
Soran
5.7K posts

Soran
@Soranlan
INTJ Scorpio ♏️ 懒得卷,所以把 AI 玩透了 AI 工具拆解 · Agent 搭建 · 懒人自动化 踩了两年坑 → 你的作弊码



这个项目能扫描出你的代码的技术栈,然后配齐Skill 比如说,你的有React、Next.js、Tailwind、Prisma,都能给你装齐Skill 而且是人工精选的Skill,没啥安全问题,支持monorepo github.com/midudev/autosk…

我也接一个,GPT-2+Seedance 2.0是当前AI圈宇宙无敌组合。 游戏动态UI,直接生成了一个视频,有互动,有声效,当然还有场景...自己联想

SOMEONE JUST BUILT A CLAUDE CODE SKILL THAT PULLS EVERY VISUAL ASSET FROM ANY COMPANY ON THE INTERNET. Logo. Color palette. Typography. Photography style. UI components. All of it extracted automatically using Serp API and Firecrawl. One prompt. One shot. A complete brand moodboard document generated in seconds. What used to take a designer half a day of manual research now takes Claude less than a minute. Drop in any company URL and walk away with everything their design team spent years building. This is what Claude Code skills actually look like when someone builds something real with them. Not a demo. Not a concept. A tool that saves hours every single time you use it. The designers who build tools like this are not being replaced by AI. They are the ones building the AI.

OPENAI JUST DROPPED GPT-IMAGE-2 AND THE INTERNET IS NOT READY. This is not another image generator. This is the first model that genuinely understands context at a visual level. Here is what it actually does: It renders accurate text inside images. Not broken garbled letters. Real readable text, signs, labels, and headlines embedded directly into the image. It follows complex multi-step visual instructions. Not just "a dog on a beach." Entire scenes with specific lighting, composition, mood, and detail executed exactly as described. It edits existing images with surgical precision. Change one element without touching anything around it. No bleeding. No artifacts. No destroyed backgrounds. It generates product photography that looks indistinguishable from a professional shoot. Real brands are already replacing their entire photo studios with this. It handles transparency, layering, and design-ready outputs natively. The gap between what AI could generate last year and what GPT-Image-2 produces right now is not incremental. It is generational. The designers who learn to use this as a tool will 10x their output overnight. The ones who ignore it will spend 2027 wondering why clients stopped calling. Screenshot this and come back in 6 months.

Dude GPT-Image-2... wtf... how.

Anthropic just pulled Claude Code from the Pro plan. Pro users wanting it need Max now. $100/month minimum. 5x jump. I'm on Max 20x so I'm fine. Flagging for anyone on Pro who's about to find out. No announcement. Just a pricing page edit.

INSTEAD OF WATCHING NETFLIX TONIGHT. Spend 9 minutes with this. The CEO of Anthropic just answered the one question every non-technical person is too afraid to ask out loud. Should I even bother learning this or is it already too late. His answer is not what you would expect from someone whose company profits every time you use AI. He is not selling you on the tools. He is telling you the truth about what is actually coming and what it means for people who are not developers. 9 minutes. No technical background needed to understand it. The people who watch this tonight will make better decisions about their career, their skills, and their time than they have made in the last 6 months. The people who skip it will still be asking the same question in a year. Completely free. Bookmark this before you open Netflix 👇

