[
  {
    "id": null,
    "name": "沈砚",
    "importance": "protagonist",
    "oneLiner": "凡骨绝脉的义庄少年，身负剑仙剑心，从砍柴到守渊——以一身凡骨，立天地。",
    "hasSheetPrompt": true,
    "hasBasePrompt": true,
    "negativePrompt": "plastic waxy skin, over-smoothed airbrushed skin, poreless doll face, perfectly symmetric face, dead lifeless eyes, helmet-like hair without flyaways, flat textureless fabric, stiff mannequin pose, extra fingers, deformed hands, watermark, text, logo, cartoon, flat vector, chibi, aged elderly face, feminine features",
    "sheetPrompt": "Single character model sheet on ONE 16:9 landscape canvas, plain white background, thin hairline rules dividing three zones. LEFT ZONE about 34% of the canvas width: one large front-facing bust portrait, head and shoulders centered like a passport photo, both shoulders fully visible, bottom edge cut straight, face fully detailed as the design reference. Lean Han Chinese boy of sixteen with roughly cut black hair, stark white temple strands, tired stubborn eyes, mended grey short tunic. RIGHT-TOP ZONE: three full-body views, front, side, back, standing side by side on one shared ground line, same face and proportions as the left bust. Worn grey short tunic, straw sandals, cloth-wrapped wooden practice sword slung on the back, cracked jade half-disc pendant at the chest. RIGHT-BOTTOM ZONE: a row of four to five small isolated close-up studies. Close-ups of the cracked jade half-disc pendant on a frayed cord, the white temple strands, callused palm and grip, the cloth wrap of the wooden practice sword, the straw-sandal weave. PROPORTIONS ARE CRITICAL: all three full bodies equal height, consistent head-to-body ratio, correct limb lengths, feet on the same ground line, headroom and footroom, never stretch or squash the figure to fit the canvas. The detail studies give way, not the figures - if the row does not fit, extend downward along the right edge. LIGHTING IN THE LEFT ZONE ONLY: soft directional key light from the upper left with natural falloff, ambient occlusion under the jaw, eye sockets, collar and neckline. LIGHTING IN THE RIGHT ZONES: flat even orthographic lighting, no directional key, no cast shadows. Han Chinese, ancient Chinese cultivation-era (xianxia), a lean sixteen-year-old male servant boy. Semi-realistic character illustration, painterly rendering with soft blended edges and visible brush texture, anatomically grounded"
  },
  {
    "id": null,
    "name": "云芷柔",
    "importance": "protagonist",
    "oneLiner": "药园里给杂役送粥的温柔师姐，为护他一世，失音、远嫁、守候，把「等」字守成奇迹。",
    "hasSheetPrompt": true,
    "hasBasePrompt": true,
    "negativePrompt": "plastic waxy skin, over-smoothed airbrushed skin, poreless doll face, perfectly symmetric face, dead lifeless eyes, helmet-like hair without flyaways, flat textureless fabric, stiff mannequin pose, extra fingers, deformed hands, watermark, text, logo, cartoon, flat vector, chibi, masculine features, harsh shadows",
    "sheetPrompt": "Single character model sheet on ONE 16:9 landscape canvas, plain white background, thin hairline rules dividing three zones. LEFT ZONE about 34% of the canvas width: one large front-facing bust portrait, head and shoulders centered like a passport photo, both shoulders fully visible, bottom edge cut straight, face fully detailed as the design reference. Gentle Han Chinese woman of eighteen with soft oval face, warm smiling eyes, fine black hair with a small wooden hairpin, pale cyan-grey robe and white inner collar. RIGHT-TOP ZONE: three full-body views, front, side, back, standing side by side on one shared ground line, same face and proportions as the left bust. Pale cyan-grey long robe with a simple cloth sash, short bamboo flute at the waist, small woven food basket in hand. RIGHT-BOTTOM ZONE: a row of four to five small isolated close-up studies. Close-ups of the small wooden hairpin, the bamboo flute, the sash knot of the pale cyan robe, the weave of the food basket, the clean white collar. PROPORTIONS ARE CRITICAL: all three full bodies equal height, consistent head-to-body ratio, correct limb lengths, feet on the same ground line, headroom and footroom, never stretch or squash the figure to fit the canvas. The detail studies give way, not the figures - if the row does not fit, extend downward along the right edge. LIGHTING IN THE LEFT ZONE ONLY: soft directional key light from the upper left with natural falloff, ambient occlusion under the jaw, eye sockets, collar and neckline. LIGHTING IN THE RIGHT ZONES: flat even orthographic lighting, no directional key, no cast shadows. Han Chinese, ancient Chinese cultivation-era (xianxia), a gentle eighteen-year-old female herbal-garden keeper. Semi-realistic character illustration, painterly rendering with soft blended edges and visible brush texture, anatomically grounded"
  },
  {
    "id": null,
    "name": "柳含烟",
    "importance": "major",
    "oneLiner": "红衣如火的魔道圣女，为救义母被困血河宗，最后叛教挡掌，把命留在了封印台前。",
    "hasSheetPrompt": true,
    "hasBasePrompt": true,
    "negativePrompt": "plastic waxy skin, over-smoothed airbrushed skin, poreless doll face, perfectly symmetric face, dead lifeless eyes, helmet-like hair without flyaways, flat textureless fabric, stiff mannequin pose, extra fingers, deformed hands, watermark, text, logo, cartoon, flat vector, chibi, sweet dollish innocence, soft girlish roundness",
    "sheetPrompt": "Single character model sheet on ONE 16:9 landscape canvas, plain white background, thin hairline rules dividing three zones. LEFT ZONE about 34% of the canvas width: one large front-facing bust portrait, head and shoulders centered like a passport photo, both shoulders fully visible, bottom edge cut straight, face fully detailed as the design reference. Striking Han Chinese woman in her early twenties with sharp cheekbones, cold frost-like eyes, arched brows, black high bun with a red ribbon and a red plum-blossom hairpin, fiery crimson robes. RIGHT-TOP ZONE: three full-body views, front, side, back, standing side by side on one shared ground line, same face and proportions as the left bust. Fiery crimson robes with dark red trim and black inner layers, wide flowing sleeves, thin silver waist chain. RIGHT-BOTTOM ZONE: a row of four to five small isolated close-up studies. Close-ups of the red plum-blossom hairpin, the silver waist chain, the dark red sleeve trim, the slender pale fingers, the crisp silk fold of the crimson robe. PROPORTIONS ARE CRITICAL: all three full bodies equal height, consistent head-to-body ratio, correct limb lengths, feet on the same ground line, headroom and footroom, never stretch or squash the figure to fit the canvas. The detail studies give way, not the figures - if the row does not fit, extend downward along the right edge. LIGHTING IN THE LEFT ZONE ONLY: soft directional key light from the upper left with natural falloff, ambient occlusion under the jaw, eye sockets, collar and neckline. LIGHTING IN THE RIGHT ZONES: flat even orthographic lighting, no directional key, no cast shadows. Han Chinese, ancient Chinese cultivation-era (xianxia), a striking female demon-cult saintess in her early twenties. Semi-realistic character illustration, painterly rendering with soft blended edges and visible brush texture, anatomically grounded"
  },
  {
    "id": null,
