Zac

Learning Japanese

[DeepBlue深藍 @DeepBlueAIX]

人妻感～
🐋🍉🥤🍋Prompt:
9:16竖版，超写实iPhone前置摄像头自拍，真实手机原片质感，像伴侣在家里随手记录的一瞬间。

25岁年轻东亚成年女性，长黑发自然披散，略微凌乱，成熟优雅气质，自然丰满身材，比例协调，身体线条符合真实人体结构。眼神放松温柔，带一点慵懒和自信，不刻意微笑。

坐在夏日客厅沙发或窗边，刚刚休息时拿起手机自拍，身体自然后靠，一只手伸向镜头，轻微手臂入镜，手机距离较近，24mm前置广角轻微变形，脸部略靠近镜头，构图随意，不对称，不像摆拍。

穿冰蓝色或薄荷绿色轻薄居家服，柔软面料，自然褶皱，真实穿着痕迹。

室内桌面有半杯柠檬茶，可口可乐，杯壁水珠，旁边随意放着切开的冰西瓜，窗边有盛夏阳光和生活杂物，营造真实居家午后氛围。

iPhone照片原片，自动曝光，真实白平衡，轻微HDR，高光自然溢出，阴影保留细节，真实皮肤纹理，轻微手机噪点，普通家庭环境，像朋友手机里的一张生活照片，没有影棚感，没有商业写真感，没有AI生成痕迹。

画面左下角保留低调艺术签名“DeepBlue”。

[极客杰尼 @seekjourney]

后悔知道晚了，通过我的 github 雷达发现一个宝藏仓库：video-shotcraft

106 个镜头配方、162 种风格、161 个动态预览，还把分镜、运镜、节奏、音效和 Remotion 实现都整理成了 Agent Skill。

最近一段时间折腾剪视频，做了一个宣传片 Skill，看来我的 HyperFrames Motion Director，又可以升级一轮了。

https://t.co/rv1hHnx7L8

我的宣传片 Skill ：https://t.co/fsF5TAXvz7

[WasifAI @doctorwasif]

Chatgpt image 2 on @openart_ai

prompt
{
"identity_ref": {
"source": "uploaded photo",
"lock": ["face", "eyes/nose/mouth", "skin tone", "hair (length/texture/color)", "accessories", "exact outfit"],
"rule": "no invented details"
},
"composition": {"format": "vertical 4:5/9:16", "figures": ["real subject", "ink-twin doodle"], "layout": "side by side, poses fully visible"},
"real_subject": {"style": "photoreal, full-body", "outfit": "match reference exactly", "expression": "playful, embarrassed, caught-out charm", "pose": "natural, toned-down version"},
"ink_twin": {
"style": "rough black sketch/charcoal doodle on wall",
"recognizability": "silhouette only (hair/outfit/pose shape)",
"avoid": ["realistic 2nd person", "flat shadow", "color cartoon"],
"energy": "amplified, chaotic vs real subject",
"extras": "optional sparse comic marks (motion lines/star/heart)"
},
"pose_rule": {
"instruction": "new playful/mischievous shared pose each time",
"avoid": ["pointing", "finger-guns", "mirrored/repeated poses"],
"relationship": "subject = subtle version; twin = amplified cartoon version, same beat"
},
"setting": {"background": "clean minimal wall (white/cream/gray)", "lighting": "soft, even, daylight-style, face visible"},
"mood": "playful, stylish, chaotic-fun, shareable",
"negative_prompt": [
"invented appearance", "outfit changes", "twin as real 2nd person",
"flat/creepy shadow", "color cartoon twin", "weak resemblance",
"clutter", "repeated/stiff/mirrored poses", "text", "watermark",
"extra limbs", "bad hands"
]
}

[Fred-New @Fred834567]

想要靠信息差赚钱的兄弟全体注意

说GitHub只是放代码的，基本没在里面翻过三个小时。

实际上，上面的信息密度比任何技术社区都高一个量级。我列举了8个信息差套利方式和仓库：👇

1、远程工作信息差套利：awesome-remote-job
这个仓库包括了上百个远程工作投递网站，甚至涵盖了远程工作面试技巧、数字游民交流社群。
想在巴厘岛边度假边赚钱，就狠狠的阅读这个仓库中的信息。
🔗：https://t.co/44LfYJbveb

2、AI赚钱信息差套利: ai-money-maker-handbook
这个项目包含了非常多的AI赚钱思路和实操，比如AI出海、AI自媒体、AI技术等。
如果你还不知道如何用AI赚钱，可以从这个开始。
🔗：https://t.co/aZH1h02lNG

3、媒体信息差套利：TelegramGroup
这个仓库整理了上万个TG群组，很多一手信息只在TG上流通，相当于给你一张地图，告诉你去哪找。
🔗：https://t.co/YypYRju58P

4、内容创作信息差套利：awesome-list
搜“awesome+领域”就能找到别人整理好的顶级资源，做内容、找选题、挖素材，全是现成的。
🔗：https://t.co/KjeFQt2RJM

