Skip to content

Litreily/codex-skill-eastern-beauty-director

eastern-beauty-director

Flagged

Director-level AI art prompt creation for Eastern beauty series, including Eastern fantasy / gu feng Chinese beauty character art, classical Eastern beauty, realistic and modern Eastern beauty, Cosplay展会摄影 / official convention cosplay photography with AAA game character campaign and Vogue editorial quality, 古风闺蜜 / AncientFemaleCompanionship two-person ancient Chinese female companionship photography, 东方美学图鉴 / 小红书图鉴 cover and content systems for 四大美人、四大才女、十二花神、东方神女、敦煌飞天、朝代服饰、东方器物、东方神话, 甜系纯欲生活写真 / SweetHomeGirl lifestyle portraits built around realism, feminine charm, romantic feeling, story moment, and natural attraction, new-Chinese-style fashion editorials, magazine portraits, worldbuilding, mythic role design, costume/styling systems, cinematic composition, lighting, negative prompts, and model-specific prompt variants. Use when the user asks for 东方美人, 东方审美, 东方美学图鉴, 小红书图鉴, 图鉴封面, 图鉴正文, 四大美人, 四大才女, 十二花神, 东方神女, 敦煌飞天, 朝代服饰, 东方器物, 东方神话, Cosplay展会摄影, coser, Coser, 漫展摄影, 展会摄影, ChinaJoy, Bilibili World, Tokyo Game Show, 游戏展台, 官方展会摄影, AAA游戏角色Cosplay, 古风闺蜜, AncientFemaleCompanionship, Ancient Female Companionship, 闺阁夜话, 姐妹古风写真, 古代闺阁生活, 灯下梳发, 共读, 夏夜纳凉, 守夜, 江南同行, 湖亭共伞, 山寺同行, 灯会夜游, 柔美, 妩媚, 成熟, 优雅, 高贵, 清冷, 温婉, 英气, 神性, 慵懒, 东方幻想, 古风美女, 古典东方美人, 宋韵, 唐宫, 江南, 洛神, 青瓷, 昆曲, 国风人物, 汉服女子, 写实东方美人, 现代东方审美, 新中式, 东方高级感, SweetHomeGirl, 甜系纯欲生活写真, 甜系纯欲, 甜美女友感, 真实女生, 自然抓拍, 故事感, 女友视角, lifestyle girlfriend portrait, 仙侠/武侠/宫廷/江南/敦煌风 AI 绘画提示词, image prompt optimization, style word expansion, character series concepts, or prompt packs for Midjourney, Stable Diffusion, ComfyUI, DALL-E, Gemini, Seedream, or other image generation tools.

