![94vanAI's tweet photo. 是不是好多天沒發🍌香蕉prompt了？
馬上搞一個！
最近潛心研究Grok去了，補上🍌prompt：
[Character & Pose 👩🦰🌃]: Ultra close-up portrait photo, photograph by playforge,CCD + strong flash, a 19s top-tier beautiful cold K-pop young woman with high ponytail hairstyle, extremely fair cold white skin, sharp prominent nose, exquisite delicate features, refined makeup, lazily sitting on the stone steps in front of Tokyo Tower. Wearing extremely minimalist pure light color outfit: white large off-shoulder short knit cardigan + light pink-purple waist-cinching suspender tight long skirt (skirt tightly stretched and clinging to body curves, slight belly pooch, full bust), Van Cleef & Arpels red jewelry necklace + jade bracelet, transparent PVC one-strap high-heel sandals. Arms casually resting on knees, legs naturally apart, eyes with slightly distant aloof gaze looking toward camera, overall posture relaxed yet full of powerful presence.
[Scene & Lighting 🌙🗼]: Thick clouds partially covering the moon, dark windy night with scattered starlight points, vibe shot of vsc,moon positioned exactly at the top spire of Tokyo Tower creating soft hazy halo glow effect, light passing through the tower's lattice structure casting unique geometric shadows on the figure and stone steps.
[Composition & Atmosphere 🎥]: 4:9 vertical composition, design by 94van,extreme low worm's eye view from ground level仰拍, ultra wide angle capturing the complete Tokyo Tower in background (occupying 99% of background frame), forming strong visual correspondence with the foreground figure. Overall palette dominated by pure white tones, silent starry sky, and deep brown of the tower, creating high-end, cold, cinematic midnight Tokyo atmosphere.
[Style 📸]: Realistic CCD photo feel, low saturation soft tones, PF style,non-uniform natural lighting, dramatic chiaroscuro layers.
[Texture ✨]: Ultra high definition skin texture, premium retouching + soft glow, lifelike realistic quality, luxurious cinematic portrait aesthetics.
[Additional text ✍️]: On the right side of the image, 5% from the edge, in every quarter section of the frame, place one thin vertical handwritten small seal script "玖肆玩AI" (four places total); bottom right corner has extremely fine semi-transparent 3D crystal handwritten "PF".](https://pbs.twimg.com/media/HD74rzpa4AAPCCH.jpg)
![MANISH1027512's tweet photo. 动作指导古小一，限时回归
公开这组动作注入提示词，去给你的模特注入灵魂吧🤪
Realistic character editing, strictly use the identity reference image as the identity reference: keep the exact same face, facial feature details and proportions, skin tone, hairstyle (color/bangs/length/volume), clothing (silhouette/material/layering), and accessories completely unchanged (no face swap, no additional person, no changes to hairstyle/clothing/accessories).
[Task Goal]
Inject a dynamic pose with a shy, girlish feeling, and use a single giant close-up peace sign hand gesture as the visual focal point; while preserving identity and outfit, only modify the pose, camera structure, cropping logic, front-back layering, and foreshortening, without adding props, and only add slight shy facial expression changes that match the pose.
[Lens & Framing]
Vertical composition 9:16, an almost full-body portrait crop, camera positioned slightly above the subject from the front, angled downward about 25–40 degrees, not a straight top-down vertical shot, distance from the foreground hand approximately 25–45 cm, with an 18–24mm wide-angle feel, and the near-large far-small perspective must be obvious; the subject as a whole occupies about 84%–92% of the frame height, the head is located in the upper-middle area of the frame, and the rear lower leg may approach the bottom edge, but it must not become a distant standing full-body shot.
[Body Axis]
The main body axis runs diagonally from lower left to upper right, with the spine forming a soft C-curve; the head tilts about 10–18 degrees toward the left side of the frame, and the chin is gently lowered about 5–10 degrees; the ribcage is in a relaxed 3/4 front view; the pelvis rotates about 15–25 degrees away from the camera; the right shoulder is slightly brought forward and sits higher than the left shoulder; the overall temperament should be a suspended, shy-girl dynamic freeze-frame, light, soft, and slightly bashful, rather than aggressive or athletic.
[Upper-Limb Perspective]
The foreground right arm extends about 70–90 degrees toward the camera relative to the torso, the right elbow bends about 95–110 degrees, and the right forearm directly enters the camera plane; the right hand forms a huge peace sign in front of the lens, with the index and middle fingers extended into a V shape, the ring finger and little finger folded inward, and the thumb assisting in holding the folded fingers, the palm at a slight diagonal angle to the lens, each hand must have exactly five fingers, joints clearly defined, with no fusion.
The left upper arm is raised to the side of the head about 45–60 degrees, the left elbow bends about 70–90 degrees, and the left forearm remains near the plane of the face; the left hand stays in a small tucked hand pose against the face, positioned beside the cheek or temple, forming a compact inward-curled little claw / half-closed pose, with the fingers naturally bent inward, not a peace sign, and not a fully open palm, and it must be clearly smaller than the foreground right hand.
[Lower-Limb Perspective]
No clearly grounded supporting foot should appear; one thigh folds upward in front of the lower torso, with the knee bent about 70–100 degrees, while the other leg extends downward and drags backward; due to perspective, the lower body as a whole must be smaller than the torso and foreground hand, and the folded thigh may obscure about 10%–20% of the pelvis.
[Depth / Occlusion]
The depth-of-field hierarchy is: nearest foreground right-hand peace sign > right forearm > head and left hand > ribcage and torso > folded thigh and pelvis > rear lower leg/foot farthest away.