5、薅羊毛信息差套利：ai-goofish-monitor
这个仓库可以实时监控同一个闲鱼商品所有商家的最新动态，并根据AI算法推荐性价比最高且货真价实的商品，让你在闲鱼中快人一步，成为赢家。
🔗：https://t.co/D1W5WRH0To

6、自动化工具信息差套利：n8n
开源自动化工具，把重复工作搭成流程后自动跑，别人每天手动两小时，你一次性搞定。
🔗：https://t.co/bpnADwlrnf

7、数据信息差套利：awesome-public-datasets
收集了海量公开数据集，做分析、做产品、做报告，数据源直接给你摆好了。
🔗：https://t.co/Ssjtza7lTA

8、量化交易信息差套利：awesome-systematic-trading
非常权威的量化交易手册，包括不同平台上关于量化交易的资源库、书籍、策略、视频等。
🔗：https://t.co/A8EEVXXNKa

这些信息差套利渠道全部是真货。每个仓库都是我翻过、用过、确认能跑通的。

但说句实话——信息差从来不值钱，执行力才值钱。GitHub上这类资源多到看不完，真正能把一个仓库吃透、跑通、变现的人，千里挑一。

有些人看到的是仓库，有些人看到的是现金流。

收藏不丢人，只收藏不打开才可惜。希望这篇能推更多人一把，把想法变成行动。

[沐阳 @yyyole]

不要把图片发给Codex！
直接让它操作PhotoShop，这才是用Codex无损P图/调色的正确方式！

[逸尘 @gengdaJ]

大家如果感兴趣，也可以尝试去https://t.co/vohiaEiD8C 学习复刻一下这个片段，分镜片段完整提示词如下：

tyle: STOP-MOTION ANIMATION — stepped, frame-by-frame motion brought to a HAND-PAINTED 2D look, a moving oil painting, NOT clay, NOT puppets, NOT 3D. True 12 frames per second, ANIMATED ON TWOS: 12 distinct hand-painted drawings per second, each pose held two frames then snapping to the next, never gliding — even at full gallop the action steps pose to pose with the strobing cadence of hand-drawn animation. Constant painterly BOIL — every frame repainted by hand, brushstrokes and outlines subtly alive frame to frame. NO smooth interpolation, NO motion blur, NO morphing, NO liquid surfaces, NO melting, NO paint-splashes — surfaces solid, motion the redrawing of strokes; real frame-by-frame animation not AI slop. Style from <<<image_1>>> — match its rendering, texture, palette and mood exactly. RIDER 1 (the victor) from <<<image_2>>> on HIS horse from <<<image_3>>>. RIDER 2 (the victim) from <<<image_4>>> on HIS horse from <<<image_5>>>. Each rider stays on his own horse as a consistent pair, never mixed, never swapped. The wolves from <<<image_6>>>. The dusk snowfield rendered from description: a pale storm-lit snow plain at dusk in a blizzard. About SIX riders and about TEN wolves; only RIDER 1 and RIDER 2 are individual characters, the other ~4 riders are anonymous silhouettes (no inputs) and the rest of the pack is a varied dark flowing mass. HYBRID: brief FIRST-PERSON POV beats on RIDER 1's grab-and-knife and on RIDER 2's bite, external for everything else. Blizzard haze and blowing snow drift SMOOTHLY; figures, horses, wolves and drawn snow-spray step on twos with secondary action.

DIRECTOR'S NOTES:

THE SCENE — a kok-boru-style wolf hunt at full gallop in the dusk blizzard, brutal and fast. A team of about SIX mounted riders works TOGETHER to corner and drive a pack of about TEN wolves, herding and cutting them off like kok-boru players. Two fates play out side by side: RIDER 1 (the victor) corners a wolf, leans from the saddle, seizes it by the scruff, and as it thrashes he KILLS it with a knife. Nearby, RIDER 2 (the victim) charges another wolf and grabs it — but it twists and savages his arm, a second wolf slams him, and the pack tears him from the saddle and swarms him. Victory and death at once. REAL PHYSICS and a visible causal chain — every wound has its on-screen cause.
KOK-BORU TEAMWORK — the riders move as a coordinated team, the horsemanship and group tactics of kok-boru turned on the wolves: they gallop in a loose formation, herding the pack, riders flanking and cutting off wolves, wheeling their horses, driving the dark stream of wolves across the snow, shouting to each other. The teamwork is readable — this is a coordinated hunt, not chaos, even in the storm.
READABILITY HIERARCHY (so the mass does not become mush) — keep it legible: RIDER 1 on horse <<<image_3>>> and RIDER 2 on horse <<<image_5>>> are the two clear FOCAL characters, distinct and in focus on their own horses; the other ~4 riders are anonymous dark silhouettes flanking and herding (no individual detail); the ~10 wolves are a varied dark flowing mass, different coat tones, sizes and strides, NEVER mirrored, NEVER cloned. Stage the two focal riders apart and clear; let the rest read as team and pack around them, with the blizzard half-veiling the deeper ranks.
RIDER 1 — THE VICTOR — with the team driving a wolf to him, RIDER 1 leans LOW off the saddle in a kok-boru reach and SEIZES the wolf by the scruff; the wolf thrashes and twists, a heavy fighting animal; RIDER 1 hauls it up and DRIVES A KNIFE down into it in a hard stepped motion, the wolf jerking and going limp — a grim, decisive kill. REAL PHYSICS: the horse's weight, the wolf's fighting mass, the effort of the haul, the knife-thrust with force.
RIDER 2 — THE VICTIM — nearby, RIDER 2 charges another wolf and leans to grab it by the scruff; the wolf TWISTS its whole heavy body mid-stride and its jaws clamp CRUSHING onto his forearm, tearing sleeve and flesh, dark blood — cause before effect; a SECOND wolf launches from his blind side and slams full-body into his chest; torn from the saddle at gallop speed he hits the snow HARD and TUMBLES with momentum, his riderless horse <<<image_5>>> galloping on, and the pack converges as a fast dark stream and SWARMS over him, a savage thrashing dark mass. Brutality staged through motion, dark mass and sound, not slow anatomical detail.
POV CRAFT (brief first-person beats) — TWO short POV beats: (a) RIDER 1's grab-and-knife — we see through his eyes: the horse's mane and neck (from <<<image_3>>>) low in frame, our own chapan-sleeved arm seizing the thrashing wolf's scruff and our hand DRIVING THE KNIFE down into it at the lens; (b) RIDER 2's bite — through his eyes: the horse's mane (from <<<image_5>>>) and our chapan-sleeved forearm reaching for the wolf, the wolf twisting BACK and its jaws clamping CRUSHING onto our forearm RIGHT AT THE LENS, dark blood across the view, the wolf's muzzle filling the frame, the view wrenching off balance. In POV the camera IS the rider's head — pitching and bouncing with the gallop, snow flying into the lens, only his sleeved arms/hands seen, never his face. Everything else is external.
CAMERA LAW (LUBEZKI handheld craft) — every shot ONE continuous motivated handheld move, never static, never locked: external shots AGGRESSIVE HANDHELD racing alongside at gallop, violently jolting, wind-whipped, horizon tilting and never level, a hard DUTCH TILT as RIDER 2 goes down; POV beats subjective first-person handheld, the camera as the rider's head bouncing with the gallop, immersed in the storm with snow flying into the lens; natural dusk storm light only, palette from the scene; never gimbal-smooth, never tripod-locked, never mechanical.
COMPOSITION LAW (frame tells the story — by thirds, NOT centered) — SPACE: the wide storm-lit snowfield as negative space, the hunt strung across it. LINE: the riders driving from one side and the wolves streaming from the other as opposing DIAGONAL lines of dynamics, the converging chase; on RIDER 2's fall a reeling dutch tilt. SHAPE: dark angular galloping masses (riders, horses, wolves) against the pale snow; the pack a dark thrashing mass trapping RIDER 2. TONE: dark figures on pale storm-lit snow, the two focal riders the points of attention; dark muted blood the only color accent. BIG VS SMALL: RIDER 2 swallowed by the converging dark pack. Off-center, by thirds, never dead-centered.
FRAMED INK STAGING — read as masses: the team's dark galloping shapes and the wolves' dark flowing mass strung diagonally across the pale field; RIDER 1 hauling and knifing one wolf, RIDER 2 frame-trapped and engulfed by the swarming pack mass. CONCEAL: the blizzard half-veils the deeper riders and wolves, hiding the count, keeping it legible. CAMERA HEIGHT: in the chaos with them, low and racing; reeling on the fall.
SECONDARY ACTION on twos — chapan skirts and fur hats whipping, horses' manes and tails streaming, wolves' fur rippling, harness swinging, breath-vapor of horses and men torn off in the wind, snow bursting from hooves and impacts as drawn powder on twos; in POV the mane and our sleeves whipping, breath-vapor and snow flying past the lens.
LIGHT — deep dusk in a blizzard, dim cold blue-grey storm light, soft, no sun, NO rays, NO beams, NO god rays, NO volumetric shafts, the figures dark masses against the pale snow. CORRECT NEUTRAL WHITE BALANCE, NOT a blue filter, muted desaturated; the blood a dark muted red, stark on the snow but never glossy, never bright.

SHOT 1 — WIDE ESTABLISHING, ~35mm, aggressive handheld racing alongside. COMPOSITION (not centered, the team hunt): the team of ~6 riders driving in a loose kok-boru formation from the LOWER-RIGHT, the dark flowing mass of ~10 wolves streaming across toward the UPPER-LEFT, opposing diagonals of the chase, the pale storm-lit snowfield streaking as negative space, the blizzard half-veiling the deeper ranks. Riders flank and cut off the wolves, wheeling their horses, herding the pack, shouting over the wind; RIDER 1 (from <<<image_2>>> on <<<image_3>>>) and RIDER 2 (from <<<image_4>>> on <<<image_5>>>) readable among them on their own horses, manes and chapans whipping on twos, snow bursting from hooves. A coordinated hunt at full gallop. A hoarse shout across the team over the wind, lips on twos:

"Qorşa! Qamap tasta!"