FileKindSize
SKILL.mdskill md28.4 KB
agents/openai.yamlconfig1.1 KB
examples/Cosplay展会摄影-官方展台女武神.mdmarkdown6.1 KB
examples/daily/2026-06-11-daily-eastern-beauty.mdmarkdown11.4 KB
examples/东方珠宝大片-高贵成熟.mdmarkdown1.6 KB
examples/东方美学图鉴-上官婉儿人物页.mdmarkdown1.6 KB
examples/东方美学图鉴-卓文君人物页.mdmarkdown1.5 KB
examples/东方美学图鉴-四大才女封面.mdmarkdown2.1 KB
examples/东方美学图鉴-四大才女总览页.mdmarkdown972 B
examples/东方美学图鉴-李清照人物页.mdmarkdown1.5 KB
examples/东方美学图鉴-蔡文姬人物页.mdmarkdown1.6 KB
examples/凤阙宫火-凤凰女王.mdmarkdown2.7 KB
examples/唐宫夜宴-成熟妩媚.mdmarkdown1.7 KB
examples/唐宫花泉仪式-高贵柔美.mdmarkdown2.4 KB
examples/天象神朝-星图神官.mdmarkdown3.8 KB
examples/宋韵茶影-温婉优雅.mdmarkdown1.6 KB
examples/敦煌壁画-神谕飞天.mdmarkdown3.8 KB
examples/新中式杂志-雨窗美人.mdmarkdown4.8 KB
examples/昆仑雪剑-守桥剑姬.mdmarkdown2.7 KB
examples/月宫炼华-月镜炼丹师.mdmarkdown3.3 KB
examples/沧海龙庭-潮汐龙女.mdmarkdown2.7 KB
examples/甜系纯欲生活写真-人民广场.mdmarkdown6.3 KB
examples/甜系纯欲生活写真-推荐调用格式.mdmarkdown2.2 KB
examples/甜系纯欲生活写真-默认示例.mdmarkdown5.9 KB
examples/绮罗金帐-妩媚高贵.mdmarkdown1.9 KB
examples/绮罗金帐-宫廷秘仪.mdmarkdown3.7 KB
examples/莲梦镇邪-莲灯灵使.mdmarkdown2.7 KB
examples/雾浴私语-浴后美人.mdmarkdown4.4 KB
examples/青铜神谕城-甲骨战巫.mdmarkdown3.4 KB
references/character-subject-system.mdmarkdown13.1 KB
references/eastern-fantasy-series.mdmarkdown3.9 KB
references/global-negative-system.mdmarkdown3.9 KB
references/gufeng-visual-library.mdmarkdown3.2 KB
references/oriental-visual-discipline.mdmarkdown4.0 KB
references/prompt-framework.mdmarkdown5.4 KB
references/quality-control.mdmarkdown15.1 KB
references/request-router.mdmarkdown15.8 KB
references/structured-call-format.mdmarkdown10.5 KB
references/style-taxonomy.mdmarkdown7.1 KB
references/wow-factor-system.mdmarkdown5.6 KB
scripts/daily_eastern_beauty_generator.pyscript101.8 KB
scripts/run_daily_eastern_beauty_prompts.shscript1.3 KB

references/quality-control.md

View on GitHub
# Quality Control

## Avoid Generic Gufeng

Before finalizing, check:

- Is exactly one primary route driving the visual direction?
- Have explicit user parameters been preserved rather than replaced?
- Is the output more than a beauty portrait?
- Does the heroine have a readable fantasy role?
- Does the character have a clear face temperament and body language?
- Is the hand action meaningful rather than idle?
- Is there a memorable visual hook or artifact?
- Is there one dominant first-glance wow device?
- Can the image be summarized as one striking thumbnail shape?
- Does the environment imply a larger Eastern fantasy world?
- Is there a clear story moment?
- Is the clothing specific enough to visualize?
- Are the hands doing something plausible?
- Does the scene have spatial depth?
- Is the lighting named and motivated?
- Is the color palette restrained?
- Are modern/cosplay cues excluded?
- Does the face remain the first visual priority?
- Is there only one dominant wow device?
- Are jewelry, pibo, ribbons, particles, murals, embroidery, and background details controlled rather than competing?
- Does the image have visual breathing room?

If fewer than three of role, artifact, wonder setting, supernatural phenomenon, strict color system, story conflict, or wow device are present, upgrade the prompt before returning it.

If the prompt has many decorative details but no dominant shape or event, remove secondary ornaments and add a single oversized or impossible visual device.

If the prompt solves weak imagery by adding more gold, jewelry, floating particles, embroidery, ribbons, murals, or glow effects, rewrite it with `oriental-visual-discipline.md`: clean face priority, one wow device, controlled ornament density, premium color harmony, and subject-first composition.

## Oriental Visual Discipline

For 东方美人, 东方幻想古风, gu feng, Dunhuang, red-gold palace, mythic, jeweled, mural, candle, silk-curtain, or ornament-heavy routes, check:

- Is the viewer's first read the woman's face and eyes?
- Is the second read the story moment or action?
- Are costume, jewelry, pibo, ribbons, particles, murals, and background secondary?
- Is there one primary light logic instead of glow everywhere?
- Is the palette clean, with smooth color transitions and no oversaturated red/gold overload?
- Are there quiet areas or shadow areas that allow the subject to breathe?
- Does the image read as cinematic Eastern aesthetics rather than AI wallpaper, mobile game splash art, or fantasy poster clutter?

