sedia tumbal proyek @masozee

All the screens from this week's AI notes concept, in one place. Which one would you open first?

keyboard_arrow_left Previous keyboard_arrow_right Next

Cloudflare 官网这排 Logo，一张图片都没用 每张卡上的品牌 Logo 都是一根根竖线拼出来的。鼠标移上去，线会散开再拼回来，旁边还会冒出几个检测框。看了下，页面里一张 Logo 图片都没有，是用 canvas 实时画出来的 做法其实不复杂。 程序先在一块看不见的画布上把 Logo 画一遍，切成很多小格子，记下每一格有多亮，存成一张亮度表。记完之后 Logo 就用不上了。屏幕上的竖线每一帧都去查这张表，亮的地方线画粗一点，暗的地方画细一点。几十根线排在一起，粗细一变，Logo 的形状就出来了，有点像用梳子在沙子上梳出图案。 Logo 本身就是线画出来的，所以线怎么动，Logo 就跟着怎么动。往下流、散开、再拼回来，都不会错位。要是在线底下垫一张图，线一动就对不上了。 先把图片转成亮度表，再让别的东西照着表来画，这个思路挺常见的，字符画和报纸上的网点印刷都是这么做的。我自己复刻的时候换成了 Apple、Google、Tesla 三个牌子，换了 Logo 的 SVG 路径。 交给 AI 做： 把下面这段复制给 Claude 或 ChatGPT，就能做出差不多的效果 👇

here it is: Create a 14.5-second, cinematic 3D fly-through of my landing page at 1440×1152, 60fps. Inputs: Landing page URL, brand name, logo mark, closing headline (a statement followed by a short italic tail), and optional 15-second music track. Visual direction: Place the webpage flat on a clean studio floor (#f9fbfb center, #f2f3f6 corners), with 24px rounded corners and soft shadows. Use one continuous camera movement: glide smoothly from the hero down the entire page, with floating cards, product mockups, photos, chips, and testimonials lifting slightly toward the camera as it passes. Maintain cinematic depth of field, realistic shadows, and subtle 3D depth. Rise overhead to reveal the entire page as one strip, then dive back into the hero, transforming it into a floating card for the final title reveal. Timeline: 0–8.2s: continuous fly-through. 7.75–9.25s: rise to overhead. 9.25–9.47s: hold the full-page view. 9.47–9.97s: dive into the hero. 9.97–14.5s: final title card with brand badge, headline, and italic tail. Use Geist Medium for the headline and Instrument Serif for the italic text. Keep four floating logo tiles around the hero throughout the transition. Motion rules: Constant forward movement with smooth acceleration and deceleration only at the beginning and end. No cuts, crossfades, sudden speed changes, sideways camera rotations, zoom reversals, glow effects, or text overlays before the final title card. Keep most floating elements 20–40px above the page. Build: Use Playwright and Chromium to capture the webpage at 1440px wide, extract individual floating elements with transparent backgrounds, and save their positions to JSON. Build a single WebGL2 scene with a time-seekable animation, realistic depth of field, mipmapped textures, soft shadows, and 16-subframe motion blur. Render the final video as H.264 MP4 at 60fps using GPU acceleration. Sync the music hit to 9.97s and add subtle whooshes at 8.66s and 9.80s. Important: If you cannot execute code on my computer, be upfront that you can only provide a live preview here and that final MP4 rendering requires Claude Code. Lock the page layout before animation to prevent section-height changes from shifting the sequence. Test a 30fps preview with optical flow to ensure smooth forward motion and consistent zoom. Workflow: First ask me for the required inputs. Capture the page and show me still previews at 0, 1.6, 2.5, 5, 6.5, 9.3, and 11 seconds. Get approval before rendering the final video.

Just updated my portfolio website. elaya.design

Cinetic v1.1 is here! Cinetic is my skill for creating good motion designs/launch videos some changes: - text elements are shorter now - calmer pacing by default - every sound is made in code with earshot this video was made with cinetic github.com/Leonxlnx/cinet…

Bikin akun faceless TikTok baru 3 hari yg lalu. Fully-automated yg handle Claude + Buffer Barusan gue cek dah nambah 40k followers aja wkwk Konten-nya full carousel ga pake video sama sekali :) I cracked TikTok distribution. I’ll scale to 30 accounts by end of the month 😂

Hanif | AI For Productivity @hanifproduktif

3 months ago

x.com/i/article/2080…

15 87 511 379K 2K

Design tip: If the app feels off but you're not sure why, it's probably the spacing.

Gara2 racun dari @radjathaher buat scale meta ads, akhirnya biar gampang gw bikin adsctl biar lebih sat set ngemanage meta ads (platform lain menyusul). Founded by SOL dan maintained by Argon dan Deepseek v4 lol. MIT. Silakan digunakan. github.com/imrenagicom/ad…

Introducing cinematic-shot An agent skill for recording hero shots for your product. Create high-quality cinematic shots with smooth 3D camera movements and stunning lighting. npx skills add 1weiho/cinematic-shot

Project card by @flohoeller Found on → bencho.dev

BENCI BANGET MALAH DIJADIIN DRAKOR PAKE AI 😭😭😭

@agnxsmar Ga ada lagi NGO GF, engineer BF dongs

Created a new skill for UI generation skeleton effect, something I’ve been using in my UI videos. These UI examples are built with transitions.dev UI Agent.

Alright, I've made the shadcn/ui skills really good at 𝙳𝙴𝚂𝙸𝙶𝙽.𝚖𝚍. Ask for a design system and it picks a preset, then builds everything: colors, tokens, type scale, elevation, styles the primitives and blocks. Start with a solid foundation and build something unique.

@coffebit Lagi diadminin mas dony iswara ya masbit

Cloudflare open sourced a security audit skill that orchestrates coding agents through six phases of reconnaissance and validation to produce independently verified, machine-readable findings. github.com/cloudflare/sec…

Dulu tiap sesi baru, selalu mulai dari konteks dan ingatan nol. Tapi sejak Agustus aku pasang OptMem ke lingkungan agentic AI, teks catatan yang ia isi sendiri tiap ada hal penting, sekarang sudah hampir seribuan baris.

@linear's Conversations on Quality is honestly one of the most worthwhile things to watch for designers and anyone who wants to raise the bar of the products they build. Make time for it