HARD CUT to SHOT 2 — RIDER 1, the victor, ~50mm, aggressive handheld then a brief POV. The team drives a wolf to RIDER 1; he leans LOW off the saddle in a kok-boru reach and SEIZES it by the scruff, the wolf thrashing and twisting, a heavy fighting animal. Brief FIRST-PERSON POV — through his eyes: the horse's mane and neck (from <<<image_3>>>) low in frame, our chapan-sleeved arm gripping the thrashing wolf's scruff, and our hand DRIVING THE KNIFE down into it at the lens in a hard stepped motion — the wolf jerking and going limp. Back to external as he hauls the limp wolf up, grim and decisive, his horse galloping on. All on twos, real weight and force.

HARD CUT to SHOT 3 — RIDER 2, the victim, ~50mm, aggressive handheld then a brief POV. Nearby, RIDER 2 (from <<<image_4>>> on <<<image_5>>>) charges another wolf and leans to grab its scruff. Brief FIRST-PERSON POV — through his eyes: the horse's mane (from <<<image_5>>>) and our chapan-sleeved forearm reaching for the wolf, the wolf TWISTING its whole body BACK toward us and its jaws clamping CRUSHING onto our forearm RIGHT AT THE LENS, dark blood whipping across the view, the wolf's muzzle filling the frame, the view WRENCHING hard sideways and down off balance — and from the edge a SECOND dark wolf LAUNCHES toward the lens —

HARD CUT on the impact to SHOT 4 — EXTERNAL, MEDIUM WIDE with a hard DUTCH TILT, ~35mm, aggressive handheld. COMPOSITION (not centered): the second wolf slams full-body into RIDER 2's chest and he is TORN from the saddle at gallop speed, hitting the snow HARD and TUMBLING, rolling over with real momentum, powder snow bursting at each impact on twos, the horizon reeling, his riderless horse (<<<image_5>>>) galloping on into the storm. The pack pours in — a fast dark stream of wolves from <<<image_6>>>, varied coats, never mirrored, never cloned, converging and SWARMING over him, a savage thrashing dark mass on the pale snow, dark blood spreading, his cries swallowed by the blizzard and the snarling, the dark pack mass trapping and framing him; beyond, the rest of the team still gallops and drives the wolves through the storm. Hold one brutal beat in the howling wind. End.

Audio: NO MUSIC, no score, no instruments — only natural diegetic sound: the roar of the blizzard, pounding four-beat gallops, riders shouting to each other across the hunt, wolves snarling, the wet crunch of the bite, RIDER 2's raw scream, the heavy tumbling impacts, the converging snarls of the swarming pack, the wind swallowing everything. No subtitles. Absolutely no music.