    "name": "血沧澜",
    "importance": "major",
    "oneLiner": "血河宗宗主、化神魔尊，布局二十年只为打开魔渊迎魔主降世——沈砚的杀母仇人，魔主降临前的关底。",
    "hasSheetPrompt": true,
    "hasBasePrompt": true,
    "negativePrompt": "plastic waxy skin, over-smoothed airbrushed skin, poreless doll face, perfectly symmetric face, dead lifeless eyes, helmet-like hair without flyaways, flat textureless fabric, stiff mannequin pose, extra fingers, deformed hands, watermark, text, logo, cartoon, flat vector, chibi, youthful face, healthy rosy skin",
    "sheetPrompt": "Single character model sheet on ONE 16:9 landscape canvas, plain white background, thin hairline rules dividing three zones. LEFT ZONE about 34% of the canvas width: one large front-facing bust portrait, head and shoulders centered like a passport photo, both shoulders fully visible, bottom edge cut straight, face fully detailed as the design reference. Sinister withered Han Chinese elderly man with sunken gaunt cheeks, sallow grey skin, bloodshot eyes with faint black qi, grey-streaked thin beard, sparse white-grey topknot with a dark iron crown, blood-red robe. RIGHT-TOP ZONE: three full-body views, front, side, back, standing side by side on one shared ground line, same face and proportions as the left bust. Floor-length blood-red robe with black embroidered cloud patterns, heavy sleeves, dark inner robes, pale clawed hands holding a blood-red spear. RIGHT-BOTTOM ZONE: a row of four to five small isolated close-up studies. Close-ups of the dark iron crown on the sparse topknot, the blood-red spear, the clawed pale hand, the black cloud embroidery on the robe, the black qi seeping from the eye corner. PROPORTIONS ARE CRITICAL: all three full bodies equal height, consistent head-to-body ratio, correct limb lengths, feet on the same ground line, headroom and footroom, never stretch or squash the figure to fit the canvas. The detail studies give way, not the figures - if the row does not fit, extend downward along the right edge. LIGHTING IN THE LEFT ZONE ONLY: soft directional key light from the upper left with natural falloff, ambient occlusion under the jaw, eye sockets, collar and neckline. LIGHTING IN THE RIGHT ZONES: flat even orthographic lighting, no directional key, no cast shadows. Han Chinese, ancient Chinese cultivation-era (xianxia), a sinister withered elderly male demon-sect sovereign. Semi-realistic character illustration, painterly rendering with soft blended edges and visible brush texture, anatomically grounded"
  },
  {
    "id": null,
    "name": "荒古魔主",
    "importance": "major",
    "oneLiner": "被剑仙独孤问以身为封印镇压千年的荒古魔头，借血沧澜之手破渊而出——全剧真正的最终BOSS。",
    "hasSheetPrompt": true,
    "hasBasePrompt": true,
    "negativePrompt": "cute chibi, cartoon, flat vector, plastic toy, overly stylized anime, human face, western fantasy demon, extra limbs, deformed anatomy, watermark, text, logo",
    "sheetPrompt": "Single character model sheet on ONE 16:9 landscape canvas, plain white background, thin hairline rules dividing three zones. LEFT ZONE about 34% of the canvas width: one large front-facing bust-scale study of the colossal demon lord head, centered like a portrait, fully rendered in detail as the design baseline. RIGHT-TOP ZONE: three FULL-BODY creature views (front / side / back) of the giant head-and-hand form side by side on one shared ground line, equal height, consistent proportions, claw base on the same ground line, headroom and footroom above and below. RIGHT-BOTTOM ZONE: a row of four to five small isolated close-up studies: the burning crimson eye, the horn protrusion, the clawed hand restrained by golden binding, the swirling qi tendrils, the broken seal fragment. Three zones separated by thin hairline rules; the creature never stretched or flattened, the detail studies give way, not the figures. LIGHTING IN THE LEFT ZONE ONLY: dramatic soft key from upper left with a cool rim light and a faint golden under-glow from the seal. LIGHTING IN THE RIGHT ZONES: flat even orthographic lighting, no directional key, no shadows. ONE consistent design across the bust study and the three views, same eye shape, same horns, same bone structure. East Asian xianxia demon lord aesthetic, colossal scale, no text, no watermark. Semi-realistic character illustration, painterly rendering with soft blended edges and visible brush texture, anatomically grounded"
  },
  {
    "id": null,
    "name": "裴无咎",
    "importance": "supporting",
    "oneLiner": "玄霄剑宗百年一遇的三灵根天骄，从俯视凡人的毒舌少年，到燃丹斩渊、替剑宗挺直脊梁的战友。",
    "hasSheetPrompt": true,
    "hasBasePrompt": true,
    "negativePrompt": "plastic or waxy skin, over-smoothed airbrushed complexion, poreless doll face, perfectly symmetrical face, dead flat eyes without specular highlight, helmet-like hair with no loose strands, flat untextured fabric with no weave or wear, stiff mannequin posing, extra fingers, malformed hands, text, watermark, signature, busy or patterned background, harsh cast shadows on the backdrop",
    "sheetPrompt": "Single character model sheet on ONE 16:9 landscape canvas, pure white background with even margins on all sides, divided by thin hairline rules into three zones. LEFT ZONE, about 34% of the canvas width: a front-facing head-and-shoulders bust portrait of a proud lean seventeen-year-old Han Chinese sword prodigy in a black silk brocade robe with silver piping and a high black hair crown as the facial design baseline — both shoulders fully visible, bottom edge cropped flat, face drawn largest and in the finest detail. RIGHT-TOP ZONE: three full-body turnaround views (front, side, back) of the same a proud lean seventeen-year-old Han Chinese sword prodigy in a black silk brocade robe with silver piping and a high black hair crown standing side by side on one shared ground line, identical face and hairstyle to the left bust, correct head-to-body proportions and limb lengths, all three figures the same height, clear headroom above and footroom below. RIGHT-BOTTOM ZONE: four to five small isolated close-up detail studies arranged in an evenly spaced single row — the silver-piped collar of the black brocade robe, the worn scabbard hilt wrapped with dark cord, the high hair crown with its ribbon knot, the narrow sword-belt buckle, one sharply arched eyebrow; if they do not fit in one row, extend them down the right edge of the canvas as a column; the detail studies give way, not the figures. PROPORTIONS ARE CRITICAL: the three full-body figures share the same height and head-to-body ratio, feet planted on the same ground line, never stretched or compressed to fit the layout. LIGHTING IN THE LEFT ZONE ONLY: a soft directional key light from the upper left with gentle falloff and subtle ambient occlusion under the chin, in the eye sockets and where the collar meets the neck, giving the head real volume. LIGHTING IN THE RIGHT ZONES: flat even orthographic lighting with no directional key and no cast shadows, so the figures stay measurable and cleanly cut out. The same face — the same features, hairstyle and expression — in the left bust and in all three right views. Semi-realistic character illustration, painterly rendering with soft blended edges and visible brush texture, anatomically grounded. Han Chinese, ancient China, cultivation-era (xianxia) fantasy setting, muted ink-blue and silver palette."