The right forearm should obscure about 10%–20% of the head and upper torso; the smaller left hand should slightly press against the contour of the face edge about 5%–10%; clear front-back layering must be preserved, and all limbs must not be flattened into the same plane.
[Hard Locks]
Only the foreground right hand is a peace sign, and it is the first priority, occupying about 35%–45% of the frame width.
The left hand against the face must remain in a small tucked hand pose and must not become a peace sign.
The 9:16 vertical composition with obvious slightly-above-front wide-angle perspective must be clear.
The head must be behind the foreground hand.
Preserve the overall diagonal suspended silhouette and the folded lower-limb relationship.
[Adaptable]
Background color, lighting atmosphere, material rendering, and environmental style may be freely adapted, but identity, outfit, 9:16 aspect ratio, camera structure, and action geometry must be locked.
[Background / Format]
Clean seamless background, real skin texture, with a slight shy expression, no heavy skin smoothing, no added props, no text, and no watermark.
Negative Prompt:
identity drift, face swap, different person, facial proportion changes, facial detail changes, skin tone changes, hairstyle changes, clothing changes, accessory changes, wrong camera angle, straight top-down vertical shot, low-angle upward shot, distant eye-level shot, telephoto flattening, perspective too weak, wrong cropping, wrong aspect ratio, non-9:16 composition, subject too small in frame, distant full-body shot, wrong orientation of head/ribcage/pelvis, symmetrical posed shot, stable standing pose, wrong center-of-gravity relationship, wrong limb direction, wrong joint angles, wrong planar relationships, foreground hand not large enough, both hands turned into peace signs, left hand turned into peace sign, foreground hand turned into open palm, foreground hand turned into fist, foreground hand turned into claw hand, incorrect peace sign, missing V-shaped gesture, wrong finger combination, wrong palm orientation, extra fingers, missing fingers, fused fingers, hand deformity, broken wrist, twisted wrist, broken elbow joint, dislocated shoulder joint, wrong front-back layering, head moving in front of foreground hand, no occlusion relationship between hand and face, floating limb misalignment, lost foreshortening, nearer limbs becoming smaller instead, feet emphasized, fierce expression, mature sexy atmosphere, overly strong athletic feeling, background clutter, extra props, text, watermark, logo, line-art style, sketch style, cartoon, anime, 3D render](https://pbs.twimg.com/media/HDm-wluXcAILau9.jpg)
![MANISH1027512's tweet photo. 动作指导古小一，限时回归
公开这组动作注入提示词，去给你的模特注入灵魂吧🤪
Realistic character editing, strictly use the identity reference image as the identity reference: keep the exact same face, facial feature details and proportions, skin tone, hairstyle (color/bangs/length/volume), clothing (silhouette/material/layering), and accessories completely unchanged (no face swap, no additional person, no changes to hairstyle/clothing/accessories).
[Task Goal]
Inject a dynamic pose with a shy, girlish feeling, and use a single giant close-up peace sign hand gesture as the visual focal point; while preserving identity and outfit, only modify the pose, camera structure, cropping logic, front-back layering, and foreshortening, without adding props, and only add slight shy facial expression changes that match the pose.
[Lens & Framing]
Vertical composition 9:16, an almost full-body portrait crop, camera positioned slightly above the subject from the front, angled downward about 25–40 degrees, not a straight top-down vertical shot, distance from the foreground hand approximately 25–45 cm, with an 18–24mm wide-angle feel, and the near-large far-small perspective must be obvious; the subject as a whole occupies about 84%–92% of the frame height, the head is located in the upper-middle area of the frame, and the rear lower leg may approach the bottom edge, but it must not become a distant standing full-body shot.
[Body Axis]
The main body axis runs diagonally from lower left to upper right, with the spine forming a soft C-curve; the head tilts about 10–18 degrees toward the left side of the frame, and the chin is gently lowered about 5–10 degrees; the ribcage is in a relaxed 3/4 front view; the pelvis rotates about 15–25 degrees away from the camera; the right shoulder is slightly brought forward and sits higher than the left shoulder; the overall temperament should be a suspended, shy-girl dynamic freeze-frame, light, soft, and slightly bashful, rather than aggressive or athletic.
[Upper-Limb Perspective]
The foreground right arm extends about 70–90 degrees toward the camera relative to the torso, the right elbow bends about 95–110 degrees, and the right forearm directly enters the camera plane; the right hand forms a huge peace sign in front of the lens, with the index and middle fingers extended into a V shape, the ring finger and little finger folded inward, and the thumb assisting in holding the folded fingers, the palm at a slight diagonal angle to the lens, each hand must have exactly five fingers, joints clearly defined, with no fusion.
The left upper arm is raised to the side of the head about 45–60 degrees, the left elbow bends about 70–90 degrees, and the left forearm remains near the plane of the face; the left hand stays in a small tucked hand pose against the face, positioned beside the cheek or temple, forming a compact inward-curled little claw / half-closed pose, with the fingers naturally bent inward, not a peace sign, and not a fully open palm, and it must be clearly smaller than the foreground right hand.
[Lower-Limb Perspective]
No clearly grounded supporting foot should appear; one thigh folds upward in front of the lower torso, with the knee bent about 70–100 degrees, while the other leg extends downward and drags backward; due to perspective, the lower body as a whole must be smaller than the torso and foreground hand, and the folded thigh may obscure about 10%–20% of the pelvis.