Constraints: stop-motion stepped cadence on twos at a true 12fps with painterly boil even at full gallop, every action stepping pose to pose never gliding, NO smooth interpolation NO motion blur NO morphing NO liquid surfaces NO melting NO paint-splashes surfaces solid motion the redrawing of strokes, blizzard haze and blowing snow smooth while all figures animals and drawn snow-spray step on twos, hand-painted oil look from <<<image_1>>> match its rendering texture palette and mood not photoreal not 3D not glossy, KOK-BORU TEAMWORK about SIX riders working as a coordinated team herding flanking and cutting off about TEN wolves the horsemanship and group tactics of kok-boru turned on the pack a coordinated hunt not chaos, READABILITY HIERARCHY RIDER 1 from <<<image_2>>> on horse <<<image_3>>> and RIDER 2 from <<<image_4>>> on horse <<<image_5>>> the two clear FOCAL characters distinct and in focus on their own horses each pair never mixed never swapped the other ~4 riders anonymous dark silhouettes no inputs the ~10 wolves from <<<image_6>>> a varied dark flowing mass different tones sizes strides NEVER mirrored NEVER cloned the blizzard half-veiling the deeper ranks to keep it legible, TWO FATES SIDE BY SIDE — RIDER 1 THE VICTOR cornering a wolf leaning low in a kok-boru reach seizing it by the scruff hauling it up and DRIVING A KNIFE down into it the wolf jerking limp a grim decisive kill, RIDER 2 THE VICTIM charging another wolf grabbing it the wolf TWISTING its whole weight and clamping its jaws CRUSHING onto his forearm dark blood the second wolf slamming full-body into his chest tearing him from the saddle the hard tumbling fall at speed the pack SWARMING him as a savage dark mass — the bite before the blood the slam before the fall, HYBRID POV brief FIRST-PERSON beats — RIDER 1's grab-and-knife through his eyes horse mane from <<<image_3>>> low in frame our chapan-sleeved arm seizing the wolf's scruff our hand DRIVING THE KNIFE down at the lens; RIDER 2's bite through his eyes horse mane from <<<image_5>>> our chapan-sleeved forearm reaching the wolf twisting back jaws clamping crushing onto our forearm AT THE LENS dark blood the muzzle filling the frame the view wrenching off balance the second wolf launching at the lens — in POV the camera IS the rider's head pitching and bouncing with the gallop snow flying into the lens only sleeved arms/hands seen never the face everything else external the cut from POV to external on the second wolf's impact, CAMERA LAW LUBEZKI external aggressive handheld racing jolting wind-whipped horizon never level hard DUTCH TILT on the fall POV subjective first-person handheld the camera as the rider's head bouncing with the gallop immersed in the storm snow into the lens one continuous motivated move never gimbal-smooth never tripod-locked never mechanical natural light only, COMPOSITION LAW by thirds NOT centered the riders driving from one side and the wolves streaming from the other as opposing DIAGONAL lines of dynamics dark angular galloping masses against pale storm-lit snow the two focal riders the points of attention dark muted blood the only color accent the pack a dark thrashing mass trapping RIDER 2 BIG VS SMALL RIDER 2 swallowed by the converging dark pack off-center, FRAMED INK the team's dark galloping shapes and the wolves' dark flowing mass strung diagonally across the pale field RIDER 1 hauling and knifing one wolf RIDER 2 frame-trapped and engulfed by the swarming pack conceal the blizzard half-veiling the deeper ranks hiding the count camera height low and racing reeling on the fall, four shots with hard cuts — (1) WIDE establishing the team driving the pack opposing diagonals kok-boru teamwork "Qorşa! Qamap tasta!", (2) RIDER 1 the victor kok-boru reach and grab brief POV our arm and knife driving down killing the wolf, (3) RIDER 2 the victim charging and grabbing brief POV the wolf twisting back and the crushing bite at the lens the second wolf launching, (4) EXTERNAL dutch-tilted the slam tearing RIDER 2 from the saddle the hard tumbling fall the pack swarming him as a savage dark mass the team still hunting beyond — SECONDARY ACTION on twos chapan skirts fur hats manes tails wolf fur harness breath-vapor all whipping with follow-through snow bursting from hooves and impacts in POV mane and sleeves whipping snow and breath-vapor past the lens, deep dusk blizzard dim cold storm light soft NO rays NO beams NO god rays NO volumetric shafts CORRECT NEUTRAL WHITE BALANCE NOT a blue filter muted desaturated dark muted blood never bright never glossy, brutality fast hard and physical staged through motion dark mass and sound not lingering gore, spoken Kazakh in Latin transliteration pronounced as written not Russian-accented, NO MUSIC absolutely no music only storm hooves shouts snarls and screams.

[Smartpig @Smartpigai]

还在为“做什么产品”发愁？

这几个独立开发者宝藏库，能让你少走半年弯路：

1、看中国开发者正在做什么
https://t.co/0HSXJOV5qT
收录大量中国独立开发者及其产品，找灵感、看赛道、研究细分需求都很方便。

2、研究一人公司怎么赚钱
https://t.co/INMFHurs9e
真实产品、收入数据、失败复盘、创业想法和工具栈都整理好了。

3、拆解能盈利的小产品
https://t.co/5nHgvmBubW

4、独立开发和 AI 出海工具箱
https://t.co/ddTheFyy0S
从开发、设计、支付、部署到增长和产品发布，基本覆盖了完整流程。

5、中文独立开发者社区
https://t.co/caHrb6h1bW
可以发布作品、交流经验、找反馈，也能看看别人踩过哪些坑。

别只收藏。

挑一个真实痛点，7 天做出 MVP，先找到第一个用户，再考虑把产品做大。

[Ihtesham Ali @ihteshamali]

Palantir sells governments a war room that costs millions a year.

So, a guy named Elie just rebuilt it, put it on GitHub, and gave it away.

It's called World Monitor.

Open it and you get a live 3D globe with 500+ news feeds pouring in across 15 categories, all summarized by AI as they land.

> Military movements.
> Economic shocks.
> Natural disasters.
> Cyber incidents.
> Flight paths.
> Shipping lanes.

+ 56 different map layers you can stack on top of each other.

It scores 31 countries on a stress index and updates the number as things happen.

It watches 29 stock exchanges, commodities, and crypto in one panel.

It runs local AI through Ollama, so you can use the whole thing without a single API key.

Native desktop app for Windows, macOS, and Linux. 25 languages. Works out of the box after one clone.