  },
  {
    "id": null,
    "name": "赵小乙",
    "importance": "supporting",
    "oneLiner": "偷馒头交朋友、记一辈子账的包子铺胖墩兄弟，从杂役到守城将军，最后以三百六十笼包子把情分还清。",
    "hasSheetPrompt": true,
    "hasBasePrompt": true,
    "negativePrompt": "plastic or waxy skin, over-smoothed airbrushed complexion, poreless doll face, perfectly symmetrical face, dead flat eyes without specular highlight, helmet-like hair with no loose strands, flat untextured fabric with no weave or wear, stiff mannequin posing, extra fingers, malformed hands, text, watermark, signature, busy or patterned background, harsh cast shadows on the backdrop",
    "sheetPrompt": "Single character model sheet on ONE 16:9 landscape canvas, pure white background with even margins on all sides, divided by thin hairline rules into three zones. LEFT ZONE, about 34% of the canvas width: a front-facing head-and-shoulders bust portrait of a plump cheerful fifteen-year-old Han Chinese servant boy in a coarse hemp tunic holding two steamed buns as the facial design baseline — both shoulders fully visible, bottom edge cropped flat, face drawn largest and in the finest detail. RIGHT-TOP ZONE: three full-body turnaround views (front, side, back) of the same a plump cheerful fifteen-year-old Han Chinese servant boy in a coarse hemp tunic holding two steamed buns standing side by side on one shared ground line, identical face and hairstyle to the left bust, correct head-to-body proportions and limb lengths, all three figures the same height, clear headroom above and footroom below. RIGHT-BOTTOM ZONE: four to five small isolated close-up detail studies arranged in an evenly spaced single row — the flour-dusted cheek, the patched knee of the hemp tunic, the cloth sash knot, a steamed bun glossy with steam, the frayed cuff; if they do not fit in one row, extend them down the right edge of the canvas as a column; the detail studies give way, not the figures. PROPORTIONS ARE CRITICAL: the three full-body figures share the same height and head-to-body ratio, feet planted on the same ground line, never stretched or compressed to fit the layout. LIGHTING IN THE LEFT ZONE ONLY: a soft directional key light from the upper left with gentle falloff and subtle ambient occlusion under the chin, in the eye sockets and where the collar meets the neck, giving the head real volume. LIGHTING IN THE RIGHT ZONES: flat even orthographic lighting with no directional key and no cast shadows, so the figures stay measurable and cleanly cut out. The same face — the same features, hairstyle and expression — in the left bust and in all three right views. Semi-realistic character illustration, painterly rendering with soft blended edges and visible brush texture, anatomically grounded. Han Chinese, ancient China, cultivation-era (xianxia) fantasy setting, warm earth-and-flour palette."
  },
  {
    "id": null,
    "name": "风清源",
    "importance": "supporting",
    "oneLiner": "须发皆白的剑痴长老，问一句「练命还是练名」便收徒，护短到「先问过老夫的剑」，最后元婴自爆为徒弟断后。",
    "hasSheetPrompt": true,
    "hasBasePrompt": true,
    "negativePrompt": "plastic or waxy skin, over-smoothed airbrushed complexion, poreless doll face, perfectly symmetrical face, dead flat eyes without specular highlight, helmet-like hair with no loose strands, flat untextured fabric with no weave or wear, stiff mannequin posing, extra fingers, malformed hands, text, watermark, signature, busy or patterned background, harsh cast shadows on the backdrop",
    "sheetPrompt": "Single character model sheet on ONE 16:9 landscape canvas, pure white background with even margins on all sides, divided by thin hairline rules into three zones. LEFT ZONE, about 34% of the canvas width: a front-facing head-and-shoulders bust portrait of a very old gaunt Han Chinese sword elder with long snow-white hair and a full white beard in a faded grey-blue linen robe as the facial design baseline — both shoulders fully visible, bottom edge cropped flat, face drawn largest and in the finest detail. RIGHT-TOP ZONE: three full-body turnaround views (front, side, back) of the same a very old gaunt Han Chinese sword elder with long snow-white hair and a full white beard in a faded grey-blue linen robe standing side by side on one shared ground line, identical face and hairstyle to the left bust, correct head-to-body proportions and limb lengths, all three figures the same height, clear headroom above and footroom below. RIGHT-BOTTOM ZONE: four to five small isolated close-up detail studies arranged in an evenly spaced single row — the liver-spotted back of a calloused hand, the wispy white beard strands, the patched sleeve, the plain sword hilt worn smooth, the deep nasolabial crease; if they do not fit in one row, extend them down the right edge of the canvas as a column; the detail studies give way, not the figures. PROPORTIONS ARE CRITICAL: the three full-body figures share the same height and head-to-body ratio, feet planted on the same ground line, never stretched or compressed to fit the layout. LIGHTING IN THE LEFT ZONE ONLY: a soft directional key light from the upper left with gentle falloff and subtle ambient occlusion under the chin, in the eye sockets and where the collar meets the neck, giving the head real volume. LIGHTING IN THE RIGHT ZONES: flat even orthographic lighting with no directional key and no cast shadows, so the figures stay measurable and cleanly cut out. The same face — the same features, hairstyle and expression — in the left bust and in all three right views. Semi-realistic character illustration, painterly rendering with soft blended edges and visible brush texture, anatomically grounded. Han Chinese, ancient China, cultivation-era (xianxia) fantasy setting, muted grey and indigo palette."
  },
  {
    "id": null,
    "name": "梁青崖",
    "importance": "supporting",
    "oneLiner": "青袍外门巡使，引沈砚入剑宗的引路人；守雁回关七日力竭殉国，临终传剑——「剑宗可以没有雁回关，但不能没有脊梁」。",
    "hasSheetPrompt": true,
    "hasBasePrompt": true,
    "negativePrompt": "plastic or waxy skin, over-smoothed airbrushed complexion, poreless doll face, perfectly symmetrical face, dead flat eyes without specular highlight, helmet-like hair with no loose strands, flat untextured fabric with no weave or wear, stiff mannequin posing, extra fingers, malformed hands, text, watermark, signature, busy or patterned background, harsh cast shadows on the backdrop",
    "sheetPrompt": "Single character model sheet on ONE 16:9 landscape canvas, pure white background with even margins on all sides, divided by thin hairline rules into three zones. LEFT ZONE, about 34% of the canvas width: a front-facing head-and-shoulders bust portrait of a weathered middle-aged Han Chinese sword cultivator in a faded blue-grey linen robe with a leather chest guard and a worn sword scabbard as the facial design baseline — both shoulders fully visible, bottom edge cropped flat, face drawn largest and in the finest detail. RIGHT-TOP ZONE: three full-body turnaround views (front, side, back) of the same a weathered middle-aged Han Chinese sword cultivator in a faded blue-grey linen robe with a leather chest guard and a worn sword scabbard standing side by side on one shared ground line, identical face and hairstyle to the left bust, correct head-to-body proportions and limb lengths, all three figures the same height, clear headroom above and footroom below. RIGHT-BOTTOM ZONE: four to five small isolated close-up detail studies arranged in an evenly spaced single row — the pale scar crossing one eyebrow, the worn scabbard edge, the leather chest-guard strap, the grey-streaked temple hair, the wrapped sword-hilt cord; if they do not fit in one row, extend them down the right edge of the canvas as a column; the detail studies give way, not the figures. PROPORTIONS ARE CRITICAL: the three full-body figures share the same height and head-to-body ratio, feet planted on the same ground line, never stretched or compressed to fit the layout. LIGHTING IN THE LEFT ZONE ONLY: a soft directional key light from the upper left with gentle falloff and subtle ambient occlusion under the chin, in the eye sockets and where the collar meets the neck, giving the head real volume. LIGHTING IN THE RIGHT ZONES: flat even orthographic lighting with no directional key and no cast shadows, so the figures stay measurable and cleanly cut out. The same face — the same features, hairstyle and expression — in the left bust and in all three right views. Semi-realistic character illustration, painterly rendering with soft blended edges and visible brush texture, anatomically grounded. Han Chinese, ancient China, cultivation-era (xianxia) fantasy setting, muted blue-grey and earth palette."