[Depth / Occlusion]
The depth-of-field hierarchy is: nearest foreground right-hand peace sign > right forearm > head and left hand > ribcage and torso > folded thigh and pelvis > rear lower leg/foot farthest away.
The right forearm should obscure about 10%–20% of the head and upper torso; the smaller left hand should slightly press against the contour of the face edge about 5%–10%; clear front-back layering must be preserved, and all limbs must not be flattened into the same plane.
[Hard Locks]
Only the foreground right hand is a peace sign, and it is the first priority, occupying about 35%–45% of the frame width.
The left hand against the face must remain in a small tucked hand pose and must not become a peace sign.
The 9:16 vertical composition with obvious slightly-above-front wide-angle perspective must be clear.
The head must be behind the foreground hand.
Preserve the overall diagonal suspended silhouette and the folded lower-limb relationship.
[Adaptable]
Background color, lighting atmosphere, material rendering, and environmental style may be freely adapted, but identity, outfit, 9:16 aspect ratio, camera structure, and action geometry must be locked.
[Background / Format]
Clean seamless background, real skin texture, with a slight shy expression, no heavy skin smoothing, no added props, no text, and no watermark.
Negative Prompt:
identity drift, face swap, different person, facial proportion changes, facial detail changes, skin tone changes, hairstyle changes, clothing changes, accessory changes, wrong camera angle, straight top-down vertical shot, low-angle upward shot, distant eye-level shot, telephoto flattening, perspective too weak, wrong cropping, wrong aspect ratio, non-9:16 composition, subject too small in frame, distant full-body shot, wrong orientation of head/ribcage/pelvis, symmetrical posed shot, stable standing pose, wrong center-of-gravity relationship, wrong limb direction, wrong joint angles, wrong planar relationships, foreground hand not large enough, both hands turned into peace signs, left hand turned into peace sign, foreground hand turned into open palm, foreground hand turned into fist, foreground hand turned into claw hand, incorrect peace sign, missing V-shaped gesture, wrong finger combination, wrong palm orientation, extra fingers, missing fingers, fused fingers, hand deformity, broken wrist, twisted wrist, broken elbow joint, dislocated shoulder joint, wrong front-back layering, head moving in front of foreground hand, no occlusion relationship between hand and face, floating limb misalignment, lost foreshortening, nearer limbs becoming smaller instead, feet emphasized, fierce expression, mature sexy atmosphere, overly strong athletic feeling, background clutter, extra props, text, watermark, logo, line-art style, sketch style, cartoon, anime, 3D render](https://pbs.twimg.com/media/HDm-vxOboAAeb_o.jpg)
Olivia
Online
Alin｜AI实战严选
@AlinMge
帮普通人筛选真正能用的AI干货
Prompt｜工具｜教程｜实战案例
亲测优先，不搬垃圾信息
目标：1000条AI实战严选
Joined January 2024
612 Following
178 Followers
550 Posts
Pinned Tweet
🚀 我开始做「普通人AI实战严选库」。
只收4类：
1. 可复制Prompt
2. 能上手的AI工具
3. 普通人看得懂的教程
4. 真实有效的案例
不收空泛资讯、标题党、大佬专属复杂玩法。
目标：1000条真正能用的AI干货。
欢迎推荐，我会筛选后收进库。
#AlinAI严选
@LmhdbAi 等其他公司做出能赚钱的成品，再用现金流去收购即将倒闭的公司更进，跳过前期投入，直接摘果子
我又回来了，自建vps+静态住宅ip，这次希望能稳定在线。
@InfiCheesy 老师，我在ytb看到您发布的vps自建节点教程。之前用快连使用claude较多，经常被封。现在快连没了，也没找到合适的代替。想自建节点。老师您现在有适合我们这种喜欢用ai工具的搭建方式推荐吗？
@JiaweiShen2568 机场很多，但没有高纯净的IP，cc封号太严重了
@libapi_ 这个确实好用啊，我手机电脑都用的这个
这条Prompt我最推荐用在3个场景：
1. 看完长文章后，不知道下一步怎么做
2. 学完教程后，不知道怎么练习
3. 想做项目，但脑子里只有碎片想法
普通人用AI，关键不是“总结”，而是让它帮你推进下一步。
AI实战严选 001/1000｜Prompt
今天这条适合：
看到很多AI工具、教程、Prompt，但收藏后不知道怎么行动的人。
它解决的问题：
以前你只是收藏一堆资料。
现在可以让AI直接帮你拆成目标、步骤、优先级和今天要做的事。
我筛选后的判断：
这条适合整理文章、课程、长视频笔记、项目方案。
不适合直接做最终决策，但很适合做第一版行动清单。
普通人使用建议：
不要只让AI“总结”，要让它告诉你“下一步该做什么”。
收藏指数：★★★★☆
【复制版】
请你把下面这段内容，整理成一个可执行清单。
要求：
1. 先总结核心目标
2. 再拆成3-5个执行模块
3. 每个模块给出具体行动步骤
4. 标注优先级：高/中/低
5. 最后给出今天最应该完成的3件事
内容如下：
【粘贴你的资料】
你最近最想用AI解决什么问题？
评论区告诉我，我优先去筛。
#AlinAI严选
我的筛选标准：
1. 普通人能不能用
2. 能不能立刻出结果
3. 是否值得收藏
4. 是否有真实应用场景
以后每条都会写清楚：
适合谁 / 解决什么问题 / 我的判断 / 使用提醒。
🚀 我开始做「普通人AI实战严选库」。
只收4类：
1. 可复制Prompt
2. 能上手的AI工具
3. 普通人看得懂的教程
4. 真实有效的案例
不收空泛资讯、标题党、大佬专属复杂玩法。
目标：1000条真正能用的AI干货。
欢迎推荐，我会筛选后收进库。
#AlinAI严选
@mubeitech 愚人节是明天，兄弟
#每日推荐 过滤90%的AI垃圾信息，收集1000条优质AI干货
欢迎评论区推荐：
1️⃣直接可抄的优质Prompt
2️⃣最新好用的AI工具网站
3️⃣实战教程 & 使用技巧
实用工具网站 ：33/1000
原创老师：@Jason23818126
#AI #AIイラスト #al教程
微信视频号、推特、油管的视频，怎么直接免费下载到本地？
今天整理了 5 个能搞定全网 90% 平台的免费下载工具。有存视频需求的朋友，建议随手码住备用。👇
1. 视频号专用：res-downloader
GitHub 上的开源工具，专门针对微信视频号开发。不需要再用手机录屏，可以直接提取原视频。
🔗 https://t.co/ah9zmdAGI6
2. 全平台高清下载：datatool
支持 4K 分辨率，网页端直接操作，能解析大部分常见的网页视频。
🔗 https://t.co/XDDRrGiDJl
3. 推特视频免登录下载：twitterxz
专用于推特，复制链接贴进去就能下。无需注册登录，支持自选分辨率。
🔗 https://t.co/7lQUnHD16Z