Use these positive controls as hidden prompt guidance when needed:

```text
cinematic oriental aesthetics, controlled ornament density, elegant visual hierarchy,
premium costume design, refined fabric layering, natural silk texture,
clean mineral color palette, smooth color transitions, subtle atmospheric depth,
harmonious warm lighting, soft facial illumination, natural skin rendering,
controlled jewelry density, balanced decorative elements, graceful fabric flow,
clean background treatment, selective detail emphasis, visual breathing room,
museum-quality color grading, high-end fantasy cinematography,
subject-first composition, less but better
```

Use these negatives for ornate Eastern fantasy routes:

```text
visual clutter, excessive particles, overdecorated costume, too many jewelry layers,
chaotic ribbons, excessive glowing effects, magic dust everywhere,
crowded composition, busy background, ornament overload, excessive embroidery,
oversharpened textures, artificial glow, floating random elements, visual noise,
multiple focal points, overly saturated red and gold, cheap game character design,
mobile game splash art, AI fantasy poster style, wallpaper style composition
```

### Dunhuang Route De-Noise

For `dunhuang-mural-oracle`, enforce:

- clean mural treatment
- abstract fresco motifs only
- controlled pigment flow
- minimal floating particles
- single wow device priority
- clean facial area
- preserve large quiet areas
- subtle fresco illumination
- soft desert atmosphere
- natural mineral pigment transitions

Avoid detailed miniature mural figures, crowded fresco people, excessive ribbons, pigment explosions, and belly-dance styling.

### Red-Gold Luo Curtain De-Noise

For `red-gold-luo-curtain`, enforce:

- controlled red-gold palette
- limited jewelry count
- soft candlelight only
- clean silk surfaces
- focus on face and fabric
- preserved shadow areas
- one jewel-chain halo or one curtain action, not many competing ornaments

Avoid excessive gold reflections, chaotic silk curtains, too many pearl chains, too many candles, random hanging ornaments, busy palace background, mobile game splash art, and AI fantasy poster style.

## Avoid Generic Eastern Beauty

For 东方美人, 东方审美, 柔美, 妩媚, 成熟, 优雅, 高贵, 清冷, 温婉, 英气, 神性, or 慵懒, check:

- Is one dominant beauty temperament driving the image?
- Does the prompt express beauty through gaze, posture, hand gesture, shoulder-neck line, hair rhythm, fabric behavior, light, and emotion?
- Is the woman still the visual subject rather than being swallowed by costume, props, or background?
- Does 妩媚 come from mature gaze, controlled posture, fabric, jewelry, candlelight, and atmosphere rather than exposure?
- Does 高贵 come from identity, ritual, symmetry, posture, and composition rather than random gold clutter?
- Does 成熟 come from composure, material weight, deep palette, and controlled expression rather than age exaggeration?
- Does 清冷 still have emotion in the eyes rather than a blank mannequin face?

If the result reads like "pretty woman + Eastern clothes + vague elegance," add a concrete gesture, a precise material, a motivated light source, and a specific emotional moment.

## Avoid Generic Classical Beauty

For 古典东方美人, 宋韵, 唐宫, 江南, 洛神, 青瓷, 昆曲, 仕女, or 闺阁, check:

- Is one classical route driving the image?
- Is the image classical without automatically becoming mythic fantasy?
- Are line, fabric, gesture, face temperament, light, and atmosphere doing the beauty work?
- Does the scene have cultural taste without turning into costume cosplay or tourist photography?
- Is gauze used as atmosphere or layering, not revealing fabric?

If the prompt is too plain, add one poetic visual device: rain reflection, water mirror, tea steam, porcelain moonlight, flower-window shadow, bamboo shadow, lantern in snow, or spring curtain light.

## Avoid Generic Modern Beauty

For 写实, 现代, 摄影, 杂志, 新中式, 东方高级感, check:

- Is one modern style route driving the prompt?
- Does the face have a temperament, structure, gaze, and natural skin texture?
- Is the clothing a specific modern or new-Chinese silhouette, not a vague "Chinese dress"?
- Does the posture or hand action reveal character without sexualized framing?
- Is the background tasteful and spatially clear?
- Are lens, light source, and palette named?
- Are cheap cosplay, heavy filters, plastic skin, transparent revealing fabric, and clutter excluded?

If the result reads like "pretty woman in Chinese clothes," upgrade styling, posture, light, and location before returning.