  },
  {
    "id": null,
    "name": "沈伯安",
    "importance": "supporting",
    "oneLiner": "守了义庄三十年的老棺材匠，把别人的儿子当亲儿子养大，最后替那孩子的娘守玉而死。",
    "hasSheetPrompt": true,
    "hasBasePrompt": true,
    "negativePrompt": "plastic waxy skin, over-smoothed skin, poreless doll face, perfectly symmetrical face, dead eyes without highlights, helmet-like rigid hair, flat textureless fabric, stiff mannequin pose, neon colors, modern clothing, western features",
    "sheetPrompt": "Single character model sheet on ONE 16:9 landscape canvas, pure white background with even margins, divided by thin hairline rules into three zones. LEFT ZONE (about 34% of the canvas width): one large head-and-shoulders bust portrait, front-facing, centered like an ID photo, both shoulders fully visible, bottom edge cut flat; the face-design baseline — an elderly Han Chinese coffin maker in ancient Chinese cultivation-era garb, stooped, gaunt, deep wrinkles, sparse gray hair, heavy-lidded eyes. LIGHTING IN THE LEFT ZONE ONLY: soft directional key light from upper left, natural falloff, ambient occlusion under the chin, eye sockets and collar-neck junction. RIGHT-TOP ZONE: three full-body views (front, side, back) standing in a row on one shared ground line, same face, hair and expression as the left bust; flat even orthographic lighting, no directional shadows, for proportion and silhouette reference; identical head-to-body ratio across all three, feet on the same ground line, headroom and footroom left clear; PROPORTIONS ARE CRITICAL, never stretch or squash the figure to fit. RIGHT-BOTTOM ZONE: four to five small isolated close-up studies in a row — calloused hands holding a chisel, wrinkled profile with gray stubble, coarse indigo tunic hem and faded hemp apron, a plain wooden coffin plank with lacquer sheen, worn straw sandals; clearly smaller than the full-body figures; if they do not fit, extend downward along the right edge — the detail studies give way, not the figures. LIGHTING IN THE RIGHT ZONES: flat even orthographic lighting, no directional key, no cast shadows. ONE consistent face across all zones. Semi-realistic character illustration, painterly rendering with soft blended edges and visible brush texture, anatomically grounded"
  },
  {
    "id": null,
    "name": "独孤问",
    "importance": "supporting",
    "oneLiner": "千年前以身封渊的剑仙，如今只剩一缕剑心、一记遗音、一缕残识，还在替他守护的人间看门。",
    "hasSheetPrompt": true,
    "hasBasePrompt": true,
    "negativePrompt": "plastic waxy skin, over-smoothed skin, poreless doll face, perfectly symmetrical face, dead eyes without highlights, helmet-like rigid hair, flat textureless fabric, stiff mannequin pose, neon colors, modern clothing, western features, heavy armor",
    "sheetPrompt": "Single character model sheet on ONE 16:9 landscape canvas, pure white background with even margins, divided by thin hairline rules into three zones. LEFT ZONE (about 34% of the canvas width): one large head-and-shoulders bust portrait, front-facing, centered like an ID photo, both shoulders fully visible, bottom edge cut flat; the face-design baseline — a Han Chinese sword immortal in ancient Chinese cultivation-era white hanfu, lean angular face, silver-streaked hair, deep-set bright eyes, faint glow at the collarbone. LIGHTING IN THE LEFT ZONE ONLY: soft directional key light from upper left, natural falloff, ambient occlusion under the chin, eye sockets and collar-neck junction. RIGHT-TOP ZONE: three full-body views (front, side, back) standing in a row on one shared ground line, same face, hair and expression as the left bust; flat even orthographic lighting, no directional shadows, for proportion and silhouette reference; identical head-to-body ratio across all three, feet on the same ground line, headroom and footroom left clear; PROPORTIONS ARE CRITICAL, never stretch or squash the figure to fit. RIGHT-BOTTOM ZONE: four to five small isolated close-up studies in a row — the jade hairpin and half-tied silver-streaked bun, the jade pendant, the glowing collarbone detail, long-fingered hands resting on a sword hilt, the worn hem of the white robe; clearly smaller than the full-body figures; if they do not fit, extend downward along the right edge — the detail studies give way, not the figures. LIGHTING IN THE RIGHT ZONES: flat even orthographic lighting, no directional key, no cast shadows. ONE consistent face across all zones. Semi-realistic character illustration, painterly rendering with soft blended edges and visible brush texture, anatomically grounded"
  },
  {
    "id": null,
    "name": "狗蛋（沈青）",
    "importance": "supporting",
    "oneLiner": "被沈砚收留的小乞丐，从劈柴、认字学起，后来背着听雪剑踏上师父走过的路，成了新一代守渊人。",
    "hasSheetPrompt": true,
    "hasBasePrompt": true,
    "negativePrompt": "plastic waxy skin, over-smoothed skin, poreless doll face, perfectly symmetrical face, dead eyes without highlights, helmet-like rigid hair, flat textureless fabric, stiff mannequin pose, neon colors, modern clothing, western features, clean idealized child",
    "sheetPrompt": "Single character model sheet on ONE 16:9 landscape canvas, pure white background with even margins, divided by thin hairline rules into three zones. LEFT ZONE (about 34% of the canvas width): one large head-and-shoulders bust portrait, front-facing, centered like an ID photo, both shoulders fully visible, bottom edge cut flat; the face-design baseline — a nine-year-old Han Chinese famine orphan boy in ancient Chinese cultivation-era riverside town clothes, round bright eyes, messy hair with straw, dirt-smudged cheeks. LIGHTING IN THE LEFT ZONE ONLY: soft directional key light from upper left, natural falloff, ambient occlusion under the chin, eye sockets and collar-neck junction. RIGHT-TOP ZONE: three full-body views (front, side, back) standing in a row on one shared ground line, same face, hair and expression as the left bust; flat even orthographic lighting, no directional shadows, for proportion and silhouette reference; identical head-to-body ratio across all three, feet on the same ground line, headroom and footroom left clear; PROPORTIONS ARE CRITICAL, never stretch or squash the figure to fit. RIGHT-BOTTOM ZONE: four to five small isolated close-up studies in a row — the worn firewood knife and rope belt, straw stuck in messy hair, patched elbow and knee of the rags, round bright-eyes expression study, dusty bare feet; clearly smaller than the full-body figures; if they do not fit, extend downward along the right edge — the detail studies give way, not the figures. LIGHTING IN THE RIGHT ZONES: flat even orthographic lighting, no directional key, no cast shadows. ONE consistent face across all zones. Semi-realistic character illustration, painterly rendering with soft blended edges and visible brush texture, anatomically grounded"
  },
  {
    "id": null,
    "name": "玄真子",
    "importance": "supporting",
    "oneLiner": "蛰伏剑宗二十年的副掌门，一身紫袍、满脸和善，背地里勾结血河宗，那句『跪了四百年，跪够了』是他的反叛宣言。",
    "hasSheetPrompt": true,
    "hasBasePrompt": true,