[爱丽丝呀！ @BTCqzy1]

Codex 5.6 真的要让剪辑师失业了！

它现在的视频制作能力提升了不止一个档次，直接一句话生成视频！

重度实测一周，筛选出这 4 个Skill，直接实现全流程闭环：

1️⃣ HyperFrames （内置插件）

负责整片策划、动效生成、画面合成与最终渲染。它能把文字构思直接转化为高水准的动态视频框架，实现全流程闭环交付。

2️⃣ Remotion （内置插件）

基于 React 模板的视频生成工具。支持通过代码批量生产高质感视频，尤其适合数字人+动效结合的自媒体内容，实现真正的工业化生产。

3️⃣ video-use

专门处理真人素材的剪辑工具。能自动识别镜头、精准卡点并完成后期剪辑，让 AI 具备了职业级的剪辑逻辑。

地址：https://t.co/QqX1MJaemQ

4️⃣ Seedance

针对顶级视频大模型的提示词工具。专门生成电影级运镜和实拍感指令，补齐了 AI 视频在审美和镜头质感上的短板。

地址：https://t.co/r4RL7z744e

现在用 Codex 5.6 + 这4个Skill，从脚本→动效→剪辑→渲染，全程AI闭环。

[灰度笔记 @GrayNoteLab]

分享一组超欲御姐风AI女生提示词

生成效果绝绝子，你们想要她都有🙌🙌

提示词操作空间大🤓 记得回来交作业～

⬇️⬇️

【整体】：完整9:16竖版时尚杂志海报，三联Layout设计，上、中、下三张画面组成同一张作品，仅用细黑色分隔线区分，不要独立边框，中间第二张作为主视觉，人物头部、肩膀及前伸手臂自然突破第一道分隔线，形成明显Layer Break跨框效果，整体具有日本高端Editorial杂志排版质感。

【人物】：22岁东亚成年女生，鹅蛋脸，白皙细腻肌肤，自然粉嫩腮红，浅灰蓝色美瞳，大眼睛，柔和卧蚕，樱粉色嘴唇，长睫毛，深栗棕色长卷发，中分偏侧刘海，大波浪长发自然垂落胸前，真实皮肤纹理，真实摄影质感。

【服装】：修身黑色无袖短款连衣裙，高克重哑光弹力针织面料（Ponte Knit），高级服装纹理，局部黑色缎面滚边，细腻布料褶皱，绝不带乳胶感、绝不带丝袜材质反光，黑色半透明连裤袜，细跟黑色凉鞋配红色鞋底，整体高级时装风格。

【场景】：日系高端Laundry Café，高级奶白色滚筒洗衣机整齐排列，粉白渐变墙面，粉色霓虹环境灯，透明亚克力桌椅，玻璃桌面，透明装饰球，水晶珠帘，透明亚克力摆件，玻璃花瓶，镀铬金属细节，镜面地面，大面积背景虚化，空间干净通透。

【镜头】：24mm广角低机位拍摄，中间主画面距离镜头约50厘米，增强透视感，上方85mm人像特写，下方50mm标准镜头，电影级镜头语言。

【动作】：人物自然坐于透明亚克力方块，一只手轻触下巴，另一只手轻扶身体侧面，身体形成自然S形线条，双腿自然收拢放置，姿态舒展优雅，眼神平静直视镜头，神态自信高级，中间主画面头部、肩膀及前臂自然突破第一张画面，形成明显Layer Break。

【第一张】：面部近景特写，眼神直视镜头，轻触下巴，长卷发自然垂落，浅灰蓝美瞳在柔光下产生细微反光，背景虚化可见粉色霓虹与奶白色洗衣机、水晶珠帘形成梦幻散景。

【第二张】：主视觉广角画面，低机位拍摄，人物自然坐于透明亚克力方块，身体向镜头轻微靠近，头部、肩膀及前伸手臂突破第一张分隔线，完整展示服装轮廓、黑色连衣裙、连裤袜与高跟凉鞋，背景保留整齐排列的洗衣机、玻璃桌椅、水晶装饰与粉色灯光。

【第三张】：中景生活照，人物轻倚透明玻璃桌，轻轻整理发丝，微微侧身回望镜头，背景保留洗衣机、水晶吊饰、透明亚克力家具、粉色霓虹灯，空间层次丰富，具有INS生活方式摄影氛围。

【光影】：正面大型柔光箱作为主光源，脸部、肩颈与腿部均匀受光，人物轮廓加入柔和粉色边缘光，裙装呈现自然织物漫反射，镜面地面产生轻微倒影，背景粉色环境光与暖白灯融合，电影级体积光，空气中充满玻璃反射与柔焦光斑。

【动态】：轻微微风吹动发尾，耳边碎发轻轻摆动，裙摆自然形成布料褶皱，人物动作定格于凝视镜头瞬间。

【构图】：人物位于黄金分割点，前景广角透视，中景突出人物，背景利用洗衣机、透明家具、水晶吊饰形成丰富空间层次，整体统一为高级时尚杂志封面设计。

【色彩】：奶油粉、樱花粉、象牙白、雅黑为主色，Kodak Portra 400胶片色调，高光柔和，肤色自然通透，电影级色彩管理。

【摄影风格】：Japanese Editorial、VOGUE、Harper's Bazaar、LARME、Luxury Fashion Campaign、Photorealistic、Ultra Detailed、Natural Skin Texture、Cinematic Lighting、8K。