4. 支持 1000+ 网站与 RSS：VidBee
GitHub 7.7k Star 的项目。支持全球上千个网站的音视频下载，带有 RSS 自动下载功能，适合批量存素材或追更。
🔗 https://t.co/0Cvqb7lWPq
5. 支持 40+ 国内平台：greenvideo
国内平台覆盖很全的在线下载工具，免安装，打开网页就能直接用。
🔗 https://t.co/a0DFeIFaZx
⚠️ 版权提示： 下载的视频版权归原作者所有，如果用于商用请注意风险。
觉得有用的话，欢迎点赞支持一下～
Prompt
复制出来它为什么失去了排版，我就不得而知了！
————————
# ROLE You are a documentary-style street photographer shooting on a Fujifilm film simulation look. Produce ONE photorealistic image with a humanistic 35mm feeling. # INPUTS (REQUIRED) Image A = face/identity reference (must match 100%) Image B = outfit reference (must match 100%) # IDENTITY + OUTFIT LOCK (CRITICAL) - Use Image A as the ONLY identity source: 100% match facial structure, proportions, skin details, hairstyle, vibe. - Use Image B as the ONLY wardrobe source: match garments, colors, fit, layers exactly. - No face drift, no reshaping, no age/ethnicity change, no outfit substitution. # LIGHT (MUST HAVE SUN FLARE) - Bright sunshine with visible rays entering the lens (real lens flare + light streaks). - Backlight or 3/4 backlight, warm highlights, gentle shadows. - Sun flare must feel organic, not CGI stickers. # EXPRESSION / MOOD - Subject smiling sweetly (甜甜的笑), joyful, relaxed, approachable. - Candid, lively, not posed-stiff. # WIND / MOTION - Wind lifts clothing slightly and moves fabric naturally. - Hair strands softly brush across the face. - Motion is subtle and believable; face remains clear. # ENVIRONMENT (SOUTHEAST ASIA, COLORFUL BUT TIDY) Background: colorful wall / house facade / parked car in a Southeast Asian street vibe. - Colorful but orderly, clean composition, not messy. - Details: warm plaster walls, pastel paint, tiled edges, simple signage, neat wires, tropical calm. - Avoid chaotic street clutter. # CAMERA / LOOK - Humanistic 35mm lens feel (classic street perspective, no wide distortion). - Documentary framing, shallow-to-medium depth of field. - Fujifilm film simulation vibe: soft highlight roll-off, gentle contrast, slightly warm tones, fine natural grain. - Handheld realism, not studio. # RANDOMIZATION ENGINE (IMPORTANT) You must NOT default to the first option in any list. You must first internally shuffle every matrix below. After shuffling, choose EXACTLY ONE item from each matrix. Do not merge multiple items from the same matrix. Do not choose the first listed item by default. If two chosen items feel too similar or redundant, resample one of them once. The final image must use exactly one selected option from each matrix below. # RANDOM CHOICE MATRICES (MODEL MUST PICK ONE FROM EACH) ## ACTION / POSE MATRIX (pick ONE) A1: walking naturally, mid-step, looking to camera A2: light jog/run, playful, smiling A3: small jump, skirt/cloth lifted by wind, joyful A4: squat/crouch to look at something, smiling up A5: holding a roadside railing while leaning forward slightly A6: leaning on a traffic light pole (red/green light pole), casual grin A7: shot through glass (window/shop glass), subject inside/outside looking at camera ## HAND PROP MATRIX (pick ONE, hold naturally) P1: iced coffee in hand (clear cup, condensation) P2: smartphone in hand (screen not readable) P3: small bouquet of flowers P4: simple street food (wrapped snack or pastry) P5: coffee + phone together (one in each hand) ## CAMERA ANGLE MATRIX (pick ONE) C1: eye-level, classic street framing C2: low angle looking up (仰拍), sun flare stronger C3: high angle looking down (俯拍), playful perspective C4: slightly side-angle with leading lines of street C5: close foreground + subject mid-ground (layered depth) ## FOREGROUND OCCLUSION MATRIX (pick ONE, partial and tasteful) F1: foreground plant leaves partially遮挡画面 F2: a passing person softly blocks a small edge of frame (motion blur, anonymous) F3: a parked car edge blocks lower corner (frame cut, realistic) F4: shot through car window edge / mirror reflection (subtle) F5: foreground object (signpost / pole) creates a narrow frame edge # COMPOSITION RULES - Keep the subject as the clear hero; occlusions must not cover the face. - Sun flare and rays must be visible; do not overexpose the face. - Keep background tidy and graphic; color blocks should read clearly. - Maintain realistic anatomy and natural motion. # QUALITY Photorealistic, magazine-grade documentary retouching (clean but textured). Fine grain, Fuji-like color, no plastic skin. # NEGATIVE PROMPT (STRICT) Do NOT: - change identity or outfit - remove sun flare / make it flat light - make background chaotic or dirty - add heavy beauty filters or AI-smoothed skin - distort hands/face/limbs - add logos/brand marks/watermarks - make prop text readable or add random textm skin tones, crisp garment rendering, high-end medium format camera feel.