    "negativePrompt": "plastic waxy skin, over-smoothed skin, poreless doll face, perfectly symmetrical face, dead eyes without highlights, helmet-like rigid hair, flat textureless fabric, stiff mannequin pose, neon colors, modern clothing, western features, cartoon villain cliche",
    "sheetPrompt": "Single character model sheet on ONE 16:9 landscape canvas, pure white background with even margins, divided by thin hairline rules into three zones. LEFT ZONE (about 34% of the canvas width): one large head-and-shoulders bust portrait, front-facing, centered like an ID photo, both shoulders fully visible, bottom edge cut flat; the face-design baseline — a Han Chinese sect deputy master in ancient Chinese cultivation-era purple silk robe, high bun with jade hairpin, trimmed goatee, faint cold smile. LIGHTING IN THE LEFT ZONE ONLY: soft directional key light from upper left, natural falloff, ambient occlusion under the chin, eye sockets and collar-neck junction. RIGHT-TOP ZONE: three full-body views (front, side, back) standing in a row on one shared ground line, same face, hair and expression as the left bust; flat even orthographic lighting, no directional shadows, for proportion and silhouette reference; identical head-to-body ratio across all three, feet on the same ground line, headroom and footroom left clear; PROPORTIONS ARE CRITICAL, never stretch or squash the figure to fit. RIGHT-BOTTOM ZONE: four to five small isolated close-up studies in a row — the jade hairpin and high bun, the trimmed goatee and faint smile, the dark brocade collar trim, the jade sect token, long sleeves with hands clasped behind the back; clearly smaller than the full-body figures; if they do not fit, extend downward along the right edge — the detail studies give way, not the figures. LIGHTING IN THE RIGHT ZONES: flat even orthographic lighting, no directional key, no cast shadows. ONE consistent face across all zones. Semi-realistic character illustration, painterly rendering with soft blended edges and visible brush texture, anatomically grounded"
  },
  {
    "id": null,
    "name": "掌门",
    "importance": "supporting",
    "oneLiner": "玄霄剑宗掌门，手中握着与沈砚胸前一模一样的另一半玉佩——魔渊之钥的另一半；卷末于藏经阁窗前道破天机，预言这剑宗的天要变了。",
    "hasSheetPrompt": true,
    "hasBasePrompt": true,
    "negativePrompt": "plastic or waxy skin, over-smoothed airbrushed complexion, poreless doll face, perfectly symmetrical face, dead flat eyes without specular highlight, helmet-like hair with no loose strands, flat untextured fabric with no weave or wear, stiff mannequin posing, extra fingers, malformed hands, text, watermark, signature, busy or patterned background, harsh cast shadows on the backdrop",
    "sheetPrompt": "Single character model sheet on ONE 16:9 landscape canvas, thin hairline rules dividing the canvas into three zones, clean white background with even margins all around. Semi-realistic character illustration, painterly rendering with soft blended edges and visible brush texture, anatomically grounded. Han Chinese, ancient Chinese cultivation-era (xianxia) setting: a solemn middle-aged sword-sect master, about fifty-five, dark indigo-azure layered robe with faint gold damask, broad sash, dark hair in a topknot with a plain jade hairpin, thin white-streaked beard. LEFT ZONE occupying about 34% of the canvas width: one large bust portrait, head and shoulders centered like an ID photo, both shoulders fully visible, bottom edge cropped straight and level — the reference face, fully detailed: composed weathered features, deep nasolabial folds, faint crow's feet, quiet authoritative eyes, pale green jade half-disc held at chest. RIGHT-TOP ZONE: full-body front / side / back three views standing in a row on one shared ground line, identical head-to-body proportions and limb lengths, feet on the same ground line, headroom above and footroom below, PROPORTIONS ARE CRITICAL — never stretch or squash the figure; the figures keep their true proportions and the detail studies give way, not the figures. RIGHT-BOTTOM ZONE: four to five small isolated close-up detail studies in an evenly spaced row, noticeably smaller than the full-body figures: the jade half-disc, the jade hairpin, the gold damask weave of the robe, the sash knot, the worn hems and sleeves. The right-zone faces must match the left-zone reference exactly: same face, same topknot, same beard, same expression. LIGHTING IN THE LEFT ZONE ONLY: a soft directional key light from the upper left with gentle falloff, subtle ambient occlusion under the chin, in the eye sockets and where the collar meets the neck, giving the head real volume. LIGHTING IN THE RIGHT ZONES: flat even orthographic lighting with no directional key and no cast shadows, so the figures stay measurable and cleanly cut out"
  },
  {
    "id": null,
    "name": "巫崖",
    "importance": "supporting",
    "oneLiner": "南疆巫王之子，练气巅峰的少主，起初轻佻嚣张地来剑宗迎亲，最终在魔渊战场率数千南疆战士反水替南疆还债，并年年给青石镇义庄送茶。",
    "hasSheetPrompt": true,
    "hasBasePrompt": true,
    "negativePrompt": "plastic or waxy skin, over-smoothed airbrushed complexion, poreless doll face, perfectly symmetrical face, dead flat eyes without specular highlight, helmet-like hair with no loose strands, flat untextured fabric with no weave or wear, stiff mannequin posing, extra fingers, malformed hands, text, watermark, signature, busy or patterned background, harsh cast shadows on the backdrop",
    "sheetPrompt": "Single character model sheet on ONE 16:9 landscape canvas, thin hairline rules dividing the canvas into three zones, clean white background with even margins all around. Semi-realistic character illustration, painterly rendering with soft blended edges and visible brush texture, anatomically grounded. Han Chinese, ancient Chinese cultivation-era (xianxia) setting, southern Nanjiang frontier: a brash teenage warrior, about seventeen, sun-darkened skin, fur-and-hide mantle over bare tanned arms, tribal indigo tattoos on shoulders and upper chest, leather cord necklace with bone beads, curved waist knife. LEFT ZONE occupying about 34% of the canvas width: one large bust portrait, head and shoulders centered like an ID photo, both shoulders fully visible, bottom edge cropped straight and level — the reference face, fully detailed: sharp bold eyebrows, sly confident grin, lively arrogant eyes, indigo tattoo marks on the bare collarbones. RIGHT-TOP ZONE: full-body front / side / back three views standing in a row on one shared ground line, identical head-to-body proportions and limb lengths, feet on the same ground line, headroom above and footroom below, PROPORTIONS ARE CRITICAL — never stretch or squash the figure; the figures keep their true proportions and the detail studies give way, not the figures. RIGHT-BOTTOM ZONE: four to five small isolated close-up detail studies in an evenly spaced row, noticeably smaller than the full-body figures: the indigo shoulder tattoos, the curved knife with wrapped grip, the bone-bead necklace, the fur-and-hide mantle clasp, the worn leather boots. The right-zone faces must match the left-zone reference exactly: same face, same grin, same hair, same expression. LIGHTING IN THE LEFT ZONE ONLY: a soft directional key light from the upper left with gentle falloff, subtle ambient occlusion under the chin, in the eye sockets and where the collar meets the neck, giving the head real volume. LIGHTING IN THE RIGHT ZONES: flat even orthographic lighting with no directional key and no cast shadows, so the figures stay measurable and cleanly cut out"