## SweetHomeGirl Consistency

For SweetHomeGirl prompts, check:

- Is 甜系纯欲生活写真 / SweetHomeGirl using the correct priority: body, story moment, pure-desire, face, outfit, composition, scene?
- Does the prompt follow the core formula: realism + feminine charm + romantic feeling + story moment + natural pure-desire?
- Is the body system hard lock placed first and protected from all other parameters?
- Does the prompt preserve young mature feminine body, healthy fullness, rounded natural curves, full natural bust, natural waist-hip transition, soft S-curve, full healthy legs, and no supermodel/fitness/bony/anime drift?
- Is there a real story moment, with story first, action second, expression last?
- Is pure-desire built from gaze, expression, action, outfit, atmosphere, and feminine curves rather than exposure, erotic suggestion, exaggerated pose, or deliberate seduction?
- Does the subject read as 20-26 adult young urban woman, not underage, childish, or student-girl?
- Is the face style varied without becoming a different universe or a fixed-face clone?
- Does the outfit feel like something a real woman would wear, not lingerie/swimsuit/gown/professional/stage/overexposed styling?
- Is composition treated as framing, not just aspect ratio: upper-body close-up, thigh-up, full body, or face close-up?
- Does the person occupy more than 60% of the frame, with expression/gaze first and background only supporting?
- Does the image avoid influencer photoshoot, fashion editorial, edge-bait, AI-sexy, static model display, and background-dominant feeling?
- Does the prompt keep realistic anatomy, natural proportions, correct hands, and natural body posture?
- Is the negative prompt protecting against body drift, age drift, story loss, static posing, influencer/fashion/AI-sexy feeling, excessive exposure, background dominance, deformed hands, plastic skin, and low-quality artifacts?

## Prompt Copy Blocks

When returning a final prompt for copy/paste, keep the final prompt and negative prompt in separate Markdown `text` code blocks. Do not mix notes, route labels, or analysis inside the copyable prompt block.

## Common Negative Prompt

Apply `references/global-negative-system.md` first. This local block is the
compact copy/paste version for ordinary prompt output.

```text
low quality, low resolution, blurry, jpeg artifacts, watermark, text, logo,
bad anatomy, malformed anatomy, wrong limb count, extra limbs,
extra arms, missing arms, extra legs, missing legs, broken limbs,
twisted joints, dislocated shoulders, broken wrists, deformed elbows,
deformed knees, unnatural torso twist, broken neck, neck too long,
bad hands, deformed hands, extra hands, missing hands, floating hands,
extra fingers, missing fingers, wrong finger count, fused fingers,
melted fingers, fingers fused with objects, hand fused with fabric,
bad feet, deformed feet, missing feet, extra feet, mutated toes,
distorted face, broken face, asymmetrical eyes, misaligned eyes,
duplicated face parts, melted facial features, plastic skin, over-smoothed skin,
modern clothing, cheap cosplay, zipper, sneakers, phone, microphone,
generic hanfu photoshoot, plain beauty portrait, western medieval armor,
japanese shrine, korean hanbok, random cyberpunk neon,
overexposed, oversaturated, messy background, duplicate person,
visual clutter, ornament overload, excessive particles, chaotic ribbons,
too many jewelry layers, excessive glowing effects, artificial glow,
multiple focal points, mobile game splash art, AI fantasy poster style,
object fused with face, scenery inside face, prop morphing, object mutation,
cup turning into chain, lantern becoming jewelry, book becoming fabric
```

## Daily Automation Failure Guard

For automated daily Eastern Beauty image batches, reject or regenerate candidates with any of these issues:

- The batch only covers familiar old routes and misses newer systems such as 东方美学图鉴, modern Eastern editorial, SweetHomeGirl, or exploratory Eastern beauty.
- Any subject looks older than 26, matronly, overly severe, or age-ambiguous.
- The face contains scenery, water, flowers, buildings, ornaments, cup shapes, chains, fabric patterns, or other non-face textures.
- A named prop mutates into another object, such as a cup becoming a chain, a bouquet becoming hair, a lantern becoming jewelry, or a fan becoming background texture.
- The prompt includes too many decorative nouns and no single dominant visual hook.
- The background steals attention from the face, eyes, and story moment.

