears in [Shot 1], [Shot 2], [Shot 3]): fully_preserved - maintain the same face, hairline, low bun and natural body proportions.
<Picture 2> ([Shot 1] composition anchor): fully_preserved - retain the fitted green qipao, closed collar, hand placement, dressing room and hanging ivory robe.
<Picture 3> ([Shot 2] composition anchor): fully_preserved - retain the full-length framing, same outfit and footwear, carved mirror and reflected room.
<Picture 4> ([Shot 3] composition anchor): fully_preserved - retain the flower room, green outfit, table, speckled vase, two lotus blossoms, one bud and courtyard view.

detailed_description:
The target video is a ten-second, 16:9, realistic live-action sequence with exactly three shots. Warm-neutral daylight, ivory whites, restrained pine green and deep brown wood remain consistent. Natural pores, fine facial hair and lip texture stay visible. Movement has ordinary physical weight, softened wrists and relaxed shoulders. The entire video is silent, without speech, vocalization, ambient sound, effects or music. There are no captions or title cards.
[Shot 1] The opening follows <Picture 2>, a close view of <Subject 1> already wearing the pine-green qipao with its collar fastening closed. Her face and hands are softly lit from the existing window side. The shot opens after an omitted dressing interval: the ivory robe now hangs behind her, and the green dress is no longer on the hanger. Her fingertips rest beside the fastening. By approximately 2.10 seconds, she gently releases them and lowers both hands several centimeters, keeping her fingers loosely curved and her wrists supple. Her eyes lift slightly along the existing off-screen direction; her chin follows only a fraction. Her shoulders stay relaxed, lips naturally settled, and breathing produces a subtle physical rise and fall without sound. The camera holds a stable close shot, preserving the collar, fingers and face together.
[Shot 2] At 00:02.500, the camera cuts to the full-length mirror composition in <Picture 3>, showing the completed green outfit, sheer black hosiery and black low-heeled pumps. The same <Subject 1> now sees her whole reflection. The hand pictured at her waist gently releases the fabric and descends along her side. She makes a tiny weight adjustment with both shoes planted, then draws her chin inward slightly while keeping her neck long and shoulders easy. Finish this motion by approximately 5.50 seconds. Her gaze remains on her reflected face; her expression stays composed. The camera holds a static shot so the carved frame and room remain stable. The reflection duplicates each hand, head and body movement with correct mirror geometry and timing, representing the same woman. Let her hand come to rest before the cut.
[Shot 3] At 00:06.000, the camera cuts to <Picture 4>, returning to the flower room after an omitted walk. <Subject 1> remains in the green qipao; one hand already rests beside the vase on the wooden table. The same two open white lotus blossoms and bud remain in the speckled vessel, with the petal bowl and boxes in their established positions. She initially looks toward the empty courtyard. After a brief pause, her eyes return to the lotus and her head follows with a small downward inclination. Her fingers relax while maintaining their contact point, leaving the nearby objects undisturbed. Complete the gaze change by approximately 9.00 seconds. The camera pulls out with small amplitude at slow speed, giving the window and courtyard slightly more space. During the final second she rests in this composed posture, with only subtle breathing and softly moving outdoor leaves. End on a fully visible image for the editor to finish.

overall_soundscape:
N/A

non_diegetic_music:
N/A