  },
  {
    "id": null,
    "name": "陆明姝",
    "importance": "supporting",
    "oneLiner": "前朝末代公主、血河宗前任圣女，为护子叛出魔道、隐姓埋名藏身义庄——以命换命，把半块玉佩与梅花帕留给了沈砚。",
    "hasSheetPrompt": true,
    "hasBasePrompt": true,
    "negativePrompt": "plastic waxy skin, over-smoothed airbrushed skin, poreless doll face, perfectly symmetric face, dead lifeless eyes, helmet-like hair without flyaways, flat textureless fabric, stiff mannequin pose, extra fingers, deformed hands, watermark, text, logo, cartoon, flat vector, chibi, western features, modern clothing",
    "sheetPrompt": "Single character model sheet on ONE 16:9 landscape canvas, plain white background, thin hairline rules dividing three zones. LEFT ZONE about 34% of the canvas width: one large front-facing bust portrait, head and shoulders centered like a passport photo, both shoulders fully visible, bottom edge cut flat, face fully rendered in detail as the facial design baseline. RIGHT-TOP ZONE: three FULL-BODY views (front / side / back) side by side on one shared ground line, equal height, consistent head-to-body ratio, feet on the same ground line, headroom and footroom above and below. RIGHT-BOTTOM ZONE: a row of four to five small isolated close-up studies: the plum-blossom handkerchief, the jade half-disc pendant, the mended grey cotton sleeve, the simple hairpin, the plain cloth shoes. Three zones separated by thin hairline rules; the figures never stretched or flattened, the detail studies give way, not the figures. LIGHTING IN THE LEFT ZONE ONLY: soft directional key from upper left, natural falloff, ambient occlusion under the chin, in the eye sockets, and at the collar. LIGHTING IN THE RIGHT ZONES: flat even orthographic lighting, no directional key, no shadows. ONE consistent face across the bust portrait and the three views, same features, same hairstyle, same expression. Han Chinese woman in her late twenties, ancient Chinese cultivation-era (xianxia) rural funeral-parlor setting, plain grey cotton clothing, not western, no text, no watermark. Semi-realistic character illustration, painterly rendering with soft blended edges and visible brush texture, anatomically grounded"
  },
  {
    "id": null,
    "name": "外门师兄",
    "importance": "minor",
    "oneLiner": "剑宗外门两位练气弟子的合并形象（孙望/郑魁）：一个夺剑羞辱、一个擂台轻视，先后被绝脉凡人沈砚当众打脸的爽点工具人。",
    "hasSheetPrompt": true,
    "hasBasePrompt": true,
    "negativePrompt": "plastic or waxy skin, over-smoothed airbrushed complexion, poreless doll face, perfectly symmetrical face, dead flat eyes without specular highlight, helmet-like hair with no loose strands, flat untextured fabric with no weave or wear, stiff mannequin posing, extra fingers, malformed hands, text, watermark, signature, busy or patterned background, harsh cast shadows on the backdrop",
    "sheetPrompt": "Single character model sheet on ONE 16:9 landscape canvas, thin hairline rules dividing the canvas into three zones, clean white background with even margins all around. Semi-realistic character illustration, painterly rendering with soft blended edges and visible brush texture, anatomically grounded. Han Chinese, ancient Chinese cultivation-era (xianxia) setting: an arrogant young sword-sect disciple, early twenties, clean pale grey-blue sect robe with a plain cloth belt, wide-bladed sword slung across his back. LEFT ZONE occupying about 34% of the canvas width: one large bust portrait, head and shoulders centered like an ID photo, both shoulders fully visible, bottom edge cropped straight and level — the reference face, fully detailed: sharp nose, sneering mouth, tilted chin, raised disdainful brows, mocking half-lidded eyes. RIGHT-TOP ZONE: full-body front / side / back three views standing in a row on one shared ground line, identical head-to-body proportions and limb lengths, feet on the same ground line, headroom above and footroom below, PROPORTIONS ARE CRITICAL — never stretch or squash the figure; the figures keep their true proportions and the detail studies give way, not the figures. RIGHT-BOTTOM ZONE: four to five small isolated close-up detail studies in an evenly spaced row, noticeably smaller than the full-body figures: the wide sword blade with its plain guard, the cloth belt knot, the grey-blue robe weave and collar, the rolled cuffs, the sneering smirk expression study. The right-zone faces must match the left-zone reference exactly: same face, same hair, same smug expression. LIGHTING IN THE LEFT ZONE ONLY: a soft directional key light from the upper left with gentle falloff, subtle ambient occlusion under the chin, in the eye sockets and where the collar meets the neck, giving the head real volume. LIGHTING IN THE RIGHT ZONES: flat even orthographic lighting with no directional key and no cast shadows, so the figures stay measurable and cleanly cut out"
  },
  {
    "id": null,
    "name": "老周头",
    "importance": "minor",
    "oneLiner": "杂役院厨房烧火的老头，年轻时是外门弟子、伤了根基跌回凡人，七十岁仍能单手提两桶水，一句话点破养气诀真意\"练的不是气，是定\"。",
    "hasSheetPrompt": true,
    "hasBasePrompt": true,
    "negativePrompt": "plastic or waxy skin, over-smoothed airbrushed complexion, poreless doll face, perfectly symmetrical face, dead flat eyes without specular highlight, helmet-like hair with no loose strands, flat untextured fabric with no weave or wear, stiff mannequin posing, extra fingers, malformed hands, text, watermark, signature, busy or patterned background, harsh cast shadows on the backdrop",
    "sheetPrompt": "Single character model sheet on ONE 16:9 landscape canvas, thin hairline rules dividing the canvas into three zones, clean white background with even margins all around. Semi-realistic character illustration, painterly rendering with soft blended edges and visible brush texture, anatomically grounded. Han Chinese, ancient Chinese cultivation-era (xianxia) setting: a wiry elderly kitchen elder, about seventy, patched coarse grey cotton tunic with a cloth apron and rolled sleeves, thin grey-white braid tied at the back of the head, fire iron in hand. LEFT ZONE occupying about 34% of the canvas width: one large bust portrait, head and shoulders centered like an ID photo, both shoulders fully visible, bottom edge cropped straight and level — the reference face, fully detailed: leathery flushed skin, sparse eyebrows, kind crinkled eyes, deep laugh lines, age spots, the grey-white braid visible behind the ear. RIGHT-TOP ZONE: full-body front / side / back three views standing in a row on one shared ground line, identical head-to-body proportions and limb lengths, feet on the same ground line, headroom above and footroom below, PROPORTIONS ARE CRITICAL — never stretch or squash the figure; the figures keep their true proportions and the detail studies give way, not the figures. RIGHT-BOTTOM ZONE: four to five small isolated close-up detail studies in an evenly spaced row, noticeably smaller than the full-body figures: the grey-white braid with its cord, the fire iron, the patched apron, the rolled sleeves with frayed cuffs, the calloused hands. The right-zone faces must match the left-zone reference exactly: same face, same braid, same kindly expression. LIGHTING IN THE LEFT ZONE ONLY: a soft directional key light from the upper left with gentle falloff, subtle ambient occlusion under the chin, in the eye sockets and where the collar meets the neck, giving the head real volume. LIGHTING IN THE RIGHT ZONES: flat even orthographic lighting with no directional key and no cast shadows, so the figures stay measurable and cleanly cut out"