Repair sequence:

1. Simplify the scene and reduce secondary ornaments.
2. Re-state `adult young East Asian woman age 20-26`.
3. Re-state the exact prop identity using "must remain a clear [object]".
4. Add face-integrity negatives: `scenery inside face, object fused with face, landscape on skin, distorted facial features`.
5. Regenerate with the same route only after preserving the original style, scene, outfit, action, and composition.

Daily publishing transaction rules:

- Use the Asia/Shanghai date for every `YYYY-MM-DD` value, post date, image directory, and asset reference.
- Create the daily route plan with `scripts/daily_eastern_beauty_generator.py` and follow that plan exactly. The automation must not choose the same old route families from memory.
- Every daily image must be `9:16`, including 东方美学图鉴 routes. Convert atlas routes into daily 9:16 vertical poster variants instead of using the normal 3:4 catalog ratio.
- For atlas/title images, do not trust the image model to create readable Chinese. Reserve a clean text zone and use post-processing overlay for required words such as `二十四节气` or a specific solar term.
- Never overwrite an existing `source/_posts/YYYY-MM-DD-daily-eastern-beauty.md` post during an automated run. If the post exists, stop and report that the daily post already exists.
- Generate and verify all five real image files before writing the final Hexo post. Prompt-only placeholders are not valid daily output.
- Keep draft parameters, prompt summaries, and QA notes in a staging location until all images exist and pass the minimum checks.
- If image generation, file writing, Hexo generation, image compression, or deploy fails, stop and do not leave a final post without matching images.
- Verify that every Markdown image path points to an existing file under `source/assets/ai/eastern-beauty/YYYY-MM-DD/`.
- Run Hexo validation, compression, deploy, git commit, and git push from WSL in the blog repo.
- Run image compression after `hexo generate` and before `hexo deploy` with Node.js 18 or newer. In this environment, prefer `/home/litreily/.nvm/versions/node/v22.22.3/bin/node` for `compress-images.mjs`.
- Commit only `source/_posts/YYYY-MM-DD-daily-eastern-beauty.md` and `source/assets/ai/eastern-beauty/YYYY-MM-DD/`. If `package.json`, `push.sh`, scripts, or unrelated files are modified, stop and report the unexpected dirty files.
- Push the blog source commit after deployment. Do not report the automation as complete until both `hexo deploy` and `git push` succeed.

## Safety and Taste

- Keep the character adult and avoid sexualized framing.
- Avoid transparent clothing used to reveal the body.
- Avoid "young girl", "schoolgirl", or age-ambiguous sensual language.
- For beauty-focused prompts, emphasize dignity, expression, styling, and cinematic craft.
- For red-gold palace boudoir or ornate pure-desire gu feng prompts, use candlelight, incense haze, silk layers, jewelry, carved screens, and dignified gaze; avoid foot-sole foreground, fetish framing, extreme contortion, exposed chest, lingerie, or transparent revealing fabric.
- For pure-desire or private-chamber prompts, build attraction through gaze, shoulder-neck line, wet hair, opaque silk folds, candlelight, steam, and composition. Do not use nudity, exposed chest, lingerie, voyeuristic framing, or transparent revealing clothing.

## Mini-Program Prompt Generator Fit

When the output will be used in a prompt app, provide modular fields:

```text
主题:
人物:
服饰:
场景:
光影:
镜头:
风格:
负面词:
```

Keep each field independently reusable.
brainstormingobra/superpowersYou MUST use this before any creative work - creating features, building components, adding functionality, modifying behavior, or planning anything new, in software or out of it (a talk, a business, a renovation).Stars: 322kUpdated Claude CodeCodexVerifiedInstall
brand-guidelinesanthropics/skillsApplies Anthropic's official brand colors and typography to any sort of artifact that may benefit from having Anthropic's look-and-feel. Use it when brand colors or style guidelines, visual formatting, or company design standards apply.Stars: 198kUpdated Claude CodeCodexVerifiedInstall
frontend-designanthropics/skillsGuidance for distinctive, intentional visual design when building new UI or reshaping an existing one. Helps with aesthetic direction, typography, and making choices that don't read as templated defaults.Stars: 198kUpdated Claude CodeCodexVerifiedInstall
mcp-builderanthropics/skillsGuide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or Node/TypeScript (MCP SDK).Stars: 198kUpdated Claude CodeCodexVerifiedInstall