我知道你对未来有很多不确定
其实不是
未来已经发生了
只是你尚未察觉
未来10年
人类的寿命会达到150岁
是真的
未来10年
人人都将获得足够的财富与自由
是真的
未来10年
人类的生活将会比之前任何时代都幸福
是真的
再给AI一些时间，一些耐心，一些包容
现在你唯一需要做的就是
好好活着
Do nothing
Don't die.
发布一个新的 Skill：baoyu-youtube-transcript
输入 YouTube URL，直接抓取视频字幕，生成带章节、发言人和封面图的文档，不需要任何 API Key。
【怎么用】
选择这个 Skill，把 YouTube 链接丢进去就行。支持完整链接、短链接、嵌入链接、Shorts 链接，甚至直接输入视频 ID 都可以。
默认输出带时间戳的 Markdown 格式，也可以导出 SRT 字幕文件。支持多语言，可以指定优先语言，也可以翻译成其他语言。
第一次抓取后会自动缓存原始数据，之后换格式、换参数都不用重新请求，秒出结果。
【工作原理】
底层调用的是 YouTube 的 InnerTube API，这是 YouTube 内部用来获取字幕数据的接口，公开可用但没有官方文档。好处是不需要 Google API Key，不需要 OAuth 认证，脚本直接发请求就能拿到字幕数据。
拿到原始字幕后，脚本会做一次智能断句处理：按句末标点（句号、问号、感叹号等）切分，跨字幕片段合并成完整句子，时间戳按字符长度等比分配，对中日韩文字做了专门适配。这样输出的文本是自然的句子，不是 YouTube 那种碎片化的逐行字幕。
【章节分割】
如果视频描述里有章节时间戳（比如 "0:00 Introduction"），脚本会自动解析，按章节把字幕分段，生成带目录的 Markdown。没有章节信息的视频，就按段落分组输出。
【说话人识别】
这是最有意思的部分。YouTube 字幕本身不带说话人信息，所以识别说话人需要 AI 后处理。
流程是这样的：先用 --speakers 参数抓取原始字幕，脚本会把视频元数据（标题、频道名、简介）和 SRT 格式的原始字幕一起输出到一个 Markdown 文件里。然后启动一个 AI 子代理（用 Claude Sonnet，够用且省成本），按预设的 Prompt 模板处理这个文件。
AI 识别说话人的逻辑分三层优先级：首先从元数据推断，视频标题通常包含嘉宾名字，频道名就是主持人；其次从对话内容判断，比如自我介绍、互相称呼；都不行就用通用标签（Speaker 1、Host 之类），保持全文一致。如果后面对话中才出现名字，会回溯更新前面所有标签。
处理完的输出是带说话人标签的分段对话，长独白会被切成 2-4 句一段，每段末尾带时间范围。
【缓存机制】
第一次运行会缓存四样东西：视频元数据（meta.json）、原始字幕片段（transcript-raw.json）、断句后的字幕（transcript-sentences.json）、视频封面图（cover.jpg）。之后不管切换格式还是重新生成，都直接用缓存，不再请求网络。加 --refresh 参数可以强制刷新。
安装命令：
$ npx skills add jimliu/baoyu-skills --skill baoyu-youtube-transcript
项目地址：https://t.co/GOjnWisJjw
是不是好多天沒發🍌香蕉prompt了？
馬上搞一個！
最近潛心研究Grok去了，補上🍌prompt：
[Character & Pose 👩🦰🌃]: Ultra close-up portrait photo, photograph by playforge,CCD + strong flash, a 19s top-tier beautiful cold K-pop young woman with high ponytail hairstyle, extremely fair cold white skin, sharp prominent nose, exquisite delicate features, refined makeup, lazily sitting on the stone steps in front of Tokyo Tower. Wearing extremely minimalist pure light color outfit: white large off-shoulder short knit cardigan + light pink-purple waist-cinching suspender tight long skirt (skirt tightly stretched and clinging to body curves, slight belly pooch, full bust), Van Cleef & Arpels red jewelry necklace + jade bracelet, transparent PVC one-strap high-heel sandals. Arms casually resting on knees, legs naturally apart, eyes with slightly distant aloof gaze looking toward camera, overall posture relaxed yet full of powerful presence.
[Scene & Lighting 🌙🗼]: Thick clouds partially covering the moon, dark windy night with scattered starlight points, vibe shot of vsc,moon positioned exactly at the top spire of Tokyo Tower creating soft hazy halo glow effect, light passing through the tower's lattice structure casting unique geometric shadows on the figure and stone steps.