  },
  {
    "id": null,
    "name": "盲眼琴师",
    "importance": "minor",
    "oneLiner": "王都听雨轩的盲眼乐正，琴声清越如雨打芭蕉，替故主守了一封遗信，最后用一条命替殿下守她的儿子。",
    "hasSheetPrompt": true,
    "hasBasePrompt": true,
    "negativePrompt": "plastic waxy skin, over-smoothed skin, poreless doll face, perfectly symmetrical face, perfectly round doll eyes, helmet-like hair, flat fabric without weave, mannequin stiff pose, distorted proportions, extra fingers, watermark, text",
    "sheetPrompt": "Single character model sheet on ONE 16:9 landscape canvas, thin hairline rules dividing it into three zones, pure white background with even margins. LEFT ZONE (about 34% of the canvas width): one bust portrait, head and shoulders, front view, centred like a passport photo, full shoulders visible, bottom edge cropped straight; this is the face-design reference — an elderly blind Han Chinese man, ancient Chinese cultivation era, fallen-dynasty capital of a river city. Clouded milk-white blind eyes, deep wrinkles, sparse white topknot, thin grey beard, faded indigo-grey hemp robe. LIGHTING IN THE LEFT ZONE ONLY: soft directional key light from upper left with natural falloff, ambient occlusion under the brow, jawline, collar and neck. RIGHT-TOP ZONE: three full-body views (front / side / back) on one shared ground line, identical proportions and face as the left bust, same clothing with visible fabric weave and believable wear; flat even orthographic lighting, no directional shadows, no cast shadows. RIGHT-BOTTOM ZONE: four to five small equal-spaced detail close-ups, clearly smaller than the full-body figures — clouded blind eye, thin callused hand over a guqin string, hemp robe weave and frayed sleeve edge, worn wooden zither. LIGHTING IN THE RIGHT ZONES: flat even orthographic lighting for clean cutout and proportion check. PROPORTIONS ARE CRITICAL: all three full-body figures equal height, same head-to-body ratio, correct limb lengths, feet on the same ground line, headroom and footroom even; never stretch or squash the figures to fit; the detail studies give way, not the figures. Semi-realistic character illustration, painterly rendering with soft blended edges and visible brush texture, anatomically grounded"
  },
  {
    "id": null,
    "name": "守渊老者",
    "importance": "minor",
    "oneLiner": "守渊一脉驻地的白发老者，等了一千年等来两玉合璧，又拼尽老命替来者断后。",
    "hasSheetPrompt": true,
    "hasBasePrompt": true,
    "negativePrompt": "plastic waxy skin, over-smoothed skin, poreless doll face, perfectly symmetrical face, dead lifeless eyes without highlights, helmet-like hair, flat fabric without weave, mannequin stiff pose, distorted proportions, extra fingers, watermark, text",
    "sheetPrompt": "Single character model sheet on ONE 16:9 landscape canvas, thin hairline rules dividing it into three zones, pure white background with even margins. LEFT ZONE (about 34% of the canvas width): one bust portrait, head and shoulders, front view, centred like a passport photo, full shoulders visible, bottom edge cropped straight; this is the face-design reference — an ancient white-haired Han Chinese ascetic elder, ancient Chinese cultivation era, keeper of a sealed abyss beneath a desolate western wasteland. Long white hair and beard, drooping eyelids, clouded piercing eyes, gaunt face, coarse grey-white hemp daoist robe with a faded sash. LIGHTING IN THE LEFT ZONE ONLY: soft directional key light from upper left with natural falloff, ambient occlusion under the brow, jawline, collar and neck. RIGHT-TOP ZONE: three full-body views (front / side / back) on one shared ground line, identical proportions and face as the left bust, same clothing with visible fabric weave and believable wear; flat even orthographic lighting, no directional shadows, no cast shadows. RIGHT-BOTTOM ZONE: four to five small equal-spaced detail close-ups, clearly smaller than the full-body figures — clouded piercing eye, long white beard texture, coarse robe weave and faded sash, aged callused hand. LIGHTING IN THE RIGHT ZONES: flat even orthographic lighting for clean cutout and proportion check. PROPORTIONS ARE CRITICAL: all three full-body figures equal height, same head-to-body ratio, correct limb lengths, feet on the same ground line, headroom and footroom even; never stretch or squash the figures to fit; the detail studies give way, not the figures. Semi-realistic character illustration, painterly rendering with soft blended edges and visible brush texture, anatomically grounded"
  },
  {
    "id": null,
    "name": "执法长老",
    "importance": "minor",
    "oneLiner": "奉掌门令围杀沈砚的剑宗执法长老，面容阴鸷，一柄赤色长剑，只认掌门令不认人。",
    "hasSheetPrompt": true,
    "hasBasePrompt": true,
    "negativePrompt": "plastic waxy skin, over-smoothed skin, poreless doll face, perfectly symmetrical face, dead lifeless eyes without highlights, helmet-like hair, flat fabric without weave, mannequin stiff pose, distorted proportions, extra fingers, watermark, text",
    "sheetPrompt": "Single character model sheet on ONE 16:9 landscape canvas, thin hairline rules dividing it into three zones, pure white background with even margins. LEFT ZONE (about 34% of the canvas width): one bust portrait, head and shoulders, front view, centred like a passport photo, full shoulders visible, bottom edge cropped straight; this is the face-design reference — a middle-aged Han Chinese cultivator, ancient Chinese cultivation era, stern law-enforcement elder of a sword sect. Gaunt face, high cheekbones, hooded cold eyes, thin short beard, dark indigo-black sect elder robe with silver cloud-and-sword trim, crimson longsword at his side. LIGHTING IN THE LEFT ZONE ONLY: soft directional key light from upper left with natural falloff, ambient occlusion under the brow, jawline, collar and neck. RIGHT-TOP ZONE: three full-body views (front / side / back) on one shared ground line, identical proportions and face as the left bust, same clothing with visible fabric weave and believable wear; flat even orthographic lighting, no directional shadows, no cast shadows. RIGHT-BOTTOM ZONE: four to five small equal-spaced detail close-ups, clearly smaller than the full-body figures — cold narrowed eye, short beard and metal hairpin, silver cloud-sword robe trim, crimson sword hilt with worn grip. LIGHTING IN THE RIGHT ZONES: flat even orthographic lighting for clean cutout and proportion check. PROPORTIONS ARE CRITICAL: all three full-body figures equal height, same head-to-body ratio, correct limb lengths, feet on the same ground line, headroom and footroom even; never stretch or squash the figures to fit; the detail studies give way, not the figures. Semi-realistic character illustration, painterly rendering with soft blended edges and visible brush texture, anatomically grounded"