【Negative Prompt】：不要动漫，不要CG，不要AI塑料皮，不要乳胶材质，不要裙子产生丝袜反光，不要错误手指，不要肢体错位，不要文字，不要Logo，不要水印，不要独立边框，不要三张独立照片，不要普通拼图效果。

[Zac @explorroad]

@AndrewBolis YT

[自己修行的才谷 @realCaigu]

谷歌前 CEO 强调了一个热知识：

如果你真的想赚钱，必须学习：理解代理、Claude Code、提示、记忆、技能、MCP和例程

然后让你的 AI 来控制你的 AI

大部分人学 AI, 就是在推特刷 AI 帖子，但其实，真正系统性的知识，大模型公司官方出的全面且免费，务必收藏这条帖子：

代理架构 https://t.co/3s0hmJbyvw
Claude Code 101 https://t.co/MWXloXUljO
Claude Code 实战 https://t.co/LFoS19rqDO
提示工程 https://t.co/rvLTIbiC3X
互动提示课程 https://t.co/PkVrmel2zb
Claude.md + 记忆 https://t.co/dnc98oyv6O
技能 https://t.co/f7gI9A3jep
MCP https://t.co/gs3rSVu7HS
例程 https://t.co/bWzywsUZRY
Claude Code 终极指南 https://t.co/wVDvVgmKGB
Awesome Claude Code https://t.co/Z7VnEIeOTJ
Anthropic 学院 https://t.co/K276utEWhm
官方 Claude Code 文档 https://t.co/ab8BUYtjlx

现在，复制全文内容，并喂给你的 Agent!

[梭哈.AI @SUOHA_AI]

想用闲鱼赚钱搞副业？用这三个工具足够了👇

GitHub 上真正靠谱的闲鱼开源工具就这几个，别人还在手动刷页面、错过好货或累死累活回消息，你直接用这些工具：就能实现 24 小时智能秒杀好货+自动化捡漏、AI 自动筛选、AI 客服智能出单 + 多账号管理

1、闲鱼智能监控工具：ai-goofish-monitor

能实时/定时监控闲鱼商品，自动筛选产品，通过 AI 分析性价比和真实度，帮你快速发现低价好货，捡漏秒拍必备

🔗：https://t.co/8RafXPchzM

2、闲鱼 AI 智能客服机器人：XianyuAutoAgent

7×24 小时自动化回复/出单，支持多专家协同决策、智能议价和上下文感知对话。卖家自动处理买家询价、砍价和出单，大幅减少手动聊天时间

🔗：https://t.co/22pNYifkkZ

3、闲鱼多账号智能管理系统：xianyu-auto-reply-fix

支持多账号管理、AI 自动回复、自动发货确认、多渠道消息通知，还提供完整 Web 管理后台。适合系统化/矩阵化运营闲鱼店铺，提升出单效率

🔗：https://t.co/RZY6VKuPxj

这些工具社区活跃、功能成熟，大多支持 Docker 一键部署，上手友好

关注我，下集出手把手部署教学😼

[鸟哥 | 蓝鸟会🕊️ @NFTCPS]

书虫们注意了！你家那堆买来吃灰的技术书，终于有救了。

GitHub上一个叫book-to-skill的开源神器刚火，狂揽2700多Star，玩法是真离谱：

1️⃣ PDF或EPUB直接丢进去
2️⃣ 自动扒目录、核心概念和套路，一键生成技能
3️⃣ 以后敲一句 /书名技能 加话题，它就替你翻书答疑

说白了，吃灰的书秒变Claude Code里随叫随到的私人顾问，这波不亏👇

https://t.co/xGvqveUuAR

[Serena @369Serena]