[Composition & Atmosphere 🎥]: 4:9 vertical composition, design by 94van,extreme low worm's eye view from ground level仰拍, ultra wide angle capturing the complete Tokyo Tower in background (occupying 99% of background frame), forming strong visual correspondence with the foreground figure. Overall palette dominated by pure white tones, silent starry sky, and deep brown of the tower, creating high-end, cold, cinematic midnight Tokyo atmosphere.
[Style 📸]: Realistic CCD photo feel, low saturation soft tones, PF style,non-uniform natural lighting, dramatic chiaroscuro layers.
[Texture ✨]: Ultra high definition skin texture, premium retouching + soft glow, lifelike realistic quality, luxurious cinematic portrait aesthetics.
[Additional text ✍️]: On the right side of the image, 5% from the edge, in every quarter section of the frame, place one thin vertical handwritten small seal script "玖肆玩AI" (four places total); bottom right corner has extremely fine semi-transparent 3D crystal handwritten "PF".
![94vanAI's tweet photo. 是不是好多天沒發🍌香蕉prompt了？
馬上搞一個！
最近潛心研究Grok去了，補上🍌prompt：
[Character & Pose 👩🦰🌃]: Ultra close-up portrait photo, photograph by playforge,CCD + strong flash, a 19s top-tier beautiful cold K-pop young woman with high ponytail hairstyle, extremely fair cold white skin, sharp prominent nose, exquisite delicate features, refined makeup, lazily sitting on the stone steps in front of Tokyo Tower. Wearing extremely minimalist pure light color outfit: white large off-shoulder short knit cardigan + light pink-purple waist-cinching suspender tight long skirt (skirt tightly stretched and clinging to body curves, slight belly pooch, full bust), Van Cleef & Arpels red jewelry necklace + jade bracelet, transparent PVC one-strap high-heel sandals. Arms casually resting on knees, legs naturally apart, eyes with slightly distant aloof gaze looking toward camera, overall posture relaxed yet full of powerful presence.
[Scene & Lighting 🌙🗼]: Thick clouds partially covering the moon, dark windy night with scattered starlight points, vibe shot of vsc,moon positioned exactly at the top spire of Tokyo Tower creating soft hazy halo glow effect, light passing through the tower's lattice structure casting unique geometric shadows on the figure and stone steps.
[Composition & Atmosphere 🎥]: 4:9 vertical composition, design by 94van,extreme low worm's eye view from ground level仰拍, ultra wide angle capturing the complete Tokyo Tower in background (occupying 99% of background frame), forming strong visual correspondence with the foreground figure. Overall palette dominated by pure white tones, silent starry sky, and deep brown of the tower, creating high-end, cold, cinematic midnight Tokyo atmosphere.
[Style 📸]: Realistic CCD photo feel, low saturation soft tones, PF style,non-uniform natural lighting, dramatic chiaroscuro layers.
[Texture ✨]: Ultra high definition skin texture, premium retouching + soft glow, lifelike realistic quality, luxurious cinematic portrait aesthetics.
[Additional text ✍️]: On the right side of the image, 5% from the edge, in every quarter section of the frame, place one thin vertical handwritten small seal script "玖肆玩AI" (four places total); bottom right corner has extremely fine semi-transparent 3D crystal handwritten "PF".](https://pbs.twimg.com/media/HD74rzsasAA9pn2.jpg)
Grok 出图效果越来越棒了
实践图片，太好用了
#每日推荐 过滤90%的AI垃圾信息，收集1000条优质AI干货
欢迎评论区推荐：
1️⃣直接可抄的优质Prompt
2️⃣最新好用的AI工具网站
3️⃣实战教程 & 使用技巧
直接可抄的优质Prompt ：29/1000
原创老师：@MANISH1027512
老师的提示词真的有东西。
#AI #AIイラスト #al教程
#每日推荐 过滤90%的AI垃圾信息，收集1000条优质AI干货
欢迎评论区推荐：
1️⃣直接可抄的优质Prompt
2️⃣最新好用的AI工具网站
3️⃣实战教程 & 使用技巧
直接可抄的优质Prompt ：29/1000
原创老师：@MANISH1027512
老师的提示词真的有东西。
#AI #AIイラスト #al教程
动作指导古小一，限时回归
公开这组动作注入提示词，去给你的模特注入灵魂吧🤪
Realistic character editing, strictly use the identity reference image as the identity reference: keep the exact same face, facial feature details and proportions, skin tone, hairstyle (color/bangs/length/volume), clothing (silhouette/material/layering), and accessories completely unchanged (no face swap, no additional person, no changes to hairstyle/clothing/accessories).
[Task Goal]
Inject a dynamic pose with a shy, girlish feeling, and use a single giant close-up peace sign hand gesture as the visual focal point; while preserving identity and outfit, only modify the pose, camera structure, cropping logic, front-back layering, and foreshortening, without adding props, and only add slight shy facial expression changes that match the pose.
[Lens & Framing]
Vertical composition 9:16, an almost full-body portrait crop, camera positioned slightly above the subject from the front, angled downward about 25–40 degrees, not a straight top-down vertical shot, distance from the foreground hand approximately 25–45 cm, with an 18–24mm wide-angle feel, and the near-large far-small perspective must be obvious; the subject as a whole occupies about 84%–92% of the frame height, the head is located in the upper-middle area of the frame, and the rear lower leg may approach the bottom edge, but it must not become a distant standing full-body shot.