  },
  {
    "id": null,
    "name": "杂役少年",
    "importance": "minor",
    "oneLiner": "百年后玄霄剑宗的砍柴杂役，从四十二刀砍倒一棵树开始，拔起听雪剑，去守那扇别人替他守了一百年的门。",
    "hasSheetPrompt": true,
    "hasBasePrompt": true,
    "negativePrompt": "plastic waxy skin, over-smoothed skin, poreless doll face, perfectly symmetrical face, dead lifeless eyes without highlights, helmet-like hair, flat fabric without weave, mannequin stiff pose, distorted proportions, extra fingers, watermark, text",
    "sheetPrompt": "Single character model sheet on ONE 16:9 landscape canvas, thin hairline rules dividing it into three zones, pure white background with even margins. LEFT ZONE (about 34% of the canvas width): one bust portrait, head and shoulders, front view, centred like a passport photo, full shoulders visible, bottom edge cropped straight; this is the face-design reference — a thin Han Chinese teenage boy of twelve or thirteen, ancient Chinese cultivation era, wood-cutting servant boy at a mountain sword sect. Messy short black hair, bright eager eyes, gap-toothed grin, washed-out grey cloth short tunic, straw sandals, chipped wood-cleaver strapped across his back. LIGHTING IN THE LEFT ZONE ONLY: soft directional key light from upper left with natural falloff, ambient occlusion under the brow, jawline, collar and neck. RIGHT-TOP ZONE: three full-body views (front / side / back) on one shared ground line, identical proportions and face as the left bust, same clothing with visible fabric weave and believable wear; flat even orthographic lighting, no directional shadows, no cast shadows. RIGHT-BOTTOM ZONE: four to five small equal-spaced detail close-ups, clearly smaller than the full-body figures — bright eye with grin, callused hand gripping the cleaver, chipped blade edge, straw sandal weave, patched grey cloth tunic. LIGHTING IN THE RIGHT ZONES: flat even orthographic lighting for clean cutout and proportion check. PROPORTIONS ARE CRITICAL: all three full-body figures equal height, same head-to-body ratio, correct limb lengths, feet on the same ground line, headroom and footroom even; never stretch or squash the figures to fit; the detail studies give way, not the figures. Semi-realistic character illustration, painterly rendering with soft blended edges and visible brush texture, anatomically grounded"
  },
  {
    "id": null,
    "name": "血袍老者",
    "importance": "minor",
    "oneLiner": "血河宗筑基巅峰的冷血执行者，两场袭杀皆由他出手，夺玉伤人后功败垂成。",
    "hasSheetPrompt": true,
    "hasBasePrompt": true,
    "negativePrompt": "plastic waxy skin, over-smoothed skin, poreless doll face, perfectly symmetrical face, dead eyes without highlights, helmet-like hair without flyaways, flat textureless fabric, mannequin pose, blurry face, extra fingers, deformed hands, watermark, text",
    "sheetPrompt": "Single character model sheet on ONE 16:9 landscape canvas for an elderly Han Chinese male cultivator from ancient cultivation-era China, fictional Da Liang dynasty. Pure white background with even margins. LEFT ZONE, about 34% of the canvas width: one large bust portrait, head and shoulders, front view, centered like a passport photo, both shoulders fully visible, bottom edge cut flat; this is the face design reference. RIGHT-TOP ZONE: three FULL-BODY views of the same figure standing in a row on one shared horizon line, front, side and back, showing the blood-red robe with black trims, the high topknot and correct proportions. RIGHT-BOTTOM ZONE: a row of four to five small isolated close-up detail studies, clearly smaller than the full-body figures: the iron hairpin, the embroidered robe sleeve cuff, the gaunt hand with prominent knuckles, the sneering mouth with short grey beard, the worn boots. Separate the zones with thin hairline rules. PROPORTIONS ARE CRITICAL: all three full-body figures identical height and head-to-body ratio, limbs correctly proportioned, feet on the same horizon, headroom and footroom, never stretch or squash the figures. The detail studies give way, not the figures. One single consistent face across all zones. LIGHTING IN THE LEFT ZONE ONLY: soft directional key light from upper left with natural falloff and ambient occlusion under the chin, eye sockets and collar. LIGHTING IN THE RIGHT ZONES: flat even orthographic lighting, no directional shadows, for clean silhouette and proportion reading. Semi-realistic character illustration, painterly rendering with soft blended edges and visible brush texture, anatomically grounded"
  },
  {
    "id": null,
    "name": "说书人",
    "importance": "minor",
    "oneLiner": "开篇散播传闻、收尾讲《凡骨传》的老说书人，用醒木见证凡人长生的传说。",
    "hasSheetPrompt": true,
    "hasBasePrompt": true,
    "negativePrompt": "plastic waxy skin, over-smoothed skin, poreless doll face, perfectly symmetrical face, dead eyes without highlights, helmet-like hair without flyaways, flat textureless fabric, mannequin pose, blurry face, extra fingers, deformed hands, watermark, text",
    "sheetPrompt": "Single character model sheet on ONE 16:9 landscape canvas for an elderly Han Chinese male storyteller from ancient cultivation-era China, fictional Da Liang dynasty. Pure white background with even margins. LEFT ZONE, about 34% of the canvas width: one large bust portrait, head and shoulders, front view, centered like a passport photo, both shoulders fully visible, bottom edge cut flat; this is the face design reference. RIGHT-TOP ZONE: three FULL-BODY views of the same figure standing in a row on one shared horizon line, front, side and back, showing the faded grey-blue linen robe, the topknot and correct proportions. RIGHT-BOTTOM ZONE: a row of four to five small isolated close-up detail studies, clearly smaller than the full-body figures: the wooden storyteller's clapper block, the tea bowl, the wooden hairpin, the trimmed grey beard with a warm smile, the cloth shoes. Separate the zones with thin hairline rules. PROPORTIONS ARE CRITICAL: all three full-body figures identical height and head-to-body ratio, limbs correctly proportioned, feet on the same horizon, headroom and footroom, never stretch or squash the figures. The detail studies give way, not the figures. One single consistent face across all zones. LIGHTING IN THE LEFT ZONE ONLY: soft directional key light from upper left with natural falloff and ambient occlusion under the chin, eye sockets and collar. LIGHTING IN THE RIGHT ZONES: flat even orthographic lighting, no directional shadows, for clean silhouette and proportion reading. Semi-realistic character illustration, painterly rendering with soft blended edges and visible brush texture, anatomically grounded"
  }
]