Codex 中文内容创作者十大必装 Skills 来了！

写作、润色、去 AI 味、配图、封面、小红书卡片，一次给你配齐。

先收藏，再慢慢装，这 10 个一个比一个实用！

1⃣ Humanizer-zh

中文创作者必备的“去 AI 味”技能。

能识别空话、套话、三段式、过度排比、AI 腔，把文章改得更像真人写的。

Github 链接：

https://t.co/zncmfoVPun

2⃣ dbskill

内容创作者的“选题诊断器”。

适合做内容诊断、商业表达、hook、小红书标题、爆款拆解，尤其适合中文自媒体和知识型创作者。

Github 链接：

https://t.co/uMvxjMoDDl

3⃣ content-research-writer

从选题、资料、提纲、引用到初稿，一条写作工作流直接跑通。

适合写公众号、Newsletter、X 长文、博客和深度文章。

Github 链接：

https://t.co/qtMIT1SsIR

4⃣ NotebookLM Claude Code Skill

资料型写作者很适合。

基于你的资料库写作，能减少胡编乱造，适合做深度文章、行业研究、课程稿和长文整理。

Github 链接：

https://t.co/P6FWXBHT1A

5⃣ khazix-skills

适合公众号长文和万字研究报告。

里面的写作流比较重，适合做 AI 热点分析、长文拆解、深度观点输出。

Github 链接：

https://t.co/eFSOvz4FX7

6⃣ ian-xiaohei-illustrations

中文文章正文配图神器。

不是随便生成一张图，而是把文章里的观点、流程、情绪、隐喻画成“小黑”风格插图。

Github 链接：

https://t.co/Lvnudzb6Pm

7⃣ guizang-social-card-skill

写完文章后，用它做小红书图文和公众号封面。

很适合把一篇长文拆成多张卡片，做成能直接发布的视觉内容。

Github 链接：

https://t.co/THKxv4fg2e

8⃣ baoyu-skills

中文内容创作者的视觉工具箱。

封面图、信息图、结构图、图解都能做，适合文章包装和观点可视化。

Github 链接：

https://t.co/INNGs7Y813

9⃣ guizang-ppt-skill

适合把文章观点做成 PPT、演讲图、公众号头图、小红书封面。

如果你经常把文字内容二次分发成演示稿或分享图，这个很有用。

Github 链接：

https://t.co/FMY3GUKphP

🔟 html-anything

把 Markdown、文案、文章变成好看的 HTML 页面、海报、卡片和 PNG。

适合做杂志风文章、小红书卡片、知识卡片和分享图。

Github 链接：

https://t.co/uyh51sScDH

如果你是中文内容创作者，强烈建议先装这 5 个：

1. Humanizer-zh
2. dbskill
3. content-research-writer
4. ian-xiaohei-illustrations
5. guizang-social-card-skill

前 3 个提升文字质量，后 2 个解决文章配图和分发包装。
别再只让 AI 帮你“写一篇文章”了。

真正提效的是：让 AI 按你的创作流程干活。

#codex #skill #内容创作

[梭哈.AI @SUOHA_AI]

https://t.co/h5KZxj1HFM

[数字生命卡兹克 @Khazix0918]

https://t.co/R7jxglREBs

[路飞 🏴‍☠️ AI 研究员🧐 @0xluffy_eth]

有人开发了一个免费简历生成器

已经在GitHub上获得2.9K星标了

仓库：https://t.co/DIvEKTNycn

[tiantian @wherecall1]

别再刷抖音了。

给你 10 个真正值得刷的网站。

1. X（推特）
信息源头。所有趋势的起点。
https://t.co/mRlVGPfXNw

2. GitHub
免费资源+项目库。
https://t.co/YmnG0sbJpA

3. YouTube
最强学习平台，没有之一。
https://t.co/OuiJM5Pe3L

4. Reddit
真实世界的声音。信息密度极高。
https://t.co/cbI9ASXioX

5. Product Hunt
新机会雷达。AI 项目发源地。
https://t.co/l837QTRPMP

6. Hacker News
硅谷工程师聚集地。认知直接提升一个层级。
https://t.co/vbhOVZS1b6

7. 少数派
高质量效率与工具内容。比普通平台更深。
https://t.co/WZK203SOtD

8. 即刻
国内少有的高信息密度社区。小圈子价值很高。
https://t.co/SmLNbbS8yc

9. 闲鱼
一个神奇的网站，买什么东西都不会溢价。
https://t.co/cEFqMljOYZ

10. B站
系统学习入口。性价比最高。
https://t.co/cY8l5kn6HQ

有人说：我没资源。

错了。

资源一直都在。

只是你把时间花在了娱乐，而不是学习。

#信息源

[梭哈.AI @SUOHA_AI]

想让你的 AI Agent 看遍全网，却总是被 API Key、付费墙、403 报错 卡得死死的？

这个工具用一条命令，就能读取、搜索 Twitter/X、Reddit、Bilibili、小红书、YouTube、GitHub 等 15+ 平台的内容，彻底告别 API 密钥！ 彻底告别付费墙！ 彻底告别各种反爬限制！

GitHub 上已有 15.4K Stars 的开源工具：Agent Reach

它的优势有👇：

→ 支持 15+ 主流平台：直接读取 Twitter/X 推文、Reddit 帖子、Bilibili 视频、小红书笔记、YouTube 字幕、GitHub 仓库、微信公众号文章、抖音视频等

→ 零 API 费用：全部基于开源 CLI，无需任何 Key 或订阅

→ 一键安装：直接给 Agent 一条指令就能自动搞定

→ 智能绕过限制：本地 Cookie + 抓取，稳定读取受限内容 → 可插拔设计：每个平台独立，随时替换升级

→ 安全可靠：本地存储 Cookie，支持安全预览模式

它的核心机制是：全部基于开源免费的 CLI 工具，通过本地 Cookie + 抓取方式访问，完全不走官方付费 API，所以零费用、本地运行，数据只留在你自己电脑上

注意：Cookie 登录的平台（如 Twitter、小红书）存在被检测封号的风险，建议用小号操作，安全第一！！

🔗 GitHub:
https://t.co/XeI6xtpPaF