[Body Axis]
The main body axis runs diagonally from lower left to upper right, with the spine forming a soft C-curve; the head tilts about 10–18 degrees toward the left side of the frame, and the chin is gently lowered about 5–10 degrees; the ribcage is in a relaxed 3/4 front view; the pelvis rotates about 15–25 degrees away from the camera; the right shoulder is slightly brought forward and sits higher than the left shoulder; the overall temperament should be a suspended, shy-girl dynamic freeze-frame, light, soft, and slightly bashful, rather than aggressive or athletic.
[Upper-Limb Perspective]
The foreground right arm extends about 70–90 degrees toward the camera relative to the torso, the right elbow bends about 95–110 degrees, and the right forearm directly enters the camera plane; the right hand forms a huge peace sign in front of the lens, with the index and middle fingers extended into a V shape, the ring finger and little finger folded inward, and the thumb assisting in holding the folded fingers, the palm at a slight diagonal angle to the lens, each hand must have exactly five fingers, joints clearly defined, with no fusion.
The left upper arm is raised to the side of the head about 45–60 degrees, the left elbow bends about 70–90 degrees, and the left forearm remains near the plane of the face; the left hand stays in a small tucked hand pose against the face, positioned beside the cheek or temple, forming a compact inward-curled little claw / half-closed pose, with the fingers naturally bent inward, not a peace sign, and not a fully open palm, and it must be clearly smaller than the foreground right hand.
[Lower-Limb Perspective]
No clearly grounded supporting foot should appear; one thigh folds upward in front of the lower torso, with the knee bent about 70–100 degrees, while the other leg extends downward and drags backward; due to perspective, the lower body as a whole must be smaller than the torso and foreground hand, and the folded thigh may obscure about 10%–20% of the pelvis.
[Depth / Occlusion]
The depth-of-field hierarchy is: nearest foreground right-hand peace sign > right forearm > head and left hand > ribcage and torso > folded thigh and pelvis > rear lower leg/foot farthest away.
The right forearm should obscure about 10%–20% of the head and upper torso; the smaller left hand should slightly press against the contour of the face edge about 5%–10%; clear front-back layering must be preserved, and all limbs must not be flattened into the same plane.
[Hard Locks]
Only the foreground right hand is a peace sign, and it is the first priority, occupying about 35%–45% of the frame width.
The left hand against the face must remain in a small tucked hand pose and must not become a peace sign.
The 9:16 vertical composition with obvious slightly-above-front wide-angle perspective must be clear.
The head must be behind the foreground hand.
Preserve the overall diagonal suspended silhouette and the folded lower-limb relationship.
[Adaptable]
Background color, lighting atmosphere, material rendering, and environmental style may be freely adapted, but identity, outfit, 9:16 aspect ratio, camera structure, and action geometry must be locked.
[Background / Format]
Clean seamless background, real skin texture, with a slight shy expression, no heavy skin smoothing, no added props, no text, and no watermark.
Negative Prompt:
identity drift, face swap, different person, facial proportion changes, facial detail changes, skin tone changes, hairstyle changes, clothing changes, accessory changes, wrong camera angle, straight top-down vertical shot, low-angle upward shot, distant eye-level shot, telephoto flattening, perspective too weak, wrong cropping, wrong aspect ratio, non-9:16 composition, subject too small in frame, distant full-body shot, wrong orientation of head/ribcage/pelvis, symmetrical posed shot, stable standing pose, wrong center-of-gravity relationship, wrong limb direction, wrong joint angles, wrong planar relationships, foreground hand not large enough, both hands turned into peace signs, left hand turned into peace sign, foreground hand turned into open palm, foreground hand turned into fist, foreground hand turned into claw hand, incorrect peace sign, missing V-shaped gesture, wrong finger combination, wrong palm orientation, extra fingers, missing fingers, fused fingers, hand deformity, broken wrist, twisted wrist, broken elbow joint, dislocated shoulder joint, wrong front-back layering, head moving in front of foreground hand, no occlusion relationship between hand and face, floating limb misalignment, lost foreshortening, nearer limbs becoming smaller instead, feet emphasized, fierce expression, mature sexy atmosphere, overly strong athletic feeling, background clutter, extra props, text, watermark, logo, line-art style, sketch style, cartoon, anime, 3D render
![MANISH1027512's tweet photo. 动作指导古小一，限时回归
公开这组动作注入提示词，去给你的模特注入灵魂吧🤪
Realistic character editing, strictly use the identity reference image as the identity reference: keep the exact same face, facial feature details and proportions, skin tone, hairstyle (color/bangs/length/volume), clothing (silhouette/material/layering), and accessories completely unchanged (no face swap, no additional person, no changes to hairstyle/clothing/accessories).
[Task Goal]
Inject a dynamic pose with a shy, girlish feeling, and use a single giant close-up peace sign hand gesture as the visual focal point; while preserving identity and outfit, only modify the pose, camera structure, cropping logic, front-back layering, and foreshortening, without adding props, and only add slight shy facial expression changes that match the pose.
[Lens & Framing]
Vertical composition 9:16, an almost full-body portrait crop, camera positioned slightly above the subject from the front, angled downward about 25–40 degrees, not a straight top-down vertical shot, distance from the foreground hand approximately 25–45 cm, with an 18–24mm wide-angle feel, and the near-large far-small perspective must be obvious; the subject as a whole occupies about 84%–92% of the frame height, the head is located in the upper-middle area of the frame, and the rear lower leg may approach the bottom edge, but it must not become a distant standing full-body shot.