生产力爆表!🤯ChatGPT-image-2.0生成手绘包装风格提案海报,广告公司设计师可以把PS卸了! 我用一个通用提示词模板,分别生成了奶粉、燕麦、键盘、电蚊拍四款产品包装提案图。如果你有实际产品,直接上传垫图,GPT会精准引用原产品图,效果会更惊艳! 通用提示词: 请为【产品名称】生成一张“手绘包装提案板风格”的产品介绍海报。 如果我提供了产品图片,请以该图片为主要参考,保留产品的核心外观特征、包装形态、颜色、材质、品牌识别和整体气质,并优先沿用产品图中的主色调与配色关系进行设计。 如果我没有提供产品图片,请根据【产品名称】自动生成符合该产品定位的产品外观与包装设计,并根据产品属性、品类特征和气质,自动搭配自然合理的主色、辅助色和背景色。 海报整体风格要求: 这不是普通电商详情页,也不是单纯棚拍海报,而是一张融合“产品成品图 + 包装设计草图 + 品牌概念开发过程”的创意提案板。整体呈现为高审美、高完成度的品牌设计展示图,像设计师正在展示一个包装概念方案。 画面要求: - 竖版构图 - 使用统一且有品牌感的主色调背景,背景颜色根据产品本身自然决定,不固定为某一种颜色 - 配色应与产品属性协调:如果有参考图,优先继承参考图的主色和品牌色;如果没有参考图,则根据产品名称自动生成合理的色彩方案 - 画面中央放置一个最完整、最吸睛的产品主视觉,具有真实质感、立体光影和商业成品效果 - 产品周围加入多个辅助视觉元素:不同角度的包装草图、结构草图、局部造型探索、手持展示小稿、包装展开思路、品牌字样排版试验 - 加入自然随性的黑色或深色手绘线稿、箭头、圈画、符号、手写中文批注,让画面具有强烈“设计过程感”和“创意提案感” - 成品图要相对精致真实,草图部分要相对随性灵动,形成“完成品 + 设计草图”的对比 - 版式看起来松弛自由,但整体要有秩序、有节奏、有视觉中心,体现高级平面设计感 - 可以适度加入与产品相关的小图形、小符号或趣味元素,增强品牌记忆点 - 最终效果应具有品牌感、创意感、过程感和视觉冲击力 配色原则: - 不要固定黄色背景 - 要根据产品类型、材质、用途、消费感受和品牌气质自动选择主色调 - 背景色、产品色、文字色、草图线稿色应彼此协调 - 可以使用高饱和单色背景,也可以使用柔和统一色系背景,只要整体有品牌感和视觉冲击力 - 色彩应自然服务于产品,而不是喧宾夺主 风格关键词: 包装设计提案板、品牌概念开发、手绘草图、设计过程感、moodboard、创意评审稿、产品成品渲染、视觉实验、商业设计海报 请特别注意: - 如果有参考产品图,优先遵循参考图,不要偏离产品本身 - 如果没有参考图,则根据产品名称自动生成合理的产品外观、包装和配色方案,并保持整体风格统一 - 画面重点是“产品介绍 + 创意设计过程展示” - 不要做成普通电商详情页,不要做成促销海报,不要过度整齐呆板,不要缺少草图和手写批注,不要廉价感,不要低幼卡通感

GPT Images 2.0 有点东西啊,估计太炸了官方都没敢说😂 YouTube 上刷到这个小哥哥生成了两本书的封面, 《从优秀到卓越》,《聪明的投资者》, 拿手机对着封面上的条形码一扫, 直接跳转到了亚马逊的真实购买页面 WTF😆😆😆 兄弟们不是画得像,牛逼的是条形码的每一根黑白条,编码都完全正确, 差一像素都扫不出来的东西,AI一次就生成对了🐂🍺 很多人以为这比生成文字简单, 其实正好相反, 文字错一个两个还能认, 条形码是严格的二进制结构, 错一格就彻底废了。 我觉得这才是真正牛逼的地方, 这次不再只是生成看起来像的东西, 它开始真正理解,这个东西在现实世界里的运行规则,Damn! 说白了,以前AI画图是给人看的, 现在AI画图是给机器看的, 是能直接和现实世界交互的。 以后你做产品包装,不用再单独找印刷厂做条码, 做海报,不用再后期贴二维码, 做门票,做会员卡,做优惠券, 一句话,AI直接给你生成能扫能用的完整版本,就说屌不屌吧! 当然硬币的另一面也很清晰,以后的假快递单,假门票,假支付码…… 也会变得和真的一模一样,扫一下还能跳转到假网站🥹🥹🥹 现实世界的信任成本又被拉高了一大截啊🙃🙃🙃 昨天大家还在惊叹它画的大米粒有多逼真,今天才反应过来, 真正的革命从来都不是好看,是能用,是能接入我们已经运行了几十年的真实世界。 一起迎接人类的群星闪耀时吧(・ω・)ノ☽



Dario is wrong. He knows absolutely nothing about the effects of technological revolutions on the labor market. Don't listen to him, Sam, Yoshua, Geoff, or me on this topic. Listen to economists who have spent their career studying this, like @Ph_Aghion , @erikbryn , @DAcemogluMIT , @amcafee , @davidautor