[Body Axis]
The main body axis runs diagonally from lower left to upper right, with the spine forming a soft C-curve; the head tilts about 10–18 degrees toward the left side of the frame, and the chin is gently lowered about 5–10 degrees; the ribcage is in a relaxed 3/4 front view; the pelvis rotates about 15–25 degrees away from the camera; the right shoulder is slightly brought forward and sits higher than the left shoulder; the overall temperament should be a suspended, shy-girl dynamic freeze-frame, light, soft, and slightly bashful, rather than aggressive or athletic.
[Upper-Limb Perspective]
The foreground right arm extends about 70–90 degrees toward the camera relative to the torso, the right elbow bends about 95–110 degrees, and the right forearm directly enters the camera plane; the right hand forms a huge peace sign in front of the lens, with the index and middle fingers extended into a V shape, the ring finger and little finger folded inward, and the thumb assisting in holding the folded fingers, the palm at a slight diagonal angle to the lens, each hand must have exactly five fingers, joints clearly defined, with no fusion.
The left upper arm is raised to the side of the head about 45–60 degrees, the left elbow bends about 70–90 degrees, and the left forearm remains near the plane of the face; the left hand stays in a small tucked hand pose against the face, positioned beside the cheek or temple, forming a compact inward-curled little claw / half-closed pose, with the fingers naturally bent inward, not a peace sign, and not a fully open palm, and it must be clearly smaller than the foreground right hand.
[Lower-Limb Perspective]
No clearly grounded supporting foot should appear; one thigh folds upward in front of the lower torso, with the knee bent about 70–100 degrees, while the other leg extends downward and drags backward; due to perspective, the lower body as a whole must be smaller than the torso and foreground hand, and the folded thigh may obscure about 10%–20% of the pelvis.
[Depth / Occlusion]
The depth-of-field hierarchy is: nearest foreground right-hand peace sign > right forearm > head and left hand > ribcage and torso > folded thigh and pelvis > rear lower leg/foot farthest away.
The right forearm should obscure about 10%–20% of the head and upper torso; the smaller left hand should slightly press against the contour of the face edge about 5%–10%; clear front-back layering must be preserved, and all limbs must not be flattened into the same plane.
[Hard Locks]
Only the foreground right hand is a peace sign, and it is the first priority, occupying about 35%–45% of the frame width.
The left hand against the face must remain in a small tucked hand pose and must not become a peace sign.
The 9:16 vertical composition with obvious slightly-above-front wide-angle perspective must be clear.
The head must be behind the foreground hand.
Preserve the overall diagonal suspended silhouette and the folded lower-limb relationship.
[Adaptable]
Background color, lighting atmosphere, material rendering, and environmental style may be freely adapted, but identity, outfit, 9:16 aspect ratio, camera structure, and action geometry must be locked.
[Background / Format]
Clean seamless background, real skin texture, with a slight shy expression, no heavy skin smoothing, no added props, no text, and no watermark.
Negative Prompt:
identity drift, face swap, different person, facial proportion changes, facial detail changes, skin tone changes, hairstyle changes, clothing changes, accessory changes, wrong camera angle, straight top-down vertical shot, low-angle upward shot, distant eye-level shot, telephoto flattening, perspective too weak, wrong cropping, wrong aspect ratio, non-9:16 composition, subject too small in frame, distant full-body shot, wrong orientation of head/ribcage/pelvis, symmetrical posed shot, stable standing pose, wrong center-of-gravity relationship, wrong limb direction, wrong joint angles, wrong planar relationships, foreground hand not large enough, both hands turned into peace signs, left hand turned into peace sign, foreground hand turned into open palm, foreground hand turned into fist, foreground hand turned into claw hand, incorrect peace sign, missing V-shaped gesture, wrong finger combination, wrong palm orientation, extra fingers, missing fingers, fused fingers, hand deformity, broken wrist, twisted wrist, broken elbow joint, dislocated shoulder joint, wrong front-back layering, head moving in front of foreground hand, no occlusion relationship between hand and face, floating limb misalignment, lost foreshortening, nearer limbs becoming smaller instead, feet emphasized, fierce expression, mature sexy atmosphere, overly strong athletic feeling, background clutter, extra props, text, watermark, logo, line-art style, sketch style, cartoon, anime, 3D render](https://pbs.twimg.com/media/HDm-xzLWQAAaZJc.jpg)
Last Seen Users on Sotwe
Trends for you
Most Popular Users
Elon Musk
@elonmusk
241.2M followers
Barack Obama
@barackobama
119.1M followers
Cristiano Ronaldo
@cristiano
112.6M followers
Donald J. Trump
@realdonaldtrump
111.8M followers
Narendra Modi
@narendramodi
107.1M followers
Rihanna
@rihanna
98.2M followers
NASA
@nasa
92.2M followers
Justin Bieber
@justinbieber
91.4M followers
KATY PERRY
@katyperry
88.9M followers
Taylor Swift
@taylorswift13
82.8M followers
Lady Gaga
@ladygaga
74.2M followers
Virat Kohli
@imvkohli
71.7M followers
Kim Kardashian
@kimkardashian
70.4M followers
YouTube
@youtube
68.8M followers
Neymar Jr
@neymarjr
64.6M followers
Bill Gates
@billgates
64.5M followers
The Ellen Show
@theellenshow
62.4M followers
Selena Gomez
@selenagomez
61.9M followers
CNN
@cnn
61.8M followers
X
@x
60.8M followers